{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 15.0,
  "global_step": 1575,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0019047619047619048,
      "grad_norm": 6.249958182433422,
      "learning_rate": 2.0833333333333333e-07,
      "loss": 1.2548,
      "step": 1
    },
    {
      "epoch": 0.0038095238095238095,
      "grad_norm": 7.3275358434995885,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 1.1995,
      "step": 2
    },
    {
      "epoch": 0.005714285714285714,
      "grad_norm": 7.962704464287428,
      "learning_rate": 6.25e-07,
      "loss": 1.3552,
      "step": 3
    },
    {
      "epoch": 0.007619047619047619,
      "grad_norm": 6.834862381995382,
      "learning_rate": 8.333333333333333e-07,
      "loss": 1.2567,
      "step": 4
    },
    {
      "epoch": 0.009523809523809525,
      "grad_norm": 6.670482016594432,
      "learning_rate": 1.0416666666666667e-06,
      "loss": 1.2178,
      "step": 5
    },
    {
      "epoch": 0.011428571428571429,
      "grad_norm": 6.228408996779406,
      "learning_rate": 1.25e-06,
      "loss": 1.4297,
      "step": 6
    },
    {
      "epoch": 0.013333333333333334,
      "grad_norm": 7.978127041228856,
      "learning_rate": 1.4583333333333335e-06,
      "loss": 1.2592,
      "step": 7
    },
    {
      "epoch": 0.015238095238095238,
      "grad_norm": 6.521728745252264,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 1.5888,
      "step": 8
    },
    {
      "epoch": 0.017142857142857144,
      "grad_norm": 6.430439812900178,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 1.3054,
      "step": 9
    },
    {
      "epoch": 0.01904761904761905,
      "grad_norm": 5.440801146286208,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 1.2474,
      "step": 10
    },
    {
      "epoch": 0.02095238095238095,
      "grad_norm": 4.484406039167107,
      "learning_rate": 2.2916666666666666e-06,
      "loss": 1.2364,
      "step": 11
    },
    {
      "epoch": 0.022857142857142857,
      "grad_norm": 4.551409690261343,
      "learning_rate": 2.5e-06,
      "loss": 1.2083,
      "step": 12
    },
    {
      "epoch": 0.024761904761904763,
      "grad_norm": 5.807313642046851,
      "learning_rate": 2.7083333333333334e-06,
      "loss": 1.4471,
      "step": 13
    },
    {
      "epoch": 0.02666666666666667,
      "grad_norm": 4.71193646005634,
      "learning_rate": 2.916666666666667e-06,
      "loss": 1.1337,
      "step": 14
    },
    {
      "epoch": 0.02857142857142857,
      "grad_norm": 4.842345764051664,
      "learning_rate": 3.125e-06,
      "loss": 1.1057,
      "step": 15
    },
    {
      "epoch": 0.030476190476190476,
      "grad_norm": 5.126936632402324,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.1551,
      "step": 16
    },
    {
      "epoch": 0.03238095238095238,
      "grad_norm": 3.9078888869265036,
      "learning_rate": 3.5416666666666673e-06,
      "loss": 1.0285,
      "step": 17
    },
    {
      "epoch": 0.03428571428571429,
      "grad_norm": 4.298236902242096,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 1.0764,
      "step": 18
    },
    {
      "epoch": 0.03619047619047619,
      "grad_norm": 3.7114863245838463,
      "learning_rate": 3.958333333333333e-06,
      "loss": 1.1386,
      "step": 19
    },
    {
      "epoch": 0.0380952380952381,
      "grad_norm": 3.9206535517036656,
      "learning_rate": 4.166666666666667e-06,
      "loss": 1.0228,
      "step": 20
    },
    {
      "epoch": 0.04,
      "grad_norm": 3.7918216997086343,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 1.1698,
      "step": 21
    },
    {
      "epoch": 0.0419047619047619,
      "grad_norm": 4.133208891764793,
      "learning_rate": 4.583333333333333e-06,
      "loss": 1.0701,
      "step": 22
    },
    {
      "epoch": 0.04380952380952381,
      "grad_norm": 3.3540327254448496,
      "learning_rate": 4.791666666666668e-06,
      "loss": 1.1161,
      "step": 23
    },
    {
      "epoch": 0.045714285714285714,
      "grad_norm": 4.539012815101767,
      "learning_rate": 5e-06,
      "loss": 1.1009,
      "step": 24
    },
    {
      "epoch": 0.047619047619047616,
      "grad_norm": 3.4906605922972735,
      "learning_rate": 5.208333333333334e-06,
      "loss": 1.0132,
      "step": 25
    },
    {
      "epoch": 0.049523809523809526,
      "grad_norm": 3.0938294959210357,
      "learning_rate": 5.416666666666667e-06,
      "loss": 1.0779,
      "step": 26
    },
    {
      "epoch": 0.05142857142857143,
      "grad_norm": 3.706381579068903,
      "learning_rate": 5.625e-06,
      "loss": 1.0345,
      "step": 27
    },
    {
      "epoch": 0.05333333333333334,
      "grad_norm": 3.0374300244693315,
      "learning_rate": 5.833333333333334e-06,
      "loss": 1.1701,
      "step": 28
    },
    {
      "epoch": 0.05523809523809524,
      "grad_norm": 3.6018268851222794,
      "learning_rate": 6.041666666666667e-06,
      "loss": 1.0996,
      "step": 29
    },
    {
      "epoch": 0.05714285714285714,
      "grad_norm": 4.078241688010774,
      "learning_rate": 6.25e-06,
      "loss": 1.0101,
      "step": 30
    },
    {
      "epoch": 0.05904761904761905,
      "grad_norm": 2.5532194989459005,
      "learning_rate": 6.458333333333334e-06,
      "loss": 1.0543,
      "step": 31
    },
    {
      "epoch": 0.06095238095238095,
      "grad_norm": 3.016442373029697,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.1279,
      "step": 32
    },
    {
      "epoch": 0.06285714285714286,
      "grad_norm": 3.6902114984607413,
      "learning_rate": 6.875e-06,
      "loss": 0.9879,
      "step": 33
    },
    {
      "epoch": 0.06476190476190476,
      "grad_norm": 2.7977555469861115,
      "learning_rate": 7.083333333333335e-06,
      "loss": 0.8825,
      "step": 34
    },
    {
      "epoch": 0.06666666666666667,
      "grad_norm": 3.510142851115604,
      "learning_rate": 7.291666666666667e-06,
      "loss": 1.023,
      "step": 35
    },
    {
      "epoch": 0.06857142857142857,
      "grad_norm": 2.8452663535625113,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.9159,
      "step": 36
    },
    {
      "epoch": 0.07047619047619047,
      "grad_norm": 3.148967190761688,
      "learning_rate": 7.708333333333334e-06,
      "loss": 0.9985,
      "step": 37
    },
    {
      "epoch": 0.07238095238095238,
      "grad_norm": 4.712123502415102,
      "learning_rate": 7.916666666666667e-06,
      "loss": 1.2326,
      "step": 38
    },
    {
      "epoch": 0.07428571428571429,
      "grad_norm": 3.653286393810381,
      "learning_rate": 8.125000000000001e-06,
      "loss": 0.9915,
      "step": 39
    },
    {
      "epoch": 0.0761904761904762,
      "grad_norm": 3.0858499783164475,
      "learning_rate": 8.333333333333334e-06,
      "loss": 1.1663,
      "step": 40
    },
    {
      "epoch": 0.07809523809523809,
      "grad_norm": 3.5143089698882473,
      "learning_rate": 8.541666666666666e-06,
      "loss": 1.0356,
      "step": 41
    },
    {
      "epoch": 0.08,
      "grad_norm": 3.2811146063203975,
      "learning_rate": 8.750000000000001e-06,
      "loss": 1.0228,
      "step": 42
    },
    {
      "epoch": 0.08190476190476191,
      "grad_norm": 3.3011687607119127,
      "learning_rate": 8.958333333333334e-06,
      "loss": 1.1049,
      "step": 43
    },
    {
      "epoch": 0.0838095238095238,
      "grad_norm": 3.210258584400171,
      "learning_rate": 9.166666666666666e-06,
      "loss": 0.8777,
      "step": 44
    },
    {
      "epoch": 0.08571428571428572,
      "grad_norm": 3.441412832078474,
      "learning_rate": 9.375000000000001e-06,
      "loss": 1.137,
      "step": 45
    },
    {
      "epoch": 0.08761904761904762,
      "grad_norm": 2.5508244468454624,
      "learning_rate": 9.583333333333335e-06,
      "loss": 0.998,
      "step": 46
    },
    {
      "epoch": 0.08952380952380952,
      "grad_norm": 3.093215332044739,
      "learning_rate": 9.791666666666666e-06,
      "loss": 1.0617,
      "step": 47
    },
    {
      "epoch": 0.09142857142857143,
      "grad_norm": 3.2469121174531073,
      "learning_rate": 1e-05,
      "loss": 0.8773,
      "step": 48
    },
    {
      "epoch": 0.09333333333333334,
      "grad_norm": 3.4805042775317765,
      "learning_rate": 9.99998941815181e-06,
      "loss": 1.1255,
      "step": 49
    },
    {
      "epoch": 0.09523809523809523,
      "grad_norm": 3.1642573627760986,
      "learning_rate": 9.999957672652028e-06,
      "loss": 1.1809,
      "step": 50
    },
    {
      "epoch": 0.09714285714285714,
      "grad_norm": 2.586521545518349,
      "learning_rate": 9.999904763635027e-06,
      "loss": 0.8819,
      "step": 51
    },
    {
      "epoch": 0.09904761904761905,
      "grad_norm": 3.1568432074306565,
      "learning_rate": 9.999830691324755e-06,
      "loss": 1.0084,
      "step": 52
    },
    {
      "epoch": 0.10095238095238095,
      "grad_norm": 3.4278518629452703,
      "learning_rate": 9.99973545603474e-06,
      "loss": 1.1908,
      "step": 53
    },
    {
      "epoch": 0.10285714285714286,
      "grad_norm": 3.1248779175425794,
      "learning_rate": 9.99961905816809e-06,
      "loss": 1.0985,
      "step": 54
    },
    {
      "epoch": 0.10476190476190476,
      "grad_norm": 2.7546079451843615,
      "learning_rate": 9.999481498217488e-06,
      "loss": 0.9914,
      "step": 55
    },
    {
      "epoch": 0.10666666666666667,
      "grad_norm": 3.038797998164753,
      "learning_rate": 9.999322776765187e-06,
      "loss": 0.9948,
      "step": 56
    },
    {
      "epoch": 0.10857142857142857,
      "grad_norm": 3.5025783028380375,
      "learning_rate": 9.999142894483013e-06,
      "loss": 1.2092,
      "step": 57
    },
    {
      "epoch": 0.11047619047619048,
      "grad_norm": 2.9501991103847724,
      "learning_rate": 9.998941852132364e-06,
      "loss": 1.1009,
      "step": 58
    },
    {
      "epoch": 0.11238095238095239,
      "grad_norm": 3.0826054414101063,
      "learning_rate": 9.998719650564197e-06,
      "loss": 1.089,
      "step": 59
    },
    {
      "epoch": 0.11428571428571428,
      "grad_norm": 3.3237722878132065,
      "learning_rate": 9.998476290719035e-06,
      "loss": 1.0187,
      "step": 60
    },
    {
      "epoch": 0.11619047619047619,
      "grad_norm": 3.2244788165958256,
      "learning_rate": 9.998211773626955e-06,
      "loss": 1.0183,
      "step": 61
    },
    {
      "epoch": 0.1180952380952381,
      "grad_norm": 3.3388884723048493,
      "learning_rate": 9.99792610040759e-06,
      "loss": 0.9636,
      "step": 62
    },
    {
      "epoch": 0.12,
      "grad_norm": 3.1562420302292793,
      "learning_rate": 9.99761927227012e-06,
      "loss": 0.9614,
      "step": 63
    },
    {
      "epoch": 0.1219047619047619,
      "grad_norm": 3.267673603904302,
      "learning_rate": 9.997291290513268e-06,
      "loss": 1.0174,
      "step": 64
    },
    {
      "epoch": 0.12380952380952381,
      "grad_norm": 2.8273452169820876,
      "learning_rate": 9.996942156525298e-06,
      "loss": 1.1283,
      "step": 65
    },
    {
      "epoch": 0.12571428571428572,
      "grad_norm": 3.8370109921118014,
      "learning_rate": 9.996571871784e-06,
      "loss": 0.9912,
      "step": 66
    },
    {
      "epoch": 0.12761904761904763,
      "grad_norm": 2.700249267077012,
      "learning_rate": 9.996180437856695e-06,
      "loss": 1.0158,
      "step": 67
    },
    {
      "epoch": 0.1295238095238095,
      "grad_norm": 3.3193444159035566,
      "learning_rate": 9.995767856400218e-06,
      "loss": 1.1443,
      "step": 68
    },
    {
      "epoch": 0.13142857142857142,
      "grad_norm": 2.5774813529521414,
      "learning_rate": 9.99533412916092e-06,
      "loss": 1.0178,
      "step": 69
    },
    {
      "epoch": 0.13333333333333333,
      "grad_norm": 2.9479981129016433,
      "learning_rate": 9.994879257974657e-06,
      "loss": 1.0722,
      "step": 70
    },
    {
      "epoch": 0.13523809523809524,
      "grad_norm": 2.886095673232496,
      "learning_rate": 9.994403244766779e-06,
      "loss": 0.9772,
      "step": 71
    },
    {
      "epoch": 0.13714285714285715,
      "grad_norm": 2.660783801859094,
      "learning_rate": 9.993906091552124e-06,
      "loss": 1.091,
      "step": 72
    },
    {
      "epoch": 0.13904761904761906,
      "grad_norm": 3.084097105820264,
      "learning_rate": 9.993387800435015e-06,
      "loss": 1.1529,
      "step": 73
    },
    {
      "epoch": 0.14095238095238094,
      "grad_norm": 4.62363254320876,
      "learning_rate": 9.992848373609241e-06,
      "loss": 0.9261,
      "step": 74
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 2.9860719936761053,
      "learning_rate": 9.992287813358057e-06,
      "loss": 1.0594,
      "step": 75
    },
    {
      "epoch": 0.14476190476190476,
      "grad_norm": 2.691342320793716,
      "learning_rate": 9.991706122054166e-06,
      "loss": 1.1088,
      "step": 76
    },
    {
      "epoch": 0.14666666666666667,
      "grad_norm": 2.9594318822617596,
      "learning_rate": 9.991103302159717e-06,
      "loss": 1.0433,
      "step": 77
    },
    {
      "epoch": 0.14857142857142858,
      "grad_norm": 3.153330979738241,
      "learning_rate": 9.990479356226289e-06,
      "loss": 1.2322,
      "step": 78
    },
    {
      "epoch": 0.15047619047619049,
      "grad_norm": 3.3307200756954303,
      "learning_rate": 9.989834286894882e-06,
      "loss": 0.9688,
      "step": 79
    },
    {
      "epoch": 0.1523809523809524,
      "grad_norm": 2.858366771965065,
      "learning_rate": 9.989168096895909e-06,
      "loss": 1.1716,
      "step": 80
    },
    {
      "epoch": 0.15428571428571428,
      "grad_norm": 3.530158228069321,
      "learning_rate": 9.988480789049176e-06,
      "loss": 1.0263,
      "step": 81
    },
    {
      "epoch": 0.15619047619047619,
      "grad_norm": 3.698678311272925,
      "learning_rate": 9.987772366263876e-06,
      "loss": 1.0112,
      "step": 82
    },
    {
      "epoch": 0.1580952380952381,
      "grad_norm": 3.9611270751486485,
      "learning_rate": 9.987042831538585e-06,
      "loss": 1.119,
      "step": 83
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.3068891014783226,
      "learning_rate": 9.986292187961227e-06,
      "loss": 1.083,
      "step": 84
    },
    {
      "epoch": 0.1619047619047619,
      "grad_norm": 4.071158215091613,
      "learning_rate": 9.98552043870908e-06,
      "loss": 1.0478,
      "step": 85
    },
    {
      "epoch": 0.16380952380952382,
      "grad_norm": 3.7596202230204834,
      "learning_rate": 9.98472758704876e-06,
      "loss": 0.9647,
      "step": 86
    },
    {
      "epoch": 0.1657142857142857,
      "grad_norm": 3.254331891200369,
      "learning_rate": 9.983913636336204e-06,
      "loss": 1.215,
      "step": 87
    },
    {
      "epoch": 0.1676190476190476,
      "grad_norm": 3.607528562832131,
      "learning_rate": 9.983078590016648e-06,
      "loss": 0.964,
      "step": 88
    },
    {
      "epoch": 0.16952380952380952,
      "grad_norm": 3.4095129182391,
      "learning_rate": 9.98222245162463e-06,
      "loss": 1.0764,
      "step": 89
    },
    {
      "epoch": 0.17142857142857143,
      "grad_norm": 3.1120779231959905,
      "learning_rate": 9.981345224783955e-06,
      "loss": 1.0883,
      "step": 90
    },
    {
      "epoch": 0.17333333333333334,
      "grad_norm": 3.4225409802187063,
      "learning_rate": 9.9804469132077e-06,
      "loss": 1.0097,
      "step": 91
    },
    {
      "epoch": 0.17523809523809525,
      "grad_norm": 4.701654702406365,
      "learning_rate": 9.979527520698182e-06,
      "loss": 1.0826,
      "step": 92
    },
    {
      "epoch": 0.17714285714285713,
      "grad_norm": 3.259130195982121,
      "learning_rate": 9.97858705114695e-06,
      "loss": 1.0748,
      "step": 93
    },
    {
      "epoch": 0.17904761904761904,
      "grad_norm": 3.0639511082253965,
      "learning_rate": 9.977625508534767e-06,
      "loss": 1.1188,
      "step": 94
    },
    {
      "epoch": 0.18095238095238095,
      "grad_norm": 2.932982742967428,
      "learning_rate": 9.976642896931592e-06,
      "loss": 1.0619,
      "step": 95
    },
    {
      "epoch": 0.18285714285714286,
      "grad_norm": 2.9226827615732605,
      "learning_rate": 9.975639220496564e-06,
      "loss": 1.02,
      "step": 96
    },
    {
      "epoch": 0.18476190476190477,
      "grad_norm": 2.364194910815137,
      "learning_rate": 9.974614483477983e-06,
      "loss": 1.0915,
      "step": 97
    },
    {
      "epoch": 0.18666666666666668,
      "grad_norm": 2.965291909603285,
      "learning_rate": 9.973568690213292e-06,
      "loss": 1.0658,
      "step": 98
    },
    {
      "epoch": 0.18857142857142858,
      "grad_norm": 3.183868457710933,
      "learning_rate": 9.972501845129064e-06,
      "loss": 1.2201,
      "step": 99
    },
    {
      "epoch": 0.19047619047619047,
      "grad_norm": 3.239515554221448,
      "learning_rate": 9.971413952740975e-06,
      "loss": 1.2212,
      "step": 100
    },
    {
      "epoch": 0.19238095238095237,
      "grad_norm": 4.114193945478856,
      "learning_rate": 9.97030501765379e-06,
      "loss": 1.1717,
      "step": 101
    },
    {
      "epoch": 0.19428571428571428,
      "grad_norm": 2.4733859830692415,
      "learning_rate": 9.96917504456134e-06,
      "loss": 1.0785,
      "step": 102
    },
    {
      "epoch": 0.1961904761904762,
      "grad_norm": 2.6776931241814,
      "learning_rate": 9.968024038246511e-06,
      "loss": 1.0503,
      "step": 103
    },
    {
      "epoch": 0.1980952380952381,
      "grad_norm": 2.8687433329822376,
      "learning_rate": 9.966852003581211e-06,
      "loss": 1.1634,
      "step": 104
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.6125526624258764,
      "learning_rate": 9.965658945526354e-06,
      "loss": 0.9906,
      "step": 105
    },
    {
      "epoch": 0.2019047619047619,
      "grad_norm": 3.344534458296087,
      "learning_rate": 9.964444869131846e-06,
      "loss": 1.085,
      "step": 106
    },
    {
      "epoch": 0.2038095238095238,
      "grad_norm": 3.222299641905838,
      "learning_rate": 9.963209779536559e-06,
      "loss": 0.9999,
      "step": 107
    },
    {
      "epoch": 0.2057142857142857,
      "grad_norm": 3.0703693594229735,
      "learning_rate": 9.961953681968296e-06,
      "loss": 0.9924,
      "step": 108
    },
    {
      "epoch": 0.20761904761904762,
      "grad_norm": 2.8728815593750983,
      "learning_rate": 9.960676581743802e-06,
      "loss": 1.0864,
      "step": 109
    },
    {
      "epoch": 0.20952380952380953,
      "grad_norm": 3.2520040087112676,
      "learning_rate": 9.959378484268702e-06,
      "loss": 1.032,
      "step": 110
    },
    {
      "epoch": 0.21142857142857144,
      "grad_norm": 2.848023660479766,
      "learning_rate": 9.958059395037505e-06,
      "loss": 0.9938,
      "step": 111
    },
    {
      "epoch": 0.21333333333333335,
      "grad_norm": 3.03683212722849,
      "learning_rate": 9.956719319633574e-06,
      "loss": 1.052,
      "step": 112
    },
    {
      "epoch": 0.21523809523809523,
      "grad_norm": 3.534186781153816,
      "learning_rate": 9.955358263729098e-06,
      "loss": 1.0652,
      "step": 113
    },
    {
      "epoch": 0.21714285714285714,
      "grad_norm": 2.8102076591175082,
      "learning_rate": 9.95397623308507e-06,
      "loss": 0.9547,
      "step": 114
    },
    {
      "epoch": 0.21904761904761905,
      "grad_norm": 2.3182734267227882,
      "learning_rate": 9.952573233551268e-06,
      "loss": 1.3116,
      "step": 115
    },
    {
      "epoch": 0.22095238095238096,
      "grad_norm": 3.689992437884337,
      "learning_rate": 9.951149271066221e-06,
      "loss": 0.9639,
      "step": 116
    },
    {
      "epoch": 0.22285714285714286,
      "grad_norm": 3.1117182277237965,
      "learning_rate": 9.949704351657192e-06,
      "loss": 1.0475,
      "step": 117
    },
    {
      "epoch": 0.22476190476190477,
      "grad_norm": 3.2894246016451336,
      "learning_rate": 9.948238481440149e-06,
      "loss": 0.996,
      "step": 118
    },
    {
      "epoch": 0.22666666666666666,
      "grad_norm": 3.095497763452724,
      "learning_rate": 9.946751666619736e-06,
      "loss": 0.9967,
      "step": 119
    },
    {
      "epoch": 0.22857142857142856,
      "grad_norm": 2.5824494132376974,
      "learning_rate": 9.945243913489255e-06,
      "loss": 1.0582,
      "step": 120
    },
    {
      "epoch": 0.23047619047619047,
      "grad_norm": 3.4476576109467914,
      "learning_rate": 9.943715228430632e-06,
      "loss": 1.0397,
      "step": 121
    },
    {
      "epoch": 0.23238095238095238,
      "grad_norm": 4.299143325553733,
      "learning_rate": 9.94216561791439e-06,
      "loss": 1.4879,
      "step": 122
    },
    {
      "epoch": 0.2342857142857143,
      "grad_norm": 2.6941969303087703,
      "learning_rate": 9.940595088499628e-06,
      "loss": 1.1719,
      "step": 123
    },
    {
      "epoch": 0.2361904761904762,
      "grad_norm": 2.9075149270426968,
      "learning_rate": 9.939003646833985e-06,
      "loss": 1.0605,
      "step": 124
    },
    {
      "epoch": 0.23809523809523808,
      "grad_norm": 2.2748068715580665,
      "learning_rate": 9.937391299653622e-06,
      "loss": 1.0057,
      "step": 125
    },
    {
      "epoch": 0.24,
      "grad_norm": 3.8992899892410335,
      "learning_rate": 9.935758053783183e-06,
      "loss": 1.0813,
      "step": 126
    },
    {
      "epoch": 0.2419047619047619,
      "grad_norm": 3.0096302665768326,
      "learning_rate": 9.934103916135772e-06,
      "loss": 1.031,
      "step": 127
    },
    {
      "epoch": 0.2438095238095238,
      "grad_norm": 3.531846236903906,
      "learning_rate": 9.932428893712923e-06,
      "loss": 1.1547,
      "step": 128
    },
    {
      "epoch": 0.24571428571428572,
      "grad_norm": 3.391883343568946,
      "learning_rate": 9.930732993604566e-06,
      "loss": 1.0361,
      "step": 129
    },
    {
      "epoch": 0.24761904761904763,
      "grad_norm": 3.5977577229850755,
      "learning_rate": 9.929016222989009e-06,
      "loss": 1.0023,
      "step": 130
    },
    {
      "epoch": 0.24952380952380954,
      "grad_norm": 3.7571434809125153,
      "learning_rate": 9.927278589132891e-06,
      "loss": 1.1339,
      "step": 131
    },
    {
      "epoch": 0.25142857142857145,
      "grad_norm": 2.5261761857089344,
      "learning_rate": 9.925520099391164e-06,
      "loss": 1.105,
      "step": 132
    },
    {
      "epoch": 0.25333333333333335,
      "grad_norm": 2.518620573737334,
      "learning_rate": 9.923740761207055e-06,
      "loss": 0.965,
      "step": 133
    },
    {
      "epoch": 0.25523809523809526,
      "grad_norm": 3.374631413724398,
      "learning_rate": 9.921940582112041e-06,
      "loss": 1.0943,
      "step": 134
    },
    {
      "epoch": 0.2571428571428571,
      "grad_norm": 2.8350601779277236,
      "learning_rate": 9.920119569725811e-06,
      "loss": 1.0307,
      "step": 135
    },
    {
      "epoch": 0.259047619047619,
      "grad_norm": 3.078359484666857,
      "learning_rate": 9.918277731756234e-06,
      "loss": 1.2088,
      "step": 136
    },
    {
      "epoch": 0.26095238095238094,
      "grad_norm": 2.8683287365781154,
      "learning_rate": 9.91641507599933e-06,
      "loss": 0.9529,
      "step": 137
    },
    {
      "epoch": 0.26285714285714284,
      "grad_norm": 2.3865571133140207,
      "learning_rate": 9.914531610339236e-06,
      "loss": 1.1474,
      "step": 138
    },
    {
      "epoch": 0.26476190476190475,
      "grad_norm": 2.7123759498479436,
      "learning_rate": 9.912627342748168e-06,
      "loss": 1.045,
      "step": 139
    },
    {
      "epoch": 0.26666666666666666,
      "grad_norm": 3.4343711498632015,
      "learning_rate": 9.910702281286401e-06,
      "loss": 1.0996,
      "step": 140
    },
    {
      "epoch": 0.26857142857142857,
      "grad_norm": 3.3983243671309165,
      "learning_rate": 9.908756434102212e-06,
      "loss": 1.141,
      "step": 141
    },
    {
      "epoch": 0.2704761904761905,
      "grad_norm": 2.9951907358912306,
      "learning_rate": 9.906789809431867e-06,
      "loss": 0.9749,
      "step": 142
    },
    {
      "epoch": 0.2723809523809524,
      "grad_norm": 2.548630418636751,
      "learning_rate": 9.904802415599578e-06,
      "loss": 1.1229,
      "step": 143
    },
    {
      "epoch": 0.2742857142857143,
      "grad_norm": 3.1479147398397562,
      "learning_rate": 9.90279426101746e-06,
      "loss": 0.9784,
      "step": 144
    },
    {
      "epoch": 0.2761904761904762,
      "grad_norm": 2.9129195354416386,
      "learning_rate": 9.900765354185508e-06,
      "loss": 1.1494,
      "step": 145
    },
    {
      "epoch": 0.2780952380952381,
      "grad_norm": 3.021162287467609,
      "learning_rate": 9.898715703691562e-06,
      "loss": 1.075,
      "step": 146
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.9927487776012365,
      "learning_rate": 9.896645318211251e-06,
      "loss": 1.0392,
      "step": 147
    },
    {
      "epoch": 0.2819047619047619,
      "grad_norm": 3.2610420257422432,
      "learning_rate": 9.894554206507982e-06,
      "loss": 0.9937,
      "step": 148
    },
    {
      "epoch": 0.2838095238095238,
      "grad_norm": 2.9699607268831065,
      "learning_rate": 9.892442377432882e-06,
      "loss": 1.1378,
      "step": 149
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 2.705524552687109,
      "learning_rate": 9.890309839924776e-06,
      "loss": 1.0683,
      "step": 150
    },
    {
      "epoch": 0.2876190476190476,
      "grad_norm": 3.3715417503889777,
      "learning_rate": 9.888156603010137e-06,
      "loss": 1.0635,
      "step": 151
    },
    {
      "epoch": 0.2895238095238095,
      "grad_norm": 2.739638304166437,
      "learning_rate": 9.885982675803056e-06,
      "loss": 0.9561,
      "step": 152
    },
    {
      "epoch": 0.2914285714285714,
      "grad_norm": 2.4241137147091987,
      "learning_rate": 9.8837880675052e-06,
      "loss": 1.1199,
      "step": 153
    },
    {
      "epoch": 0.29333333333333333,
      "grad_norm": 3.2548663265967885,
      "learning_rate": 9.881572787405774e-06,
      "loss": 1.0592,
      "step": 154
    },
    {
      "epoch": 0.29523809523809524,
      "grad_norm": 2.8847430466560593,
      "learning_rate": 9.87933684488148e-06,
      "loss": 1.0127,
      "step": 155
    },
    {
      "epoch": 0.29714285714285715,
      "grad_norm": 4.163773017576036,
      "learning_rate": 9.877080249396482e-06,
      "loss": 1.1959,
      "step": 156
    },
    {
      "epoch": 0.29904761904761906,
      "grad_norm": 2.43449383239491,
      "learning_rate": 9.874803010502359e-06,
      "loss": 0.9392,
      "step": 157
    },
    {
      "epoch": 0.30095238095238097,
      "grad_norm": 3.192868562187782,
      "learning_rate": 9.872505137838069e-06,
      "loss": 1.1568,
      "step": 158
    },
    {
      "epoch": 0.3028571428571429,
      "grad_norm": 3.4387041337301625,
      "learning_rate": 9.87018664112991e-06,
      "loss": 1.0076,
      "step": 159
    },
    {
      "epoch": 0.3047619047619048,
      "grad_norm": 2.645228572202429,
      "learning_rate": 9.867847530191472e-06,
      "loss": 1.1138,
      "step": 160
    },
    {
      "epoch": 0.30666666666666664,
      "grad_norm": 2.7359173146917413,
      "learning_rate": 9.8654878149236e-06,
      "loss": 1.0449,
      "step": 161
    },
    {
      "epoch": 0.30857142857142855,
      "grad_norm": 2.9870908289014735,
      "learning_rate": 9.863107505314359e-06,
      "loss": 1.0193,
      "step": 162
    },
    {
      "epoch": 0.31047619047619046,
      "grad_norm": 2.574232205962119,
      "learning_rate": 9.860706611438975e-06,
      "loss": 1.082,
      "step": 163
    },
    {
      "epoch": 0.31238095238095237,
      "grad_norm": 2.986482890199796,
      "learning_rate": 9.858285143459804e-06,
      "loss": 0.93,
      "step": 164
    },
    {
      "epoch": 0.3142857142857143,
      "grad_norm": 3.536206243869805,
      "learning_rate": 9.855843111626295e-06,
      "loss": 1.1111,
      "step": 165
    },
    {
      "epoch": 0.3161904761904762,
      "grad_norm": 2.686673444877999,
      "learning_rate": 9.853380526274928e-06,
      "loss": 1.2395,
      "step": 166
    },
    {
      "epoch": 0.3180952380952381,
      "grad_norm": 4.611197609530781,
      "learning_rate": 9.850897397829182e-06,
      "loss": 1.1801,
      "step": 167
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.911101177768665,
      "learning_rate": 9.848393736799496e-06,
      "loss": 1.1439,
      "step": 168
    },
    {
      "epoch": 0.3219047619047619,
      "grad_norm": 2.422871904775949,
      "learning_rate": 9.845869553783214e-06,
      "loss": 1.0528,
      "step": 169
    },
    {
      "epoch": 0.3238095238095238,
      "grad_norm": 2.6621360710125095,
      "learning_rate": 9.843324859464547e-06,
      "loss": 1.0423,
      "step": 170
    },
    {
      "epoch": 0.32571428571428573,
      "grad_norm": 3.7362699560222805,
      "learning_rate": 9.840759664614516e-06,
      "loss": 1.1698,
      "step": 171
    },
    {
      "epoch": 0.32761904761904764,
      "grad_norm": 2.1733989857511644,
      "learning_rate": 9.838173980090928e-06,
      "loss": 1.0468,
      "step": 172
    },
    {
      "epoch": 0.3295238095238095,
      "grad_norm": 3.018576608971308,
      "learning_rate": 9.835567816838309e-06,
      "loss": 0.9909,
      "step": 173
    },
    {
      "epoch": 0.3314285714285714,
      "grad_norm": 3.2360372541399913,
      "learning_rate": 9.832941185887868e-06,
      "loss": 1.0334,
      "step": 174
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 3.5138295004700524,
      "learning_rate": 9.83029409835745e-06,
      "loss": 1.1439,
      "step": 175
    },
    {
      "epoch": 0.3352380952380952,
      "grad_norm": 2.596761327319496,
      "learning_rate": 9.827626565451487e-06,
      "loss": 0.9764,
      "step": 176
    },
    {
      "epoch": 0.33714285714285713,
      "grad_norm": 3.617953130813023,
      "learning_rate": 9.82493859846095e-06,
      "loss": 1.0809,
      "step": 177
    },
    {
      "epoch": 0.33904761904761904,
      "grad_norm": 2.6372949321239822,
      "learning_rate": 9.8222302087633e-06,
      "loss": 0.9211,
      "step": 178
    },
    {
      "epoch": 0.34095238095238095,
      "grad_norm": 3.00182530804867,
      "learning_rate": 9.819501407822446e-06,
      "loss": 1.1826,
      "step": 179
    },
    {
      "epoch": 0.34285714285714286,
      "grad_norm": 2.816028166000998,
      "learning_rate": 9.816752207188695e-06,
      "loss": 0.9066,
      "step": 180
    },
    {
      "epoch": 0.34476190476190477,
      "grad_norm": 3.566149344336227,
      "learning_rate": 9.81398261849869e-06,
      "loss": 1.013,
      "step": 181
    },
    {
      "epoch": 0.3466666666666667,
      "grad_norm": 2.6508129562771887,
      "learning_rate": 9.811192653475382e-06,
      "loss": 1.0807,
      "step": 182
    },
    {
      "epoch": 0.3485714285714286,
      "grad_norm": 2.403667822417291,
      "learning_rate": 9.808382323927962e-06,
      "loss": 0.8976,
      "step": 183
    },
    {
      "epoch": 0.3504761904761905,
      "grad_norm": 2.5863595375751274,
      "learning_rate": 9.805551641751826e-06,
      "loss": 1.0593,
      "step": 184
    },
    {
      "epoch": 0.3523809523809524,
      "grad_norm": 2.49618027877372,
      "learning_rate": 9.802700618928513e-06,
      "loss": 1.178,
      "step": 185
    },
    {
      "epoch": 0.35428571428571426,
      "grad_norm": 3.4940607815837548,
      "learning_rate": 9.799829267525656e-06,
      "loss": 0.9083,
      "step": 186
    },
    {
      "epoch": 0.35619047619047617,
      "grad_norm": 2.4823680559712673,
      "learning_rate": 9.79693759969694e-06,
      "loss": 0.9226,
      "step": 187
    },
    {
      "epoch": 0.3580952380952381,
      "grad_norm": 2.8744244733166857,
      "learning_rate": 9.794025627682042e-06,
      "loss": 0.9735,
      "step": 188
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.3597483861310424,
      "learning_rate": 9.791093363806576e-06,
      "loss": 1.0966,
      "step": 189
    },
    {
      "epoch": 0.3619047619047619,
      "grad_norm": 3.3794514928646207,
      "learning_rate": 9.788140820482056e-06,
      "loss": 1.1086,
      "step": 190
    },
    {
      "epoch": 0.3638095238095238,
      "grad_norm": 2.237115025130188,
      "learning_rate": 9.785168010205822e-06,
      "loss": 0.9749,
      "step": 191
    },
    {
      "epoch": 0.3657142857142857,
      "grad_norm": 2.612924523749599,
      "learning_rate": 9.782174945561009e-06,
      "loss": 0.9963,
      "step": 192
    },
    {
      "epoch": 0.3676190476190476,
      "grad_norm": 3.4133183920646397,
      "learning_rate": 9.77916163921648e-06,
      "loss": 1.1557,
      "step": 193
    },
    {
      "epoch": 0.36952380952380953,
      "grad_norm": 3.1580639931199945,
      "learning_rate": 9.77612810392677e-06,
      "loss": 1.2647,
      "step": 194
    },
    {
      "epoch": 0.37142857142857144,
      "grad_norm": 2.678879647534476,
      "learning_rate": 9.773074352532048e-06,
      "loss": 1.0786,
      "step": 195
    },
    {
      "epoch": 0.37333333333333335,
      "grad_norm": 2.775674376589696,
      "learning_rate": 9.770000397958045e-06,
      "loss": 1.0459,
      "step": 196
    },
    {
      "epoch": 0.37523809523809526,
      "grad_norm": 3.302196763192064,
      "learning_rate": 9.766906253216011e-06,
      "loss": 1.0103,
      "step": 197
    },
    {
      "epoch": 0.37714285714285717,
      "grad_norm": 2.9143701416841172,
      "learning_rate": 9.763791931402652e-06,
      "loss": 1.1111,
      "step": 198
    },
    {
      "epoch": 0.379047619047619,
      "grad_norm": 2.291599732156606,
      "learning_rate": 9.760657445700082e-06,
      "loss": 1.0406,
      "step": 199
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 3.189432428360578,
      "learning_rate": 9.757502809375761e-06,
      "loss": 1.1519,
      "step": 200
    },
    {
      "epoch": 0.38285714285714284,
      "grad_norm": 2.6118089279457126,
      "learning_rate": 9.754328035782442e-06,
      "loss": 1.0255,
      "step": 201
    },
    {
      "epoch": 0.38476190476190475,
      "grad_norm": 3.3669356516916578,
      "learning_rate": 9.751133138358111e-06,
      "loss": 1.1824,
      "step": 202
    },
    {
      "epoch": 0.38666666666666666,
      "grad_norm": 2.791504768510238,
      "learning_rate": 9.747918130625941e-06,
      "loss": 1.0649,
      "step": 203
    },
    {
      "epoch": 0.38857142857142857,
      "grad_norm": 2.557484611681563,
      "learning_rate": 9.744683026194221e-06,
      "loss": 0.9955,
      "step": 204
    },
    {
      "epoch": 0.3904761904761905,
      "grad_norm": 2.9051992741904074,
      "learning_rate": 9.7414278387563e-06,
      "loss": 0.9927,
      "step": 205
    },
    {
      "epoch": 0.3923809523809524,
      "grad_norm": 3.1402167980474696,
      "learning_rate": 9.738152582090543e-06,
      "loss": 1.0115,
      "step": 206
    },
    {
      "epoch": 0.3942857142857143,
      "grad_norm": 2.6748264866511993,
      "learning_rate": 9.734857270060254e-06,
      "loss": 1.1955,
      "step": 207
    },
    {
      "epoch": 0.3961904761904762,
      "grad_norm": 5.605572655010524,
      "learning_rate": 9.731541916613631e-06,
      "loss": 0.9863,
      "step": 208
    },
    {
      "epoch": 0.3980952380952381,
      "grad_norm": 4.49082904136764,
      "learning_rate": 9.7282065357837e-06,
      "loss": 1.3237,
      "step": 209
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.017378218982703,
      "learning_rate": 9.724851141688258e-06,
      "loss": 1.2822,
      "step": 210
    },
    {
      "epoch": 0.40190476190476193,
      "grad_norm": 3.4271261376555286,
      "learning_rate": 9.721475748529815e-06,
      "loss": 1.135,
      "step": 211
    },
    {
      "epoch": 0.4038095238095238,
      "grad_norm": 2.8093618506873317,
      "learning_rate": 9.71808037059553e-06,
      "loss": 1.0961,
      "step": 212
    },
    {
      "epoch": 0.4057142857142857,
      "grad_norm": 3.9587987374950617,
      "learning_rate": 9.714665022257152e-06,
      "loss": 1.2021,
      "step": 213
    },
    {
      "epoch": 0.4076190476190476,
      "grad_norm": 3.235909397764408,
      "learning_rate": 9.711229717970961e-06,
      "loss": 1.1709,
      "step": 214
    },
    {
      "epoch": 0.4095238095238095,
      "grad_norm": 2.9294343889888976,
      "learning_rate": 9.707774472277702e-06,
      "loss": 1.1405,
      "step": 215
    },
    {
      "epoch": 0.4114285714285714,
      "grad_norm": 2.8743119961776658,
      "learning_rate": 9.704299299802531e-06,
      "loss": 1.2047,
      "step": 216
    },
    {
      "epoch": 0.41333333333333333,
      "grad_norm": 3.740667872605965,
      "learning_rate": 9.700804215254946e-06,
      "loss": 1.0581,
      "step": 217
    },
    {
      "epoch": 0.41523809523809524,
      "grad_norm": 3.490121897988836,
      "learning_rate": 9.69728923342873e-06,
      "loss": 1.1561,
      "step": 218
    },
    {
      "epoch": 0.41714285714285715,
      "grad_norm": 3.3639327787089846,
      "learning_rate": 9.693754369201885e-06,
      "loss": 1.1634,
      "step": 219
    },
    {
      "epoch": 0.41904761904761906,
      "grad_norm": 3.0200761314830262,
      "learning_rate": 9.690199637536566e-06,
      "loss": 1.1064,
      "step": 220
    },
    {
      "epoch": 0.42095238095238097,
      "grad_norm": 3.5667974140280463,
      "learning_rate": 9.68662505347903e-06,
      "loss": 1.1377,
      "step": 221
    },
    {
      "epoch": 0.4228571428571429,
      "grad_norm": 3.7220254675880025,
      "learning_rate": 9.683030632159556e-06,
      "loss": 1.1443,
      "step": 222
    },
    {
      "epoch": 0.4247619047619048,
      "grad_norm": 2.595476925697928,
      "learning_rate": 9.679416388792393e-06,
      "loss": 1.092,
      "step": 223
    },
    {
      "epoch": 0.4266666666666667,
      "grad_norm": 3.853813921258193,
      "learning_rate": 9.675782338675692e-06,
      "loss": 1.0147,
      "step": 224
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 3.471695450855635,
      "learning_rate": 9.67212849719144e-06,
      "loss": 1.0765,
      "step": 225
    },
    {
      "epoch": 0.43047619047619046,
      "grad_norm": 2.8103206233957367,
      "learning_rate": 9.668454879805394e-06,
      "loss": 1.1322,
      "step": 226
    },
    {
      "epoch": 0.43238095238095237,
      "grad_norm": 2.8355047244496703,
      "learning_rate": 9.664761502067019e-06,
      "loss": 1.1095,
      "step": 227
    },
    {
      "epoch": 0.4342857142857143,
      "grad_norm": 3.114785788041395,
      "learning_rate": 9.661048379609419e-06,
      "loss": 1.1004,
      "step": 228
    },
    {
      "epoch": 0.4361904761904762,
      "grad_norm": 3.021879404601287,
      "learning_rate": 9.657315528149275e-06,
      "loss": 0.9346,
      "step": 229
    },
    {
      "epoch": 0.4380952380952381,
      "grad_norm": 3.026803250823136,
      "learning_rate": 9.653562963486774e-06,
      "loss": 1.0449,
      "step": 230
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.4206628902791025,
      "learning_rate": 9.649790701505541e-06,
      "loss": 1.1101,
      "step": 231
    },
    {
      "epoch": 0.4419047619047619,
      "grad_norm": 2.6551465295801306,
      "learning_rate": 9.64599875817258e-06,
      "loss": 1.0131,
      "step": 232
    },
    {
      "epoch": 0.4438095238095238,
      "grad_norm": 2.854467341451607,
      "learning_rate": 9.6421871495382e-06,
      "loss": 1.1845,
      "step": 233
    },
    {
      "epoch": 0.44571428571428573,
      "grad_norm": 2.827217189746745,
      "learning_rate": 9.638355891735944e-06,
      "loss": 1.0443,
      "step": 234
    },
    {
      "epoch": 0.44761904761904764,
      "grad_norm": 3.006208737372956,
      "learning_rate": 9.634505000982529e-06,
      "loss": 1.0076,
      "step": 235
    },
    {
      "epoch": 0.44952380952380955,
      "grad_norm": 2.5796022640509086,
      "learning_rate": 9.630634493577767e-06,
      "loss": 1.0163,
      "step": 236
    },
    {
      "epoch": 0.4514285714285714,
      "grad_norm": 2.7106342118537485,
      "learning_rate": 9.626744385904512e-06,
      "loss": 0.9616,
      "step": 237
    },
    {
      "epoch": 0.4533333333333333,
      "grad_norm": 2.2487090446859574,
      "learning_rate": 9.622834694428574e-06,
      "loss": 0.953,
      "step": 238
    },
    {
      "epoch": 0.4552380952380952,
      "grad_norm": 3.3958126730237983,
      "learning_rate": 9.618905435698658e-06,
      "loss": 1.0927,
      "step": 239
    },
    {
      "epoch": 0.45714285714285713,
      "grad_norm": 2.305056273629618,
      "learning_rate": 9.61495662634629e-06,
      "loss": 1.0678,
      "step": 240
    },
    {
      "epoch": 0.45904761904761904,
      "grad_norm": 3.119636116341919,
      "learning_rate": 9.610988283085752e-06,
      "loss": 1.0628,
      "step": 241
    },
    {
      "epoch": 0.46095238095238095,
      "grad_norm": 3.4556491676683976,
      "learning_rate": 9.607000422714006e-06,
      "loss": 1.0585,
      "step": 242
    },
    {
      "epoch": 0.46285714285714286,
      "grad_norm": 3.4346262774329124,
      "learning_rate": 9.602993062110626e-06,
      "loss": 0.995,
      "step": 243
    },
    {
      "epoch": 0.46476190476190476,
      "grad_norm": 3.470708055444413,
      "learning_rate": 9.598966218237723e-06,
      "loss": 1.1668,
      "step": 244
    },
    {
      "epoch": 0.4666666666666667,
      "grad_norm": 2.7204516702659953,
      "learning_rate": 9.594919908139878e-06,
      "loss": 1.0181,
      "step": 245
    },
    {
      "epoch": 0.4685714285714286,
      "grad_norm": 2.978286252335797,
      "learning_rate": 9.590854148944067e-06,
      "loss": 1.0025,
      "step": 246
    },
    {
      "epoch": 0.4704761904761905,
      "grad_norm": 2.3895185947413604,
      "learning_rate": 9.586768957859589e-06,
      "loss": 1.0165,
      "step": 247
    },
    {
      "epoch": 0.4723809523809524,
      "grad_norm": 3.1874291056224178,
      "learning_rate": 9.58266435217799e-06,
      "loss": 0.9626,
      "step": 248
    },
    {
      "epoch": 0.4742857142857143,
      "grad_norm": 2.2712287800543205,
      "learning_rate": 9.578540349273e-06,
      "loss": 1.0621,
      "step": 249
    },
    {
      "epoch": 0.47619047619047616,
      "grad_norm": 2.643855435385986,
      "learning_rate": 9.574396966600445e-06,
      "loss": 1.102,
      "step": 250
    },
    {
      "epoch": 0.4780952380952381,
      "grad_norm": 2.5617281013572866,
      "learning_rate": 9.570234221698185e-06,
      "loss": 0.9841,
      "step": 251
    },
    {
      "epoch": 0.48,
      "grad_norm": 3.1302368487531282,
      "learning_rate": 9.566052132186032e-06,
      "loss": 0.9949,
      "step": 252
    },
    {
      "epoch": 0.4819047619047619,
      "grad_norm": 2.5440685943414665,
      "learning_rate": 9.561850715765684e-06,
      "loss": 1.2165,
      "step": 253
    },
    {
      "epoch": 0.4838095238095238,
      "grad_norm": 3.1267589828104905,
      "learning_rate": 9.557629990220637e-06,
      "loss": 1.0531,
      "step": 254
    },
    {
      "epoch": 0.4857142857142857,
      "grad_norm": 2.4958404703914208,
      "learning_rate": 9.553389973416125e-06,
      "loss": 1.028,
      "step": 255
    },
    {
      "epoch": 0.4876190476190476,
      "grad_norm": 3.2773918473984898,
      "learning_rate": 9.549130683299032e-06,
      "loss": 1.1487,
      "step": 256
    },
    {
      "epoch": 0.4895238095238095,
      "grad_norm": 2.4843863440219383,
      "learning_rate": 9.544852137897824e-06,
      "loss": 1.0878,
      "step": 257
    },
    {
      "epoch": 0.49142857142857144,
      "grad_norm": 3.0899480036220863,
      "learning_rate": 9.540554355322467e-06,
      "loss": 1.1547,
      "step": 258
    },
    {
      "epoch": 0.49333333333333335,
      "grad_norm": 2.4873063573655694,
      "learning_rate": 9.536237353764354e-06,
      "loss": 1.0231,
      "step": 259
    },
    {
      "epoch": 0.49523809523809526,
      "grad_norm": 2.4731566104989335,
      "learning_rate": 9.531901151496227e-06,
      "loss": 1.1443,
      "step": 260
    },
    {
      "epoch": 0.49714285714285716,
      "grad_norm": 3.644355028600177,
      "learning_rate": 9.527545766872102e-06,
      "loss": 1.0975,
      "step": 261
    },
    {
      "epoch": 0.4990476190476191,
      "grad_norm": 2.463223924477943,
      "learning_rate": 9.523171218327184e-06,
      "loss": 1.0939,
      "step": 262
    },
    {
      "epoch": 0.5009523809523809,
      "grad_norm": 3.66085062531903,
      "learning_rate": 9.518777524377796e-06,
      "loss": 1.2054,
      "step": 263
    },
    {
      "epoch": 0.5028571428571429,
      "grad_norm": 3.4454973268157447,
      "learning_rate": 9.514364703621299e-06,
      "loss": 1.0681,
      "step": 264
    },
    {
      "epoch": 0.5047619047619047,
      "grad_norm": 4.078934398255164,
      "learning_rate": 9.509932774736016e-06,
      "loss": 0.9327,
      "step": 265
    },
    {
      "epoch": 0.5066666666666667,
      "grad_norm": 3.6449246722536577,
      "learning_rate": 9.505481756481145e-06,
      "loss": 1.0902,
      "step": 266
    },
    {
      "epoch": 0.5085714285714286,
      "grad_norm": 3.0639601396623344,
      "learning_rate": 9.501011667696683e-06,
      "loss": 1.0834,
      "step": 267
    },
    {
      "epoch": 0.5104761904761905,
      "grad_norm": 2.3245739748017638,
      "learning_rate": 9.496522527303353e-06,
      "loss": 1.0887,
      "step": 268
    },
    {
      "epoch": 0.5123809523809524,
      "grad_norm": 2.649201264888993,
      "learning_rate": 9.492014354302516e-06,
      "loss": 1.0926,
      "step": 269
    },
    {
      "epoch": 0.5142857142857142,
      "grad_norm": 3.3699942327060137,
      "learning_rate": 9.48748716777609e-06,
      "loss": 1.1117,
      "step": 270
    },
    {
      "epoch": 0.5161904761904762,
      "grad_norm": 2.3315268640810807,
      "learning_rate": 9.482940986886479e-06,
      "loss": 1.1532,
      "step": 271
    },
    {
      "epoch": 0.518095238095238,
      "grad_norm": 3.2872520805798287,
      "learning_rate": 9.47837583087648e-06,
      "loss": 1.168,
      "step": 272
    },
    {
      "epoch": 0.52,
      "grad_norm": 3.175259967164933,
      "learning_rate": 9.47379171906921e-06,
      "loss": 1.1615,
      "step": 273
    },
    {
      "epoch": 0.5219047619047619,
      "grad_norm": 2.31119179225052,
      "learning_rate": 9.469188670868015e-06,
      "loss": 1.0789,
      "step": 274
    },
    {
      "epoch": 0.5238095238095238,
      "grad_norm": 3.665629221607055,
      "learning_rate": 9.4645667057564e-06,
      "loss": 1.1591,
      "step": 275
    },
    {
      "epoch": 0.5257142857142857,
      "grad_norm": 2.838895713766497,
      "learning_rate": 9.459925843297938e-06,
      "loss": 1.05,
      "step": 276
    },
    {
      "epoch": 0.5276190476190477,
      "grad_norm": 3.593344894023993,
      "learning_rate": 9.45526610313619e-06,
      "loss": 1.0057,
      "step": 277
    },
    {
      "epoch": 0.5295238095238095,
      "grad_norm": 3.4688559822187073,
      "learning_rate": 9.450587504994621e-06,
      "loss": 1.2173,
      "step": 278
    },
    {
      "epoch": 0.5314285714285715,
      "grad_norm": 3.221192760938516,
      "learning_rate": 9.445890068676518e-06,
      "loss": 1.2644,
      "step": 279
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 2.673413979534444,
      "learning_rate": 9.441173814064903e-06,
      "loss": 0.8941,
      "step": 280
    },
    {
      "epoch": 0.5352380952380953,
      "grad_norm": 2.7644194263380126,
      "learning_rate": 9.436438761122453e-06,
      "loss": 1.1562,
      "step": 281
    },
    {
      "epoch": 0.5371428571428571,
      "grad_norm": 3.415965862086195,
      "learning_rate": 9.431684929891412e-06,
      "loss": 1.2217,
      "step": 282
    },
    {
      "epoch": 0.539047619047619,
      "grad_norm": 3.07491566363796,
      "learning_rate": 9.42691234049351e-06,
      "loss": 1.0962,
      "step": 283
    },
    {
      "epoch": 0.540952380952381,
      "grad_norm": 2.9432367183464327,
      "learning_rate": 9.42212101312987e-06,
      "loss": 0.971,
      "step": 284
    },
    {
      "epoch": 0.5428571428571428,
      "grad_norm": 3.3779697079478272,
      "learning_rate": 9.417310968080934e-06,
      "loss": 1.1105,
      "step": 285
    },
    {
      "epoch": 0.5447619047619048,
      "grad_norm": 3.1745310820670407,
      "learning_rate": 9.412482225706368e-06,
      "loss": 0.9272,
      "step": 286
    },
    {
      "epoch": 0.5466666666666666,
      "grad_norm": 3.0696325027617806,
      "learning_rate": 9.407634806444981e-06,
      "loss": 0.9993,
      "step": 287
    },
    {
      "epoch": 0.5485714285714286,
      "grad_norm": 3.1633035674577417,
      "learning_rate": 9.402768730814633e-06,
      "loss": 1.1908,
      "step": 288
    },
    {
      "epoch": 0.5504761904761905,
      "grad_norm": 3.096994879166552,
      "learning_rate": 9.397884019412154e-06,
      "loss": 1.1474,
      "step": 289
    },
    {
      "epoch": 0.5523809523809524,
      "grad_norm": 2.2278564335531676,
      "learning_rate": 9.392980692913251e-06,
      "loss": 0.9674,
      "step": 290
    },
    {
      "epoch": 0.5542857142857143,
      "grad_norm": 2.578548275724668,
      "learning_rate": 9.388058772072433e-06,
      "loss": 0.9822,
      "step": 291
    },
    {
      "epoch": 0.5561904761904762,
      "grad_norm": 3.1635229335006945,
      "learning_rate": 9.383118277722902e-06,
      "loss": 1.1093,
      "step": 292
    },
    {
      "epoch": 0.5580952380952381,
      "grad_norm": 3.071945428911023,
      "learning_rate": 9.378159230776487e-06,
      "loss": 1.213,
      "step": 293
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.3121878371586146,
      "learning_rate": 9.373181652223536e-06,
      "loss": 1.0306,
      "step": 294
    },
    {
      "epoch": 0.5619047619047619,
      "grad_norm": 3.249984941625326,
      "learning_rate": 9.368185563132845e-06,
      "loss": 1.0258,
      "step": 295
    },
    {
      "epoch": 0.5638095238095238,
      "grad_norm": 2.250049366695023,
      "learning_rate": 9.363170984651554e-06,
      "loss": 1.1255,
      "step": 296
    },
    {
      "epoch": 0.5657142857142857,
      "grad_norm": 3.282853646317834,
      "learning_rate": 9.358137938005067e-06,
      "loss": 1.0138,
      "step": 297
    },
    {
      "epoch": 0.5676190476190476,
      "grad_norm": 4.102435203986516,
      "learning_rate": 9.35308644449696e-06,
      "loss": 1.1136,
      "step": 298
    },
    {
      "epoch": 0.5695238095238095,
      "grad_norm": 3.190359078403304,
      "learning_rate": 9.348016525508886e-06,
      "loss": 1.0097,
      "step": 299
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 2.8055099383728406,
      "learning_rate": 9.342928202500492e-06,
      "loss": 1.0414,
      "step": 300
    },
    {
      "epoch": 0.5733333333333334,
      "grad_norm": 3.226391622091831,
      "learning_rate": 9.337821497009321e-06,
      "loss": 1.1383,
      "step": 301
    },
    {
      "epoch": 0.5752380952380952,
      "grad_norm": 2.7242755307654507,
      "learning_rate": 9.332696430650726e-06,
      "loss": 0.9899,
      "step": 302
    },
    {
      "epoch": 0.5771428571428572,
      "grad_norm": 2.7396710737709706,
      "learning_rate": 9.327553025117774e-06,
      "loss": 1.0438,
      "step": 303
    },
    {
      "epoch": 0.579047619047619,
      "grad_norm": 2.3803496855726767,
      "learning_rate": 9.322391302181167e-06,
      "loss": 1.1477,
      "step": 304
    },
    {
      "epoch": 0.580952380952381,
      "grad_norm": 2.657112971537801,
      "learning_rate": 9.317211283689126e-06,
      "loss": 1.1747,
      "step": 305
    },
    {
      "epoch": 0.5828571428571429,
      "grad_norm": 3.2004740781756698,
      "learning_rate": 9.312012991567322e-06,
      "loss": 1.0265,
      "step": 306
    },
    {
      "epoch": 0.5847619047619048,
      "grad_norm": 2.7345693906589466,
      "learning_rate": 9.306796447818768e-06,
      "loss": 0.8831,
      "step": 307
    },
    {
      "epoch": 0.5866666666666667,
      "grad_norm": 2.581473591303195,
      "learning_rate": 9.301561674523736e-06,
      "loss": 1.3194,
      "step": 308
    },
    {
      "epoch": 0.5885714285714285,
      "grad_norm": 3.0062420990531775,
      "learning_rate": 9.296308693839652e-06,
      "loss": 1.0905,
      "step": 309
    },
    {
      "epoch": 0.5904761904761905,
      "grad_norm": 3.15466633587545,
      "learning_rate": 9.291037528001021e-06,
      "loss": 0.8329,
      "step": 310
    },
    {
      "epoch": 0.5923809523809523,
      "grad_norm": 3.0856341101198232,
      "learning_rate": 9.285748199319307e-06,
      "loss": 1.0536,
      "step": 311
    },
    {
      "epoch": 0.5942857142857143,
      "grad_norm": 2.2216358207731584,
      "learning_rate": 9.280440730182863e-06,
      "loss": 0.9541,
      "step": 312
    },
    {
      "epoch": 0.5961904761904762,
      "grad_norm": 2.842861446009651,
      "learning_rate": 9.275115143056819e-06,
      "loss": 1.0248,
      "step": 313
    },
    {
      "epoch": 0.5980952380952381,
      "grad_norm": 3.8297523835881,
      "learning_rate": 9.269771460482998e-06,
      "loss": 1.1637,
      "step": 314
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.685139474137949,
      "learning_rate": 9.264409705079819e-06,
      "loss": 1.1308,
      "step": 315
    },
    {
      "epoch": 0.6019047619047619,
      "grad_norm": 3.2078014043192438,
      "learning_rate": 9.25902989954219e-06,
      "loss": 1.1837,
      "step": 316
    },
    {
      "epoch": 0.6038095238095238,
      "grad_norm": 3.6262754832843958,
      "learning_rate": 9.253632066641427e-06,
      "loss": 1.1233,
      "step": 317
    },
    {
      "epoch": 0.6057142857142858,
      "grad_norm": 2.8480071200284254,
      "learning_rate": 9.248216229225148e-06,
      "loss": 0.997,
      "step": 318
    },
    {
      "epoch": 0.6076190476190476,
      "grad_norm": 2.965424675109055,
      "learning_rate": 9.242782410217182e-06,
      "loss": 1.0191,
      "step": 319
    },
    {
      "epoch": 0.6095238095238096,
      "grad_norm": 3.4390134855353938,
      "learning_rate": 9.237330632617469e-06,
      "loss": 1.0756,
      "step": 320
    },
    {
      "epoch": 0.6114285714285714,
      "grad_norm": 2.520790848512735,
      "learning_rate": 9.231860919501958e-06,
      "loss": 1.1408,
      "step": 321
    },
    {
      "epoch": 0.6133333333333333,
      "grad_norm": 3.0228482634517806,
      "learning_rate": 9.226373294022524e-06,
      "loss": 1.1168,
      "step": 322
    },
    {
      "epoch": 0.6152380952380953,
      "grad_norm": 2.8478464921869926,
      "learning_rate": 9.220867779406853e-06,
      "loss": 1.0273,
      "step": 323
    },
    {
      "epoch": 0.6171428571428571,
      "grad_norm": 3.403570442777272,
      "learning_rate": 9.215344398958351e-06,
      "loss": 1.0575,
      "step": 324
    },
    {
      "epoch": 0.6190476190476191,
      "grad_norm": 3.599725555546533,
      "learning_rate": 9.209803176056048e-06,
      "loss": 1.1933,
      "step": 325
    },
    {
      "epoch": 0.6209523809523809,
      "grad_norm": 3.276388578728829,
      "learning_rate": 9.204244134154499e-06,
      "loss": 1.1533,
      "step": 326
    },
    {
      "epoch": 0.6228571428571429,
      "grad_norm": 2.608627014756356,
      "learning_rate": 9.198667296783674e-06,
      "loss": 1.3496,
      "step": 327
    },
    {
      "epoch": 0.6247619047619047,
      "grad_norm": 3.300183900694987,
      "learning_rate": 9.193072687548875e-06,
      "loss": 1.2422,
      "step": 328
    },
    {
      "epoch": 0.6266666666666667,
      "grad_norm": 2.77850058830544,
      "learning_rate": 9.187460330130624e-06,
      "loss": 1.0319,
      "step": 329
    },
    {
      "epoch": 0.6285714285714286,
      "grad_norm": 3.376930054441844,
      "learning_rate": 9.181830248284565e-06,
      "loss": 1.3129,
      "step": 330
    },
    {
      "epoch": 0.6304761904761905,
      "grad_norm": 4.122798968695337,
      "learning_rate": 9.176182465841368e-06,
      "loss": 1.0245,
      "step": 331
    },
    {
      "epoch": 0.6323809523809524,
      "grad_norm": 2.7632021808150795,
      "learning_rate": 9.170517006706623e-06,
      "loss": 1.2431,
      "step": 332
    },
    {
      "epoch": 0.6342857142857142,
      "grad_norm": 3.0319102483966196,
      "learning_rate": 9.164833894860743e-06,
      "loss": 1.0567,
      "step": 333
    },
    {
      "epoch": 0.6361904761904762,
      "grad_norm": 2.414195345453208,
      "learning_rate": 9.159133154358856e-06,
      "loss": 1.1851,
      "step": 334
    },
    {
      "epoch": 0.638095238095238,
      "grad_norm": 3.4637316101079736,
      "learning_rate": 9.153414809330712e-06,
      "loss": 1.1527,
      "step": 335
    },
    {
      "epoch": 0.64,
      "grad_norm": 2.458716682265978,
      "learning_rate": 9.147678883980572e-06,
      "loss": 1.2253,
      "step": 336
    },
    {
      "epoch": 0.6419047619047619,
      "grad_norm": 3.0209779102998393,
      "learning_rate": 9.141925402587118e-06,
      "loss": 1.0282,
      "step": 337
    },
    {
      "epoch": 0.6438095238095238,
      "grad_norm": 2.884184752305691,
      "learning_rate": 9.136154389503332e-06,
      "loss": 1.082,
      "step": 338
    },
    {
      "epoch": 0.6457142857142857,
      "grad_norm": 2.8859380779810513,
      "learning_rate": 9.130365869156408e-06,
      "loss": 0.9967,
      "step": 339
    },
    {
      "epoch": 0.6476190476190476,
      "grad_norm": 4.099940868149714,
      "learning_rate": 9.124559866047646e-06,
      "loss": 1.1936,
      "step": 340
    },
    {
      "epoch": 0.6495238095238095,
      "grad_norm": 2.6309654217829745,
      "learning_rate": 9.118736404752342e-06,
      "loss": 1.0655,
      "step": 341
    },
    {
      "epoch": 0.6514285714285715,
      "grad_norm": 3.1773333103707446,
      "learning_rate": 9.112895509919687e-06,
      "loss": 1.1259,
      "step": 342
    },
    {
      "epoch": 0.6533333333333333,
      "grad_norm": 3.1167716935695893,
      "learning_rate": 9.10703720627267e-06,
      "loss": 1.1632,
      "step": 343
    },
    {
      "epoch": 0.6552380952380953,
      "grad_norm": 3.3279493562876197,
      "learning_rate": 9.101161518607962e-06,
      "loss": 1.1378,
      "step": 344
    },
    {
      "epoch": 0.6571428571428571,
      "grad_norm": 3.1981939157817947,
      "learning_rate": 9.095268471795813e-06,
      "loss": 1.1373,
      "step": 345
    },
    {
      "epoch": 0.659047619047619,
      "grad_norm": 2.8603014674607428,
      "learning_rate": 9.08935809077996e-06,
      "loss": 1.0392,
      "step": 346
    },
    {
      "epoch": 0.660952380952381,
      "grad_norm": 2.724259741971536,
      "learning_rate": 9.0834304005775e-06,
      "loss": 1.1743,
      "step": 347
    },
    {
      "epoch": 0.6628571428571428,
      "grad_norm": 2.9885426816774316,
      "learning_rate": 9.077485426278802e-06,
      "loss": 1.1454,
      "step": 348
    },
    {
      "epoch": 0.6647619047619048,
      "grad_norm": 2.504779329263032,
      "learning_rate": 9.071523193047392e-06,
      "loss": 1.0482,
      "step": 349
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 3.044001947744392,
      "learning_rate": 9.065543726119847e-06,
      "loss": 1.0667,
      "step": 350
    },
    {
      "epoch": 0.6685714285714286,
      "grad_norm": 2.412752952896956,
      "learning_rate": 9.059547050805696e-06,
      "loss": 1.0208,
      "step": 351
    },
    {
      "epoch": 0.6704761904761904,
      "grad_norm": 2.7191213066147517,
      "learning_rate": 9.053533192487298e-06,
      "loss": 1.0417,
      "step": 352
    },
    {
      "epoch": 0.6723809523809524,
      "grad_norm": 2.8814004911974176,
      "learning_rate": 9.047502176619749e-06,
      "loss": 1.1503,
      "step": 353
    },
    {
      "epoch": 0.6742857142857143,
      "grad_norm": 2.805131777342699,
      "learning_rate": 9.041454028730767e-06,
      "loss": 1.0081,
      "step": 354
    },
    {
      "epoch": 0.6761904761904762,
      "grad_norm": 2.565200413286613,
      "learning_rate": 9.035388774420584e-06,
      "loss": 1.1812,
      "step": 355
    },
    {
      "epoch": 0.6780952380952381,
      "grad_norm": 2.8238095599448245,
      "learning_rate": 9.029306439361841e-06,
      "loss": 1.0914,
      "step": 356
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.6113097414494755,
      "learning_rate": 9.023207049299473e-06,
      "loss": 1.1112,
      "step": 357
    },
    {
      "epoch": 0.6819047619047619,
      "grad_norm": 3.1869036634410777,
      "learning_rate": 9.017090630050616e-06,
      "loss": 1.0125,
      "step": 358
    },
    {
      "epoch": 0.6838095238095238,
      "grad_norm": 3.0223119368517186,
      "learning_rate": 9.010957207504471e-06,
      "loss": 1.0529,
      "step": 359
    },
    {
      "epoch": 0.6857142857142857,
      "grad_norm": 2.80424218698681,
      "learning_rate": 9.004806807622219e-06,
      "loss": 1.009,
      "step": 360
    },
    {
      "epoch": 0.6876190476190476,
      "grad_norm": 3.259013929402233,
      "learning_rate": 8.998639456436898e-06,
      "loss": 1.1255,
      "step": 361
    },
    {
      "epoch": 0.6895238095238095,
      "grad_norm": 3.460844692910995,
      "learning_rate": 8.992455180053298e-06,
      "loss": 1.0553,
      "step": 362
    },
    {
      "epoch": 0.6914285714285714,
      "grad_norm": 3.663541327578872,
      "learning_rate": 8.98625400464785e-06,
      "loss": 1.1177,
      "step": 363
    },
    {
      "epoch": 0.6933333333333334,
      "grad_norm": 2.8186318281692113,
      "learning_rate": 8.98003595646851e-06,
      "loss": 1.0084,
      "step": 364
    },
    {
      "epoch": 0.6952380952380952,
      "grad_norm": 3.817357178726652,
      "learning_rate": 8.973801061834657e-06,
      "loss": 1.1294,
      "step": 365
    },
    {
      "epoch": 0.6971428571428572,
      "grad_norm": 3.0000571353197762,
      "learning_rate": 8.967549347136974e-06,
      "loss": 1.0939,
      "step": 366
    },
    {
      "epoch": 0.699047619047619,
      "grad_norm": 2.489047116573603,
      "learning_rate": 8.96128083883734e-06,
      "loss": 1.0108,
      "step": 367
    },
    {
      "epoch": 0.700952380952381,
      "grad_norm": 2.2374284799304758,
      "learning_rate": 8.954995563468713e-06,
      "loss": 0.9346,
      "step": 368
    },
    {
      "epoch": 0.7028571428571428,
      "grad_norm": 2.858066130943614,
      "learning_rate": 8.948693547635029e-06,
      "loss": 1.1154,
      "step": 369
    },
    {
      "epoch": 0.7047619047619048,
      "grad_norm": 2.5246337088575377,
      "learning_rate": 8.942374818011074e-06,
      "loss": 1.154,
      "step": 370
    },
    {
      "epoch": 0.7066666666666667,
      "grad_norm": 3.2249688145142525,
      "learning_rate": 8.936039401342388e-06,
      "loss": 1.1242,
      "step": 371
    },
    {
      "epoch": 0.7085714285714285,
      "grad_norm": 3.1953262833622103,
      "learning_rate": 8.929687324445135e-06,
      "loss": 1.1208,
      "step": 372
    },
    {
      "epoch": 0.7104761904761905,
      "grad_norm": 3.4019986996229905,
      "learning_rate": 8.923318614206e-06,
      "loss": 1.0769,
      "step": 373
    },
    {
      "epoch": 0.7123809523809523,
      "grad_norm": 2.9160497148249647,
      "learning_rate": 8.916933297582073e-06,
      "loss": 1.045,
      "step": 374
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 2.560075764013695,
      "learning_rate": 8.910531401600735e-06,
      "loss": 1.1377,
      "step": 375
    },
    {
      "epoch": 0.7161904761904762,
      "grad_norm": 2.8083474146713256,
      "learning_rate": 8.904112953359542e-06,
      "loss": 1.1011,
      "step": 376
    },
    {
      "epoch": 0.7180952380952381,
      "grad_norm": 2.5673621742011026,
      "learning_rate": 8.897677980026114e-06,
      "loss": 0.9616,
      "step": 377
    },
    {
      "epoch": 0.72,
      "grad_norm": 3.7642909388622736,
      "learning_rate": 8.891226508838013e-06,
      "loss": 1.0577,
      "step": 378
    },
    {
      "epoch": 0.7219047619047619,
      "grad_norm": 3.364840838181064,
      "learning_rate": 8.884758567102636e-06,
      "loss": 1.075,
      "step": 379
    },
    {
      "epoch": 0.7238095238095238,
      "grad_norm": 2.6838256528900337,
      "learning_rate": 8.878274182197092e-06,
      "loss": 1.025,
      "step": 380
    },
    {
      "epoch": 0.7257142857142858,
      "grad_norm": 3.0219185424141095,
      "learning_rate": 8.871773381568093e-06,
      "loss": 0.9733,
      "step": 381
    },
    {
      "epoch": 0.7276190476190476,
      "grad_norm": 2.542427915477238,
      "learning_rate": 8.865256192731835e-06,
      "loss": 1.0816,
      "step": 382
    },
    {
      "epoch": 0.7295238095238096,
      "grad_norm": 3.335494765427363,
      "learning_rate": 8.858722643273877e-06,
      "loss": 1.2553,
      "step": 383
    },
    {
      "epoch": 0.7314285714285714,
      "grad_norm": 4.050862992491009,
      "learning_rate": 8.852172760849029e-06,
      "loss": 1.1365,
      "step": 384
    },
    {
      "epoch": 0.7333333333333333,
      "grad_norm": 3.4436727393115127,
      "learning_rate": 8.845606573181241e-06,
      "loss": 1.1239,
      "step": 385
    },
    {
      "epoch": 0.7352380952380952,
      "grad_norm": 2.82839942474682,
      "learning_rate": 8.839024108063466e-06,
      "loss": 1.0783,
      "step": 386
    },
    {
      "epoch": 0.7371428571428571,
      "grad_norm": 2.7206052968291377,
      "learning_rate": 8.83242539335757e-06,
      "loss": 1.1512,
      "step": 387
    },
    {
      "epoch": 0.7390476190476191,
      "grad_norm": 3.510052220359161,
      "learning_rate": 8.825810456994184e-06,
      "loss": 1.124,
      "step": 388
    },
    {
      "epoch": 0.7409523809523809,
      "grad_norm": 2.967050518609841,
      "learning_rate": 8.819179326972615e-06,
      "loss": 1.1152,
      "step": 389
    },
    {
      "epoch": 0.7428571428571429,
      "grad_norm": 3.379005317255521,
      "learning_rate": 8.812532031360707e-06,
      "loss": 1.149,
      "step": 390
    },
    {
      "epoch": 0.7447619047619047,
      "grad_norm": 3.2123171616309336,
      "learning_rate": 8.805868598294728e-06,
      "loss": 1.0989,
      "step": 391
    },
    {
      "epoch": 0.7466666666666667,
      "grad_norm": 2.422084944704448,
      "learning_rate": 8.79918905597925e-06,
      "loss": 1.0587,
      "step": 392
    },
    {
      "epoch": 0.7485714285714286,
      "grad_norm": 3.4402615362346713,
      "learning_rate": 8.792493432687042e-06,
      "loss": 1.0559,
      "step": 393
    },
    {
      "epoch": 0.7504761904761905,
      "grad_norm": 2.427929273885637,
      "learning_rate": 8.785781756758923e-06,
      "loss": 1.0436,
      "step": 394
    },
    {
      "epoch": 0.7523809523809524,
      "grad_norm": 3.282090778146958,
      "learning_rate": 8.779054056603672e-06,
      "loss": 1.0019,
      "step": 395
    },
    {
      "epoch": 0.7542857142857143,
      "grad_norm": 3.5061662461280627,
      "learning_rate": 8.77231036069789e-06,
      "loss": 1.0609,
      "step": 396
    },
    {
      "epoch": 0.7561904761904762,
      "grad_norm": 2.552904246872334,
      "learning_rate": 8.765550697585879e-06,
      "loss": 0.9617,
      "step": 397
    },
    {
      "epoch": 0.758095238095238,
      "grad_norm": 3.3596424805810585,
      "learning_rate": 8.758775095879535e-06,
      "loss": 1.2109,
      "step": 398
    },
    {
      "epoch": 0.76,
      "grad_norm": 3.4348345089003622,
      "learning_rate": 8.751983584258213e-06,
      "loss": 1.1735,
      "step": 399
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 2.231821616463513,
      "learning_rate": 8.745176191468611e-06,
      "loss": 0.9655,
      "step": 400
    },
    {
      "epoch": 0.7638095238095238,
      "grad_norm": 2.6109892200645644,
      "learning_rate": 8.738352946324646e-06,
      "loss": 1.0254,
      "step": 401
    },
    {
      "epoch": 0.7657142857142857,
      "grad_norm": 2.980437739721515,
      "learning_rate": 8.731513877707336e-06,
      "loss": 1.1012,
      "step": 402
    },
    {
      "epoch": 0.7676190476190476,
      "grad_norm": 3.0083473362483533,
      "learning_rate": 8.724659014564676e-06,
      "loss": 0.9967,
      "step": 403
    },
    {
      "epoch": 0.7695238095238095,
      "grad_norm": 2.558923124814242,
      "learning_rate": 8.717788385911514e-06,
      "loss": 1.1585,
      "step": 404
    },
    {
      "epoch": 0.7714285714285715,
      "grad_norm": 2.3051163211822887,
      "learning_rate": 8.710902020829433e-06,
      "loss": 1.0719,
      "step": 405
    },
    {
      "epoch": 0.7733333333333333,
      "grad_norm": 2.41821834923688,
      "learning_rate": 8.703999948466617e-06,
      "loss": 1.0255,
      "step": 406
    },
    {
      "epoch": 0.7752380952380953,
      "grad_norm": 3.14367022632351,
      "learning_rate": 8.697082198037739e-06,
      "loss": 1.1382,
      "step": 407
    },
    {
      "epoch": 0.7771428571428571,
      "grad_norm": 2.9683955939402864,
      "learning_rate": 8.690148798823835e-06,
      "loss": 0.9404,
      "step": 408
    },
    {
      "epoch": 0.7790476190476191,
      "grad_norm": 2.529142172327783,
      "learning_rate": 8.683199780172175e-06,
      "loss": 1.2332,
      "step": 409
    },
    {
      "epoch": 0.780952380952381,
      "grad_norm": 2.4658388877395625,
      "learning_rate": 8.676235171496144e-06,
      "loss": 0.9892,
      "step": 410
    },
    {
      "epoch": 0.7828571428571428,
      "grad_norm": 2.8283641793738896,
      "learning_rate": 8.669255002275112e-06,
      "loss": 0.984,
      "step": 411
    },
    {
      "epoch": 0.7847619047619048,
      "grad_norm": 3.133647096585687,
      "learning_rate": 8.662259302054319e-06,
      "loss": 0.9927,
      "step": 412
    },
    {
      "epoch": 0.7866666666666666,
      "grad_norm": 2.758729735555187,
      "learning_rate": 8.655248100444737e-06,
      "loss": 1.1519,
      "step": 413
    },
    {
      "epoch": 0.7885714285714286,
      "grad_norm": 3.189001386891881,
      "learning_rate": 8.648221427122957e-06,
      "loss": 1.2207,
      "step": 414
    },
    {
      "epoch": 0.7904761904761904,
      "grad_norm": 2.9108592393995343,
      "learning_rate": 8.641179311831053e-06,
      "loss": 1.1063,
      "step": 415
    },
    {
      "epoch": 0.7923809523809524,
      "grad_norm": 3.6297074994123513,
      "learning_rate": 8.634121784376465e-06,
      "loss": 1.1972,
      "step": 416
    },
    {
      "epoch": 0.7942857142857143,
      "grad_norm": 3.0166932780857576,
      "learning_rate": 8.627048874631868e-06,
      "loss": 1.0153,
      "step": 417
    },
    {
      "epoch": 0.7961904761904762,
      "grad_norm": 2.352420567560322,
      "learning_rate": 8.619960612535038e-06,
      "loss": 0.8948,
      "step": 418
    },
    {
      "epoch": 0.7980952380952381,
      "grad_norm": 3.903440171260907,
      "learning_rate": 8.612857028088748e-06,
      "loss": 1.1302,
      "step": 419
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.9690330836261984,
      "learning_rate": 8.605738151360615e-06,
      "loss": 1.0636,
      "step": 420
    },
    {
      "epoch": 0.8019047619047619,
      "grad_norm": 3.242096843945755,
      "learning_rate": 8.598604012482989e-06,
      "loss": 1.1882,
      "step": 421
    },
    {
      "epoch": 0.8038095238095239,
      "grad_norm": 2.601107974340937,
      "learning_rate": 8.59145464165282e-06,
      "loss": 0.9328,
      "step": 422
    },
    {
      "epoch": 0.8057142857142857,
      "grad_norm": 3.236040533371977,
      "learning_rate": 8.584290069131532e-06,
      "loss": 1.1176,
      "step": 423
    },
    {
      "epoch": 0.8076190476190476,
      "grad_norm": 3.1387425308191363,
      "learning_rate": 8.577110325244889e-06,
      "loss": 0.9882,
      "step": 424
    },
    {
      "epoch": 0.8095238095238095,
      "grad_norm": 2.7150796170310576,
      "learning_rate": 8.569915440382877e-06,
      "loss": 0.8892,
      "step": 425
    },
    {
      "epoch": 0.8114285714285714,
      "grad_norm": 3.065969460883179,
      "learning_rate": 8.562705444999568e-06,
      "loss": 1.0418,
      "step": 426
    },
    {
      "epoch": 0.8133333333333334,
      "grad_norm": 3.0573807259121804,
      "learning_rate": 8.555480369612993e-06,
      "loss": 1.0592,
      "step": 427
    },
    {
      "epoch": 0.8152380952380952,
      "grad_norm": 2.3194057375292894,
      "learning_rate": 8.54824024480501e-06,
      "loss": 1.1651,
      "step": 428
    },
    {
      "epoch": 0.8171428571428572,
      "grad_norm": 2.529280513955374,
      "learning_rate": 8.540985101221182e-06,
      "loss": 1.1728,
      "step": 429
    },
    {
      "epoch": 0.819047619047619,
      "grad_norm": 2.8174939455132675,
      "learning_rate": 8.533714969570642e-06,
      "loss": 1.0767,
      "step": 430
    },
    {
      "epoch": 0.820952380952381,
      "grad_norm": 3.4046233317345966,
      "learning_rate": 8.526429880625957e-06,
      "loss": 1.1272,
      "step": 431
    },
    {
      "epoch": 0.8228571428571428,
      "grad_norm": 3.1126606357344992,
      "learning_rate": 8.51912986522301e-06,
      "loss": 1.1608,
      "step": 432
    },
    {
      "epoch": 0.8247619047619048,
      "grad_norm": 2.711030441114635,
      "learning_rate": 8.511814954260868e-06,
      "loss": 1.0115,
      "step": 433
    },
    {
      "epoch": 0.8266666666666667,
      "grad_norm": 2.615811959017416,
      "learning_rate": 8.504485178701635e-06,
      "loss": 0.8514,
      "step": 434
    },
    {
      "epoch": 0.8285714285714286,
      "grad_norm": 3.13353585968966,
      "learning_rate": 8.497140569570344e-06,
      "loss": 1.1303,
      "step": 435
    },
    {
      "epoch": 0.8304761904761905,
      "grad_norm": 3.333661923768568,
      "learning_rate": 8.48978115795481e-06,
      "loss": 1.2062,
      "step": 436
    },
    {
      "epoch": 0.8323809523809523,
      "grad_norm": 3.1005760988695132,
      "learning_rate": 8.482406975005506e-06,
      "loss": 0.9391,
      "step": 437
    },
    {
      "epoch": 0.8342857142857143,
      "grad_norm": 2.9439892789366433,
      "learning_rate": 8.47501805193542e-06,
      "loss": 1.0021,
      "step": 438
    },
    {
      "epoch": 0.8361904761904762,
      "grad_norm": 3.1430386500120466,
      "learning_rate": 8.46761442001994e-06,
      "loss": 1.1745,
      "step": 439
    },
    {
      "epoch": 0.8380952380952381,
      "grad_norm": 3.152245132905031,
      "learning_rate": 8.46019611059671e-06,
      "loss": 0.9255,
      "step": 440
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.6845203586161714,
      "learning_rate": 8.4527631550655e-06,
      "loss": 1.0977,
      "step": 441
    },
    {
      "epoch": 0.8419047619047619,
      "grad_norm": 3.096279054161254,
      "learning_rate": 8.445315584888073e-06,
      "loss": 1.1743,
      "step": 442
    },
    {
      "epoch": 0.8438095238095238,
      "grad_norm": 3.3066419584820257,
      "learning_rate": 8.43785343158805e-06,
      "loss": 1.0291,
      "step": 443
    },
    {
      "epoch": 0.8457142857142858,
      "grad_norm": 2.912561027411646,
      "learning_rate": 8.43037672675078e-06,
      "loss": 1.1286,
      "step": 444
    },
    {
      "epoch": 0.8476190476190476,
      "grad_norm": 3.365607195342652,
      "learning_rate": 8.422885502023207e-06,
      "loss": 1.0388,
      "step": 445
    },
    {
      "epoch": 0.8495238095238096,
      "grad_norm": 2.9397530588051612,
      "learning_rate": 8.41537978911373e-06,
      "loss": 1.1248,
      "step": 446
    },
    {
      "epoch": 0.8514285714285714,
      "grad_norm": 2.524457250295759,
      "learning_rate": 8.40785961979208e-06,
      "loss": 1.1046,
      "step": 447
    },
    {
      "epoch": 0.8533333333333334,
      "grad_norm": 2.5066645738851054,
      "learning_rate": 8.400325025889167e-06,
      "loss": 1.0707,
      "step": 448
    },
    {
      "epoch": 0.8552380952380952,
      "grad_norm": 2.1343550534829556,
      "learning_rate": 8.392776039296967e-06,
      "loss": 1.1798,
      "step": 449
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 2.4107377275946273,
      "learning_rate": 8.385212691968367e-06,
      "loss": 1.0357,
      "step": 450
    },
    {
      "epoch": 0.8590476190476191,
      "grad_norm": 3.225104794505188,
      "learning_rate": 8.37763501591705e-06,
      "loss": 1.0195,
      "step": 451
    },
    {
      "epoch": 0.8609523809523809,
      "grad_norm": 3.923484239965437,
      "learning_rate": 8.37004304321734e-06,
      "loss": 0.9942,
      "step": 452
    },
    {
      "epoch": 0.8628571428571429,
      "grad_norm": 2.5883022229346375,
      "learning_rate": 8.362436806004077e-06,
      "loss": 0.9948,
      "step": 453
    },
    {
      "epoch": 0.8647619047619047,
      "grad_norm": 3.1611212611466124,
      "learning_rate": 8.354816336472483e-06,
      "loss": 1.1588,
      "step": 454
    },
    {
      "epoch": 0.8666666666666667,
      "grad_norm": 2.227224884930765,
      "learning_rate": 8.347181666878016e-06,
      "loss": 1.051,
      "step": 455
    },
    {
      "epoch": 0.8685714285714285,
      "grad_norm": 3.343904034490303,
      "learning_rate": 8.339532829536243e-06,
      "loss": 1.1178,
      "step": 456
    },
    {
      "epoch": 0.8704761904761905,
      "grad_norm": 2.7042791692096837,
      "learning_rate": 8.331869856822699e-06,
      "loss": 0.9584,
      "step": 457
    },
    {
      "epoch": 0.8723809523809524,
      "grad_norm": 2.6112940045283892,
      "learning_rate": 8.324192781172748e-06,
      "loss": 1.1043,
      "step": 458
    },
    {
      "epoch": 0.8742857142857143,
      "grad_norm": 2.9972603993374625,
      "learning_rate": 8.316501635081452e-06,
      "loss": 0.9461,
      "step": 459
    },
    {
      "epoch": 0.8761904761904762,
      "grad_norm": 3.2457696049052602,
      "learning_rate": 8.308796451103425e-06,
      "loss": 1.049,
      "step": 460
    },
    {
      "epoch": 0.878095238095238,
      "grad_norm": 3.537077072294635,
      "learning_rate": 8.301077261852702e-06,
      "loss": 1.0805,
      "step": 461
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.9377251873072283,
      "learning_rate": 8.293344100002596e-06,
      "loss": 1.0289,
      "step": 462
    },
    {
      "epoch": 0.8819047619047619,
      "grad_norm": 3.3156194597464483,
      "learning_rate": 8.285596998285572e-06,
      "loss": 1.0314,
      "step": 463
    },
    {
      "epoch": 0.8838095238095238,
      "grad_norm": 2.6050334252906815,
      "learning_rate": 8.277835989493086e-06,
      "loss": 1.269,
      "step": 464
    },
    {
      "epoch": 0.8857142857142857,
      "grad_norm": 2.615504751195819,
      "learning_rate": 8.270061106475466e-06,
      "loss": 1.293,
      "step": 465
    },
    {
      "epoch": 0.8876190476190476,
      "grad_norm": 3.7859955032929973,
      "learning_rate": 8.262272382141764e-06,
      "loss": 1.0463,
      "step": 466
    },
    {
      "epoch": 0.8895238095238095,
      "grad_norm": 2.9860498670291804,
      "learning_rate": 8.254469849459623e-06,
      "loss": 1.0942,
      "step": 467
    },
    {
      "epoch": 0.8914285714285715,
      "grad_norm": 2.7344778901565,
      "learning_rate": 8.246653541455125e-06,
      "loss": 0.9873,
      "step": 468
    },
    {
      "epoch": 0.8933333333333333,
      "grad_norm": 2.80940900597449,
      "learning_rate": 8.238823491212666e-06,
      "loss": 1.067,
      "step": 469
    },
    {
      "epoch": 0.8952380952380953,
      "grad_norm": 3.4775010761402703,
      "learning_rate": 8.230979731874809e-06,
      "loss": 1.1605,
      "step": 470
    },
    {
      "epoch": 0.8971428571428571,
      "grad_norm": 3.065434576482116,
      "learning_rate": 8.223122296642139e-06,
      "loss": 0.9812,
      "step": 471
    },
    {
      "epoch": 0.8990476190476191,
      "grad_norm": 2.7304169187074074,
      "learning_rate": 8.21525121877313e-06,
      "loss": 1.1026,
      "step": 472
    },
    {
      "epoch": 0.900952380952381,
      "grad_norm": 2.7940523187258823,
      "learning_rate": 8.207366531584007e-06,
      "loss": 1.244,
      "step": 473
    },
    {
      "epoch": 0.9028571428571428,
      "grad_norm": 3.0019016221120047,
      "learning_rate": 8.199468268448591e-06,
      "loss": 1.103,
      "step": 474
    },
    {
      "epoch": 0.9047619047619048,
      "grad_norm": 3.488895058692748,
      "learning_rate": 8.191556462798171e-06,
      "loss": 0.9693,
      "step": 475
    },
    {
      "epoch": 0.9066666666666666,
      "grad_norm": 2.052419927964473,
      "learning_rate": 8.183631148121362e-06,
      "loss": 1.0691,
      "step": 476
    },
    {
      "epoch": 0.9085714285714286,
      "grad_norm": 2.784756056228296,
      "learning_rate": 8.175692357963949e-06,
      "loss": 0.9683,
      "step": 477
    },
    {
      "epoch": 0.9104761904761904,
      "grad_norm": 3.312667416922426,
      "learning_rate": 8.167740125928764e-06,
      "loss": 1.0703,
      "step": 478
    },
    {
      "epoch": 0.9123809523809524,
      "grad_norm": 2.1762321202342987,
      "learning_rate": 8.15977448567553e-06,
      "loss": 1.0664,
      "step": 479
    },
    {
      "epoch": 0.9142857142857143,
      "grad_norm": 2.856371298100056,
      "learning_rate": 8.151795470920729e-06,
      "loss": 1.0039,
      "step": 480
    },
    {
      "epoch": 0.9161904761904762,
      "grad_norm": 3.808057593484791,
      "learning_rate": 8.143803115437444e-06,
      "loss": 1.0565,
      "step": 481
    },
    {
      "epoch": 0.9180952380952381,
      "grad_norm": 2.4083168883883492,
      "learning_rate": 8.13579745305524e-06,
      "loss": 0.9804,
      "step": 482
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.458277457932072,
      "learning_rate": 8.12777851765999e-06,
      "loss": 0.9108,
      "step": 483
    },
    {
      "epoch": 0.9219047619047619,
      "grad_norm": 2.2041917678077008,
      "learning_rate": 8.119746343193764e-06,
      "loss": 0.8584,
      "step": 484
    },
    {
      "epoch": 0.9238095238095239,
      "grad_norm": 3.085155154904451,
      "learning_rate": 8.111700963654658e-06,
      "loss": 1.1146,
      "step": 485
    },
    {
      "epoch": 0.9257142857142857,
      "grad_norm": 2.5494148735504507,
      "learning_rate": 8.103642413096667e-06,
      "loss": 1.0932,
      "step": 486
    },
    {
      "epoch": 0.9276190476190476,
      "grad_norm": 2.516089543568002,
      "learning_rate": 8.095570725629537e-06,
      "loss": 1.1081,
      "step": 487
    },
    {
      "epoch": 0.9295238095238095,
      "grad_norm": 3.3273604855594607,
      "learning_rate": 8.08748593541861e-06,
      "loss": 1.0552,
      "step": 488
    },
    {
      "epoch": 0.9314285714285714,
      "grad_norm": 2.4935896236090453,
      "learning_rate": 8.079388076684703e-06,
      "loss": 1.0423,
      "step": 489
    },
    {
      "epoch": 0.9333333333333333,
      "grad_norm": 3.3186717714278116,
      "learning_rate": 8.071277183703935e-06,
      "loss": 1.1353,
      "step": 490
    },
    {
      "epoch": 0.9352380952380952,
      "grad_norm": 3.031819536929874,
      "learning_rate": 8.063153290807604e-06,
      "loss": 0.9664,
      "step": 491
    },
    {
      "epoch": 0.9371428571428572,
      "grad_norm": 3.3105066660602565,
      "learning_rate": 8.05501643238203e-06,
      "loss": 1.1321,
      "step": 492
    },
    {
      "epoch": 0.939047619047619,
      "grad_norm": 2.7712863621364865,
      "learning_rate": 8.046866642868414e-06,
      "loss": 1.0302,
      "step": 493
    },
    {
      "epoch": 0.940952380952381,
      "grad_norm": 2.9784754051557254,
      "learning_rate": 8.03870395676269e-06,
      "loss": 1.0708,
      "step": 494
    },
    {
      "epoch": 0.9428571428571428,
      "grad_norm": 3.74313100234932,
      "learning_rate": 8.030528408615377e-06,
      "loss": 1.304,
      "step": 495
    },
    {
      "epoch": 0.9447619047619048,
      "grad_norm": 2.34322724592221,
      "learning_rate": 8.022340033031443e-06,
      "loss": 0.9794,
      "step": 496
    },
    {
      "epoch": 0.9466666666666667,
      "grad_norm": 2.7200841196093117,
      "learning_rate": 8.014138864670144e-06,
      "loss": 1.1417,
      "step": 497
    },
    {
      "epoch": 0.9485714285714286,
      "grad_norm": 3.0826039786111097,
      "learning_rate": 8.005924938244889e-06,
      "loss": 1.2513,
      "step": 498
    },
    {
      "epoch": 0.9504761904761905,
      "grad_norm": 2.72441910058809,
      "learning_rate": 7.997698288523085e-06,
      "loss": 1.0193,
      "step": 499
    },
    {
      "epoch": 0.9523809523809523,
      "grad_norm": 2.5720275928169496,
      "learning_rate": 7.989458950326e-06,
      "loss": 1.119,
      "step": 500
    },
    {
      "epoch": 0.9542857142857143,
      "grad_norm": 2.8001662682136987,
      "learning_rate": 7.9812069585286e-06,
      "loss": 1.2259,
      "step": 501
    },
    {
      "epoch": 0.9561904761904761,
      "grad_norm": 2.6252536854215247,
      "learning_rate": 7.972942348059415e-06,
      "loss": 1.0859,
      "step": 502
    },
    {
      "epoch": 0.9580952380952381,
      "grad_norm": 2.892537843168099,
      "learning_rate": 7.964665153900389e-06,
      "loss": 1.1428,
      "step": 503
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.650168201967266,
      "learning_rate": 7.956375411086723e-06,
      "loss": 1.242,
      "step": 504
    },
    {
      "epoch": 0.9619047619047619,
      "grad_norm": 2.4635114767396047,
      "learning_rate": 7.948073154706741e-06,
      "loss": 1.031,
      "step": 505
    },
    {
      "epoch": 0.9638095238095238,
      "grad_norm": 2.61394095575774,
      "learning_rate": 7.939758419901728e-06,
      "loss": 0.9719,
      "step": 506
    },
    {
      "epoch": 0.9657142857142857,
      "grad_norm": 2.8577982697972075,
      "learning_rate": 7.931431241865787e-06,
      "loss": 1.0958,
      "step": 507
    },
    {
      "epoch": 0.9676190476190476,
      "grad_norm": 2.9964565356802186,
      "learning_rate": 7.923091655845696e-06,
      "loss": 1.135,
      "step": 508
    },
    {
      "epoch": 0.9695238095238096,
      "grad_norm": 2.6243394304546706,
      "learning_rate": 7.914739697140742e-06,
      "loss": 1.1313,
      "step": 509
    },
    {
      "epoch": 0.9714285714285714,
      "grad_norm": 4.129503983689747,
      "learning_rate": 7.906375401102594e-06,
      "loss": 1.1608,
      "step": 510
    },
    {
      "epoch": 0.9733333333333334,
      "grad_norm": 3.0240446885326886,
      "learning_rate": 7.897998803135134e-06,
      "loss": 1.1815,
      "step": 511
    },
    {
      "epoch": 0.9752380952380952,
      "grad_norm": 2.7395588200297247,
      "learning_rate": 7.889609938694317e-06,
      "loss": 1.0299,
      "step": 512
    },
    {
      "epoch": 0.9771428571428571,
      "grad_norm": 2.812468027721245,
      "learning_rate": 7.88120884328802e-06,
      "loss": 1.0421,
      "step": 513
    },
    {
      "epoch": 0.979047619047619,
      "grad_norm": 3.2398794565010833,
      "learning_rate": 7.87279555247589e-06,
      "loss": 0.9394,
      "step": 514
    },
    {
      "epoch": 0.9809523809523809,
      "grad_norm": 2.285044181961201,
      "learning_rate": 7.86437010186919e-06,
      "loss": 0.933,
      "step": 515
    },
    {
      "epoch": 0.9828571428571429,
      "grad_norm": 3.162605788900836,
      "learning_rate": 7.85593252713066e-06,
      "loss": 1.0602,
      "step": 516
    },
    {
      "epoch": 0.9847619047619047,
      "grad_norm": 3.6290752229883885,
      "learning_rate": 7.847482863974351e-06,
      "loss": 1.1061,
      "step": 517
    },
    {
      "epoch": 0.9866666666666667,
      "grad_norm": 3.0548556746777455,
      "learning_rate": 7.839021148165485e-06,
      "loss": 1.0494,
      "step": 518
    },
    {
      "epoch": 0.9885714285714285,
      "grad_norm": 3.2934083286725335,
      "learning_rate": 7.8305474155203e-06,
      "loss": 1.0653,
      "step": 519
    },
    {
      "epoch": 0.9904761904761905,
      "grad_norm": 2.7343965449931935,
      "learning_rate": 7.822061701905896e-06,
      "loss": 1.1359,
      "step": 520
    },
    {
      "epoch": 0.9923809523809524,
      "grad_norm": 2.8988997887295374,
      "learning_rate": 7.813564043240087e-06,
      "loss": 0.9217,
      "step": 521
    },
    {
      "epoch": 0.9942857142857143,
      "grad_norm": 3.1683223411795214,
      "learning_rate": 7.805054475491247e-06,
      "loss": 1.0363,
      "step": 522
    },
    {
      "epoch": 0.9961904761904762,
      "grad_norm": 2.3955236534328908,
      "learning_rate": 7.796533034678155e-06,
      "loss": 1.174,
      "step": 523
    },
    {
      "epoch": 0.9980952380952381,
      "grad_norm": 3.368355854935716,
      "learning_rate": 7.787999756869849e-06,
      "loss": 1.0466,
      "step": 524
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.8272177401137437,
      "learning_rate": 7.779454678185472e-06,
      "loss": 1.1049,
      "step": 525
    },
    {
      "epoch": 1.0019047619047619,
      "grad_norm": 3.278351794680604,
      "learning_rate": 7.770897834794111e-06,
      "loss": 0.78,
      "step": 526
    },
    {
      "epoch": 1.0038095238095237,
      "grad_norm": 3.5979306600958822,
      "learning_rate": 7.762329262914654e-06,
      "loss": 0.7994,
      "step": 527
    },
    {
      "epoch": 1.0057142857142858,
      "grad_norm": 2.459774911918089,
      "learning_rate": 7.753748998815633e-06,
      "loss": 0.591,
      "step": 528
    },
    {
      "epoch": 1.0076190476190476,
      "grad_norm": 2.943287897207152,
      "learning_rate": 7.745157078815068e-06,
      "loss": 0.7519,
      "step": 529
    },
    {
      "epoch": 1.0095238095238095,
      "grad_norm": 3.0078645478736665,
      "learning_rate": 7.736553539280315e-06,
      "loss": 0.8874,
      "step": 530
    },
    {
      "epoch": 1.0114285714285713,
      "grad_norm": 3.920537090277711,
      "learning_rate": 7.727938416627915e-06,
      "loss": 0.8486,
      "step": 531
    },
    {
      "epoch": 1.0133333333333334,
      "grad_norm": 3.4362938183059257,
      "learning_rate": 7.719311747323436e-06,
      "loss": 0.7709,
      "step": 532
    },
    {
      "epoch": 1.0152380952380953,
      "grad_norm": 2.8989669198690793,
      "learning_rate": 7.710673567881322e-06,
      "loss": 0.6786,
      "step": 533
    },
    {
      "epoch": 1.0171428571428571,
      "grad_norm": 3.132026672205312,
      "learning_rate": 7.70202391486473e-06,
      "loss": 0.6602,
      "step": 534
    },
    {
      "epoch": 1.019047619047619,
      "grad_norm": 3.2494009229771903,
      "learning_rate": 7.693362824885388e-06,
      "loss": 0.6235,
      "step": 535
    },
    {
      "epoch": 1.020952380952381,
      "grad_norm": 4.816066275924408,
      "learning_rate": 7.684690334603432e-06,
      "loss": 0.764,
      "step": 536
    },
    {
      "epoch": 1.022857142857143,
      "grad_norm": 2.913912105130997,
      "learning_rate": 7.676006480727255e-06,
      "loss": 0.7966,
      "step": 537
    },
    {
      "epoch": 1.0247619047619048,
      "grad_norm": 3.1319662563763835,
      "learning_rate": 7.667311300013341e-06,
      "loss": 0.6839,
      "step": 538
    },
    {
      "epoch": 1.0266666666666666,
      "grad_norm": 2.6749682101403445,
      "learning_rate": 7.658604829266125e-06,
      "loss": 0.6437,
      "step": 539
    },
    {
      "epoch": 1.0285714285714285,
      "grad_norm": 2.7598153061420927,
      "learning_rate": 7.64988710533783e-06,
      "loss": 0.8466,
      "step": 540
    },
    {
      "epoch": 1.0304761904761905,
      "grad_norm": 3.2302960082380268,
      "learning_rate": 7.641158165128305e-06,
      "loss": 0.7511,
      "step": 541
    },
    {
      "epoch": 1.0323809523809524,
      "grad_norm": 3.2816320724084114,
      "learning_rate": 7.632418045584881e-06,
      "loss": 0.602,
      "step": 542
    },
    {
      "epoch": 1.0342857142857143,
      "grad_norm": 3.043399357123019,
      "learning_rate": 7.623666783702204e-06,
      "loss": 0.5299,
      "step": 543
    },
    {
      "epoch": 1.036190476190476,
      "grad_norm": 3.7126576358318895,
      "learning_rate": 7.614904416522082e-06,
      "loss": 0.6604,
      "step": 544
    },
    {
      "epoch": 1.0380952380952382,
      "grad_norm": 3.293538551297259,
      "learning_rate": 7.6061309811333335e-06,
      "loss": 0.6533,
      "step": 545
    },
    {
      "epoch": 1.04,
      "grad_norm": 3.475721598976322,
      "learning_rate": 7.597346514671624e-06,
      "loss": 0.8101,
      "step": 546
    },
    {
      "epoch": 1.041904761904762,
      "grad_norm": 2.8801230420633104,
      "learning_rate": 7.588551054319306e-06,
      "loss": 0.7262,
      "step": 547
    },
    {
      "epoch": 1.0438095238095237,
      "grad_norm": 3.053328328032738,
      "learning_rate": 7.579744637305273e-06,
      "loss": 0.6568,
      "step": 548
    },
    {
      "epoch": 1.0457142857142858,
      "grad_norm": 2.559613550403858,
      "learning_rate": 7.5709273009047914e-06,
      "loss": 0.5685,
      "step": 549
    },
    {
      "epoch": 1.0476190476190477,
      "grad_norm": 2.2234064518033416,
      "learning_rate": 7.562099082439345e-06,
      "loss": 0.6424,
      "step": 550
    },
    {
      "epoch": 1.0495238095238095,
      "grad_norm": 2.686130863329895,
      "learning_rate": 7.553260019276485e-06,
      "loss": 0.5465,
      "step": 551
    },
    {
      "epoch": 1.0514285714285714,
      "grad_norm": 3.121543742475663,
      "learning_rate": 7.544410148829657e-06,
      "loss": 0.7379,
      "step": 552
    },
    {
      "epoch": 1.0533333333333332,
      "grad_norm": 3.075061107546592,
      "learning_rate": 7.535549508558059e-06,
      "loss": 0.7253,
      "step": 553
    },
    {
      "epoch": 1.0552380952380953,
      "grad_norm": 3.183575567094095,
      "learning_rate": 7.526678135966468e-06,
      "loss": 0.661,
      "step": 554
    },
    {
      "epoch": 1.0571428571428572,
      "grad_norm": 3.72039021918403,
      "learning_rate": 7.517796068605094e-06,
      "loss": 0.5585,
      "step": 555
    },
    {
      "epoch": 1.059047619047619,
      "grad_norm": 2.8858758418603156,
      "learning_rate": 7.508903344069409e-06,
      "loss": 0.7614,
      "step": 556
    },
    {
      "epoch": 1.0609523809523809,
      "grad_norm": 3.0022490934253914,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.5634,
      "step": 557
    },
    {
      "epoch": 1.062857142857143,
      "grad_norm": 2.7282699336617178,
      "learning_rate": 7.491086074082402e-06,
      "loss": 0.7474,
      "step": 558
    },
    {
      "epoch": 1.0647619047619048,
      "grad_norm": 2.9043729096498616,
      "learning_rate": 7.482161604046936e-06,
      "loss": 0.6343,
      "step": 559
    },
    {
      "epoch": 1.0666666666666667,
      "grad_norm": 3.259445858074413,
      "learning_rate": 7.47322662766856e-06,
      "loss": 0.5934,
      "step": 560
    },
    {
      "epoch": 1.0685714285714285,
      "grad_norm": 3.7749218119290853,
      "learning_rate": 7.4642811827666965e-06,
      "loss": 0.6366,
      "step": 561
    },
    {
      "epoch": 1.0704761904761906,
      "grad_norm": 2.954117102936782,
      "learning_rate": 7.455325307205084e-06,
      "loss": 0.6264,
      "step": 562
    },
    {
      "epoch": 1.0723809523809524,
      "grad_norm": 2.7256178855732074,
      "learning_rate": 7.4463590388916075e-06,
      "loss": 0.6833,
      "step": 563
    },
    {
      "epoch": 1.0742857142857143,
      "grad_norm": 2.928500454110158,
      "learning_rate": 7.4373824157781425e-06,
      "loss": 0.6062,
      "step": 564
    },
    {
      "epoch": 1.0761904761904761,
      "grad_norm": 2.6743871619584203,
      "learning_rate": 7.428395475860398e-06,
      "loss": 0.7074,
      "step": 565
    },
    {
      "epoch": 1.078095238095238,
      "grad_norm": 2.4180749186593253,
      "learning_rate": 7.419398257177741e-06,
      "loss": 0.7505,
      "step": 566
    },
    {
      "epoch": 1.08,
      "grad_norm": 2.4380295507604144,
      "learning_rate": 7.410390797813058e-06,
      "loss": 0.6839,
      "step": 567
    },
    {
      "epoch": 1.081904761904762,
      "grad_norm": 3.1902766944170438,
      "learning_rate": 7.401373135892574e-06,
      "loss": 0.7149,
      "step": 568
    },
    {
      "epoch": 1.0838095238095238,
      "grad_norm": 2.9368239492577586,
      "learning_rate": 7.3923453095857015e-06,
      "loss": 0.682,
      "step": 569
    },
    {
      "epoch": 1.0857142857142856,
      "grad_norm": 3.1126220587391202,
      "learning_rate": 7.383307357104874e-06,
      "loss": 0.6853,
      "step": 570
    },
    {
      "epoch": 1.0876190476190477,
      "grad_norm": 3.308310990512469,
      "learning_rate": 7.374259316705388e-06,
      "loss": 0.7989,
      "step": 571
    },
    {
      "epoch": 1.0895238095238096,
      "grad_norm": 2.813660310937022,
      "learning_rate": 7.365201226685241e-06,
      "loss": 0.6996,
      "step": 572
    },
    {
      "epoch": 1.0914285714285714,
      "grad_norm": 2.90914066558398,
      "learning_rate": 7.356133125384966e-06,
      "loss": 0.5593,
      "step": 573
    },
    {
      "epoch": 1.0933333333333333,
      "grad_norm": 3.9102161897815413,
      "learning_rate": 7.3470550511874704e-06,
      "loss": 0.7346,
      "step": 574
    },
    {
      "epoch": 1.0952380952380953,
      "grad_norm": 2.3314594413614906,
      "learning_rate": 7.337967042517876e-06,
      "loss": 0.5884,
      "step": 575
    },
    {
      "epoch": 1.0971428571428572,
      "grad_norm": 3.7294915583346055,
      "learning_rate": 7.3288691378433556e-06,
      "loss": 0.8694,
      "step": 576
    },
    {
      "epoch": 1.099047619047619,
      "grad_norm": 2.241611532747276,
      "learning_rate": 7.319761375672964e-06,
      "loss": 0.7215,
      "step": 577
    },
    {
      "epoch": 1.100952380952381,
      "grad_norm": 3.0065054180640094,
      "learning_rate": 7.310643794557489e-06,
      "loss": 0.6533,
      "step": 578
    },
    {
      "epoch": 1.1028571428571428,
      "grad_norm": 2.5124703183128965,
      "learning_rate": 7.30151643308927e-06,
      "loss": 0.7418,
      "step": 579
    },
    {
      "epoch": 1.1047619047619048,
      "grad_norm": 3.1635163327358726,
      "learning_rate": 7.292379329902049e-06,
      "loss": 0.6065,
      "step": 580
    },
    {
      "epoch": 1.1066666666666667,
      "grad_norm": 3.058968244142787,
      "learning_rate": 7.283232523670805e-06,
      "loss": 0.7095,
      "step": 581
    },
    {
      "epoch": 1.1085714285714285,
      "grad_norm": 3.2744676662799383,
      "learning_rate": 7.274076053111579e-06,
      "loss": 0.6943,
      "step": 582
    },
    {
      "epoch": 1.1104761904761904,
      "grad_norm": 3.150223760389491,
      "learning_rate": 7.2649099569813264e-06,
      "loss": 0.6957,
      "step": 583
    },
    {
      "epoch": 1.1123809523809525,
      "grad_norm": 3.0837214519697955,
      "learning_rate": 7.255734274077742e-06,
      "loss": 0.7712,
      "step": 584
    },
    {
      "epoch": 1.1142857142857143,
      "grad_norm": 3.2997911569532996,
      "learning_rate": 7.246549043239098e-06,
      "loss": 0.6497,
      "step": 585
    },
    {
      "epoch": 1.1161904761904762,
      "grad_norm": 3.629950209092766,
      "learning_rate": 7.237354303344083e-06,
      "loss": 0.697,
      "step": 586
    },
    {
      "epoch": 1.118095238095238,
      "grad_norm": 2.1997562865480624,
      "learning_rate": 7.228150093311634e-06,
      "loss": 0.6353,
      "step": 587
    },
    {
      "epoch": 1.12,
      "grad_norm": 3.35340362360685,
      "learning_rate": 7.218936452100771e-06,
      "loss": 0.6252,
      "step": 588
    },
    {
      "epoch": 1.121904761904762,
      "grad_norm": 2.5596578364560845,
      "learning_rate": 7.209713418710437e-06,
      "loss": 1.01,
      "step": 589
    },
    {
      "epoch": 1.1238095238095238,
      "grad_norm": 3.0831464502774755,
      "learning_rate": 7.200481032179328e-06,
      "loss": 0.5934,
      "step": 590
    },
    {
      "epoch": 1.1257142857142857,
      "grad_norm": 3.1422531731636396,
      "learning_rate": 7.1912393315857245e-06,
      "loss": 0.7947,
      "step": 591
    },
    {
      "epoch": 1.1276190476190475,
      "grad_norm": 4.120755328615173,
      "learning_rate": 7.1819883560473395e-06,
      "loss": 0.7441,
      "step": 592
    },
    {
      "epoch": 1.1295238095238096,
      "grad_norm": 2.782211801179133,
      "learning_rate": 7.172728144721138e-06,
      "loss": 0.6375,
      "step": 593
    },
    {
      "epoch": 1.1314285714285715,
      "grad_norm": 2.951628940328863,
      "learning_rate": 7.163458736803184e-06,
      "loss": 0.7329,
      "step": 594
    },
    {
      "epoch": 1.1333333333333333,
      "grad_norm": 3.9215786491138362,
      "learning_rate": 7.154180171528462e-06,
      "loss": 0.6429,
      "step": 595
    },
    {
      "epoch": 1.1352380952380952,
      "grad_norm": 3.7546044540408747,
      "learning_rate": 7.144892488170719e-06,
      "loss": 0.6793,
      "step": 596
    },
    {
      "epoch": 1.1371428571428572,
      "grad_norm": 3.0558176842479634,
      "learning_rate": 7.135595726042299e-06,
      "loss": 0.7436,
      "step": 597
    },
    {
      "epoch": 1.139047619047619,
      "grad_norm": 3.0034861847596446,
      "learning_rate": 7.126289924493971e-06,
      "loss": 0.6186,
      "step": 598
    },
    {
      "epoch": 1.140952380952381,
      "grad_norm": 3.2711780797566945,
      "learning_rate": 7.116975122914767e-06,
      "loss": 0.6456,
      "step": 599
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 2.76214778973002,
      "learning_rate": 7.107651360731812e-06,
      "loss": 0.5644,
      "step": 600
    },
    {
      "epoch": 1.1447619047619049,
      "grad_norm": 4.036426978694489,
      "learning_rate": 7.098318677410163e-06,
      "loss": 0.6523,
      "step": 601
    },
    {
      "epoch": 1.1466666666666667,
      "grad_norm": 3.028625517033124,
      "learning_rate": 7.088977112452634e-06,
      "loss": 0.7852,
      "step": 602
    },
    {
      "epoch": 1.1485714285714286,
      "grad_norm": 2.5915779061337045,
      "learning_rate": 7.079626705399636e-06,
      "loss": 0.6592,
      "step": 603
    },
    {
      "epoch": 1.1504761904761904,
      "grad_norm": 3.0988196962230634,
      "learning_rate": 7.070267495829e-06,
      "loss": 0.7292,
      "step": 604
    },
    {
      "epoch": 1.1523809523809523,
      "grad_norm": 3.06446152236741,
      "learning_rate": 7.060899523355822e-06,
      "loss": 0.6567,
      "step": 605
    },
    {
      "epoch": 1.1542857142857144,
      "grad_norm": 3.192339444894251,
      "learning_rate": 7.051522827632288e-06,
      "loss": 0.715,
      "step": 606
    },
    {
      "epoch": 1.1561904761904762,
      "grad_norm": 3.1922700149619097,
      "learning_rate": 7.042137448347504e-06,
      "loss": 0.7048,
      "step": 607
    },
    {
      "epoch": 1.158095238095238,
      "grad_norm": 2.914801277992762,
      "learning_rate": 7.032743425227336e-06,
      "loss": 0.7237,
      "step": 608
    },
    {
      "epoch": 1.16,
      "grad_norm": 2.8583540958618325,
      "learning_rate": 7.023340798034234e-06,
      "loss": 0.664,
      "step": 609
    },
    {
      "epoch": 1.161904761904762,
      "grad_norm": 2.8405966340374302,
      "learning_rate": 7.013929606567067e-06,
      "loss": 0.8086,
      "step": 610
    },
    {
      "epoch": 1.1638095238095238,
      "grad_norm": 2.96081352241689,
      "learning_rate": 7.004509890660955e-06,
      "loss": 0.7387,
      "step": 611
    },
    {
      "epoch": 1.1657142857142857,
      "grad_norm": 2.851385859021923,
      "learning_rate": 6.995081690187097e-06,
      "loss": 0.7045,
      "step": 612
    },
    {
      "epoch": 1.1676190476190476,
      "grad_norm": 2.783273572615768,
      "learning_rate": 6.985645045052611e-06,
      "loss": 0.5878,
      "step": 613
    },
    {
      "epoch": 1.1695238095238096,
      "grad_norm": 2.6671554453652178,
      "learning_rate": 6.976199995200354e-06,
      "loss": 0.7057,
      "step": 614
    },
    {
      "epoch": 1.1714285714285715,
      "grad_norm": 2.72720234773299,
      "learning_rate": 6.966746580608761e-06,
      "loss": 0.6468,
      "step": 615
    },
    {
      "epoch": 1.1733333333333333,
      "grad_norm": 2.7363828971637454,
      "learning_rate": 6.95728484129167e-06,
      "loss": 0.6577,
      "step": 616
    },
    {
      "epoch": 1.1752380952380952,
      "grad_norm": 3.0876827077248183,
      "learning_rate": 6.947814817298155e-06,
      "loss": 0.7497,
      "step": 617
    },
    {
      "epoch": 1.177142857142857,
      "grad_norm": 3.330842295767064,
      "learning_rate": 6.93833654871236e-06,
      "loss": 0.6806,
      "step": 618
    },
    {
      "epoch": 1.1790476190476191,
      "grad_norm": 2.9509529527079117,
      "learning_rate": 6.928850075653326e-06,
      "loss": 0.682,
      "step": 619
    },
    {
      "epoch": 1.180952380952381,
      "grad_norm": 2.5296803223351705,
      "learning_rate": 6.919355438274817e-06,
      "loss": 0.6394,
      "step": 620
    },
    {
      "epoch": 1.1828571428571428,
      "grad_norm": 2.6766781705506006,
      "learning_rate": 6.909852676765161e-06,
      "loss": 0.6751,
      "step": 621
    },
    {
      "epoch": 1.1847619047619047,
      "grad_norm": 2.6692385415608926,
      "learning_rate": 6.900341831347067e-06,
      "loss": 0.6577,
      "step": 622
    },
    {
      "epoch": 1.1866666666666668,
      "grad_norm": 2.895449612452324,
      "learning_rate": 6.8908229422774666e-06,
      "loss": 0.7271,
      "step": 623
    },
    {
      "epoch": 1.1885714285714286,
      "grad_norm": 2.572667406001419,
      "learning_rate": 6.881296049847333e-06,
      "loss": 0.6749,
      "step": 624
    },
    {
      "epoch": 1.1904761904761905,
      "grad_norm": 2.3550347849478284,
      "learning_rate": 6.8717611943815185e-06,
      "loss": 0.7193,
      "step": 625
    },
    {
      "epoch": 1.1923809523809523,
      "grad_norm": 2.9313104534606502,
      "learning_rate": 6.862218416238582e-06,
      "loss": 0.7198,
      "step": 626
    },
    {
      "epoch": 1.1942857142857144,
      "grad_norm": 2.684143088959432,
      "learning_rate": 6.852667755810614e-06,
      "loss": 0.7345,
      "step": 627
    },
    {
      "epoch": 1.1961904761904762,
      "grad_norm": 2.57853016776272,
      "learning_rate": 6.84310925352307e-06,
      "loss": 0.6601,
      "step": 628
    },
    {
      "epoch": 1.198095238095238,
      "grad_norm": 3.244096053912211,
      "learning_rate": 6.8335429498346e-06,
      "loss": 0.5982,
      "step": 629
    },
    {
      "epoch": 1.2,
      "grad_norm": 2.904330283712291,
      "learning_rate": 6.823968885236869e-06,
      "loss": 0.6148,
      "step": 630
    },
    {
      "epoch": 1.2019047619047618,
      "grad_norm": 2.813833695416442,
      "learning_rate": 6.814387100254399e-06,
      "loss": 0.7887,
      "step": 631
    },
    {
      "epoch": 1.2038095238095239,
      "grad_norm": 3.013400807925236,
      "learning_rate": 6.804797635444389e-06,
      "loss": 0.7622,
      "step": 632
    },
    {
      "epoch": 1.2057142857142857,
      "grad_norm": 3.247727385129131,
      "learning_rate": 6.795200531396542e-06,
      "loss": 0.661,
      "step": 633
    },
    {
      "epoch": 1.2076190476190476,
      "grad_norm": 3.2811715636074616,
      "learning_rate": 6.785595828732896e-06,
      "loss": 0.681,
      "step": 634
    },
    {
      "epoch": 1.2095238095238094,
      "grad_norm": 2.9902563399511664,
      "learning_rate": 6.775983568107653e-06,
      "loss": 0.7117,
      "step": 635
    },
    {
      "epoch": 1.2114285714285715,
      "grad_norm": 2.243652325070851,
      "learning_rate": 6.766363790207009e-06,
      "loss": 0.5765,
      "step": 636
    },
    {
      "epoch": 1.2133333333333334,
      "grad_norm": 2.5735329670900917,
      "learning_rate": 6.756736535748974e-06,
      "loss": 0.7017,
      "step": 637
    },
    {
      "epoch": 1.2152380952380952,
      "grad_norm": 3.80153409494151,
      "learning_rate": 6.747101845483206e-06,
      "loss": 0.7943,
      "step": 638
    },
    {
      "epoch": 1.217142857142857,
      "grad_norm": 2.6745618332686085,
      "learning_rate": 6.7374597601908365e-06,
      "loss": 0.6102,
      "step": 639
    },
    {
      "epoch": 1.2190476190476192,
      "grad_norm": 3.5009554061622836,
      "learning_rate": 6.7278103206843005e-06,
      "loss": 0.7141,
      "step": 640
    },
    {
      "epoch": 1.220952380952381,
      "grad_norm": 3.296137766731785,
      "learning_rate": 6.718153567807157e-06,
      "loss": 0.7761,
      "step": 641
    },
    {
      "epoch": 1.2228571428571429,
      "grad_norm": 2.3029113743264755,
      "learning_rate": 6.7084895424339254e-06,
      "loss": 0.7787,
      "step": 642
    },
    {
      "epoch": 1.2247619047619047,
      "grad_norm": 2.699693904526732,
      "learning_rate": 6.6988182854699045e-06,
      "loss": 0.6739,
      "step": 643
    },
    {
      "epoch": 1.2266666666666666,
      "grad_norm": 2.7706641633621936,
      "learning_rate": 6.6891398378510034e-06,
      "loss": 0.9543,
      "step": 644
    },
    {
      "epoch": 1.2285714285714286,
      "grad_norm": 3.114602507663263,
      "learning_rate": 6.679454240543567e-06,
      "loss": 0.7415,
      "step": 645
    },
    {
      "epoch": 1.2304761904761905,
      "grad_norm": 3.748018389276035,
      "learning_rate": 6.669761534544205e-06,
      "loss": 0.6831,
      "step": 646
    },
    {
      "epoch": 1.2323809523809524,
      "grad_norm": 2.932045476585695,
      "learning_rate": 6.6600617608796126e-06,
      "loss": 0.566,
      "step": 647
    },
    {
      "epoch": 1.2342857142857142,
      "grad_norm": 3.4254357212361652,
      "learning_rate": 6.650354960606405e-06,
      "loss": 0.7355,
      "step": 648
    },
    {
      "epoch": 1.2361904761904763,
      "grad_norm": 3.2171574185270275,
      "learning_rate": 6.640641174810934e-06,
      "loss": 0.7746,
      "step": 649
    },
    {
      "epoch": 1.2380952380952381,
      "grad_norm": 3.25023258863064,
      "learning_rate": 6.630920444609127e-06,
      "loss": 0.5981,
      "step": 650
    },
    {
      "epoch": 1.24,
      "grad_norm": 3.4511631219101995,
      "learning_rate": 6.621192811146297e-06,
      "loss": 0.6065,
      "step": 651
    },
    {
      "epoch": 1.2419047619047618,
      "grad_norm": 2.4793514657437696,
      "learning_rate": 6.611458315596979e-06,
      "loss": 0.7514,
      "step": 652
    },
    {
      "epoch": 1.243809523809524,
      "grad_norm": 2.842659106653368,
      "learning_rate": 6.601716999164759e-06,
      "loss": 0.6749,
      "step": 653
    },
    {
      "epoch": 1.2457142857142858,
      "grad_norm": 2.447668155196946,
      "learning_rate": 6.591968903082087e-06,
      "loss": 0.7702,
      "step": 654
    },
    {
      "epoch": 1.2476190476190476,
      "grad_norm": 2.4860929729475623,
      "learning_rate": 6.582214068610111e-06,
      "loss": 0.6024,
      "step": 655
    },
    {
      "epoch": 1.2495238095238095,
      "grad_norm": 2.7927351396236917,
      "learning_rate": 6.572452537038504e-06,
      "loss": 0.805,
      "step": 656
    },
    {
      "epoch": 1.2514285714285713,
      "grad_norm": 2.330341372653887,
      "learning_rate": 6.562684349685283e-06,
      "loss": 0.5915,
      "step": 657
    },
    {
      "epoch": 1.2533333333333334,
      "grad_norm": 2.6353598138337517,
      "learning_rate": 6.55290954789664e-06,
      "loss": 0.6989,
      "step": 658
    },
    {
      "epoch": 1.2552380952380953,
      "grad_norm": 3.4826327690358707,
      "learning_rate": 6.54312817304676e-06,
      "loss": 0.672,
      "step": 659
    },
    {
      "epoch": 1.2571428571428571,
      "grad_norm": 3.9266385131697135,
      "learning_rate": 6.533340266537656e-06,
      "loss": 0.7152,
      "step": 660
    },
    {
      "epoch": 1.259047619047619,
      "grad_norm": 3.311353164852456,
      "learning_rate": 6.52354586979898e-06,
      "loss": 0.6168,
      "step": 661
    },
    {
      "epoch": 1.2609523809523808,
      "grad_norm": 2.6540745304501554,
      "learning_rate": 6.513745024287861e-06,
      "loss": 0.6757,
      "step": 662
    },
    {
      "epoch": 1.262857142857143,
      "grad_norm": 2.5763142149127174,
      "learning_rate": 6.503937771488724e-06,
      "loss": 0.5506,
      "step": 663
    },
    {
      "epoch": 1.2647619047619048,
      "grad_norm": 2.7276428205823327,
      "learning_rate": 6.494124152913113e-06,
      "loss": 0.6259,
      "step": 664
    },
    {
      "epoch": 1.2666666666666666,
      "grad_norm": 3.5755199923419405,
      "learning_rate": 6.484304210099517e-06,
      "loss": 0.7152,
      "step": 665
    },
    {
      "epoch": 1.2685714285714287,
      "grad_norm": 3.7130504152492914,
      "learning_rate": 6.474477984613192e-06,
      "loss": 0.6839,
      "step": 666
    },
    {
      "epoch": 1.2704761904761905,
      "grad_norm": 3.1016323384414277,
      "learning_rate": 6.464645518045991e-06,
      "loss": 0.8442,
      "step": 667
    },
    {
      "epoch": 1.2723809523809524,
      "grad_norm": 2.374826299733099,
      "learning_rate": 6.454806852016179e-06,
      "loss": 0.6381,
      "step": 668
    },
    {
      "epoch": 1.2742857142857142,
      "grad_norm": 2.20642420777723,
      "learning_rate": 6.444962028168266e-06,
      "loss": 0.7221,
      "step": 669
    },
    {
      "epoch": 1.276190476190476,
      "grad_norm": 3.4295788673888477,
      "learning_rate": 6.435111088172823e-06,
      "loss": 0.617,
      "step": 670
    },
    {
      "epoch": 1.2780952380952382,
      "grad_norm": 2.655800684712996,
      "learning_rate": 6.425254073726311e-06,
      "loss": 0.6748,
      "step": 671
    },
    {
      "epoch": 1.28,
      "grad_norm": 3.077761933655665,
      "learning_rate": 6.415391026550903e-06,
      "loss": 0.6371,
      "step": 672
    },
    {
      "epoch": 1.2819047619047619,
      "grad_norm": 2.7718500409566946,
      "learning_rate": 6.405521988394305e-06,
      "loss": 0.7329,
      "step": 673
    },
    {
      "epoch": 1.2838095238095237,
      "grad_norm": 3.4275730335541748,
      "learning_rate": 6.395647001029586e-06,
      "loss": 0.7096,
      "step": 674
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 2.9976507369396623,
      "learning_rate": 6.385766106254987e-06,
      "loss": 0.6088,
      "step": 675
    },
    {
      "epoch": 1.2876190476190477,
      "grad_norm": 3.1837382805231202,
      "learning_rate": 6.375879345893763e-06,
      "loss": 0.725,
      "step": 676
    },
    {
      "epoch": 1.2895238095238095,
      "grad_norm": 2.5997684793341698,
      "learning_rate": 6.365986761793992e-06,
      "loss": 0.6453,
      "step": 677
    },
    {
      "epoch": 1.2914285714285714,
      "grad_norm": 2.6334802846490284,
      "learning_rate": 6.356088395828404e-06,
      "loss": 0.67,
      "step": 678
    },
    {
      "epoch": 1.2933333333333334,
      "grad_norm": 2.8475343588361572,
      "learning_rate": 6.346184289894202e-06,
      "loss": 0.718,
      "step": 679
    },
    {
      "epoch": 1.2952380952380953,
      "grad_norm": 3.293427470182712,
      "learning_rate": 6.336274485912881e-06,
      "loss": 0.8292,
      "step": 680
    },
    {
      "epoch": 1.2971428571428572,
      "grad_norm": 3.9640201165292663,
      "learning_rate": 6.326359025830061e-06,
      "loss": 0.6829,
      "step": 681
    },
    {
      "epoch": 1.299047619047619,
      "grad_norm": 2.75330361872787,
      "learning_rate": 6.3164379516152995e-06,
      "loss": 0.8595,
      "step": 682
    },
    {
      "epoch": 1.3009523809523809,
      "grad_norm": 3.3356147097372935,
      "learning_rate": 6.306511305261914e-06,
      "loss": 0.6987,
      "step": 683
    },
    {
      "epoch": 1.302857142857143,
      "grad_norm": 4.349370713917045,
      "learning_rate": 6.29657912878681e-06,
      "loss": 0.762,
      "step": 684
    },
    {
      "epoch": 1.3047619047619048,
      "grad_norm": 3.0295474631604953,
      "learning_rate": 6.2866414642303054e-06,
      "loss": 0.5335,
      "step": 685
    },
    {
      "epoch": 1.3066666666666666,
      "grad_norm": 2.912987585897963,
      "learning_rate": 6.2766983536559404e-06,
      "loss": 0.6494,
      "step": 686
    },
    {
      "epoch": 1.3085714285714285,
      "grad_norm": 3.3460715358833086,
      "learning_rate": 6.266749839150309e-06,
      "loss": 0.6154,
      "step": 687
    },
    {
      "epoch": 1.3104761904761904,
      "grad_norm": 2.965154041904548,
      "learning_rate": 6.256795962822881e-06,
      "loss": 0.6863,
      "step": 688
    },
    {
      "epoch": 1.3123809523809524,
      "grad_norm": 3.554960125513718,
      "learning_rate": 6.2468367668058195e-06,
      "loss": 0.8489,
      "step": 689
    },
    {
      "epoch": 1.3142857142857143,
      "grad_norm": 3.117852691852733,
      "learning_rate": 6.236872293253803e-06,
      "loss": 0.8616,
      "step": 690
    },
    {
      "epoch": 1.3161904761904761,
      "grad_norm": 3.39993417136851,
      "learning_rate": 6.22690258434385e-06,
      "loss": 0.6842,
      "step": 691
    },
    {
      "epoch": 1.3180952380952382,
      "grad_norm": 3.016683033095873,
      "learning_rate": 6.2169276822751425e-06,
      "loss": 0.7003,
      "step": 692
    },
    {
      "epoch": 1.32,
      "grad_norm": 2.535887025186407,
      "learning_rate": 6.206947629268838e-06,
      "loss": 0.6624,
      "step": 693
    },
    {
      "epoch": 1.321904761904762,
      "grad_norm": 3.193173944123266,
      "learning_rate": 6.196962467567897e-06,
      "loss": 0.7883,
      "step": 694
    },
    {
      "epoch": 1.3238095238095238,
      "grad_norm": 2.833232000100724,
      "learning_rate": 6.186972239436911e-06,
      "loss": 0.6097,
      "step": 695
    },
    {
      "epoch": 1.3257142857142856,
      "grad_norm": 3.816531669777505,
      "learning_rate": 6.1769769871619035e-06,
      "loss": 0.7582,
      "step": 696
    },
    {
      "epoch": 1.3276190476190477,
      "grad_norm": 2.727282902700753,
      "learning_rate": 6.166976753050176e-06,
      "loss": 0.7223,
      "step": 697
    },
    {
      "epoch": 1.3295238095238096,
      "grad_norm": 2.47254221153791,
      "learning_rate": 6.156971579430111e-06,
      "loss": 0.7326,
      "step": 698
    },
    {
      "epoch": 1.3314285714285714,
      "grad_norm": 3.3277609741062713,
      "learning_rate": 6.146961508651002e-06,
      "loss": 0.6196,
      "step": 699
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 2.488166322666529,
      "learning_rate": 6.136946583082866e-06,
      "loss": 0.5561,
      "step": 700
    },
    {
      "epoch": 1.3352380952380951,
      "grad_norm": 3.5366781350749448,
      "learning_rate": 6.126926845116272e-06,
      "loss": 0.6788,
      "step": 701
    },
    {
      "epoch": 1.3371428571428572,
      "grad_norm": 2.8607003125373467,
      "learning_rate": 6.11690233716216e-06,
      "loss": 0.7793,
      "step": 702
    },
    {
      "epoch": 1.339047619047619,
      "grad_norm": 3.0435458215778866,
      "learning_rate": 6.106873101651659e-06,
      "loss": 0.7998,
      "step": 703
    },
    {
      "epoch": 1.340952380952381,
      "grad_norm": 3.2575729935155735,
      "learning_rate": 6.096839181035905e-06,
      "loss": 0.5916,
      "step": 704
    },
    {
      "epoch": 1.342857142857143,
      "grad_norm": 2.9063740802818665,
      "learning_rate": 6.08680061778587e-06,
      "loss": 0.741,
      "step": 705
    },
    {
      "epoch": 1.3447619047619048,
      "grad_norm": 3.9752753569114696,
      "learning_rate": 6.076757454392176e-06,
      "loss": 0.6629,
      "step": 706
    },
    {
      "epoch": 1.3466666666666667,
      "grad_norm": 2.7449874531756864,
      "learning_rate": 6.066709733364914e-06,
      "loss": 0.7444,
      "step": 707
    },
    {
      "epoch": 1.3485714285714285,
      "grad_norm": 3.4410814124933573,
      "learning_rate": 6.056657497233465e-06,
      "loss": 0.658,
      "step": 708
    },
    {
      "epoch": 1.3504761904761904,
      "grad_norm": 3.7547841103755286,
      "learning_rate": 6.046600788546328e-06,
      "loss": 0.7313,
      "step": 709
    },
    {
      "epoch": 1.3523809523809525,
      "grad_norm": 2.816106699814712,
      "learning_rate": 6.036539649870928e-06,
      "loss": 0.6924,
      "step": 710
    },
    {
      "epoch": 1.3542857142857143,
      "grad_norm": 2.7898661832286744,
      "learning_rate": 6.026474123793436e-06,
      "loss": 0.7206,
      "step": 711
    },
    {
      "epoch": 1.3561904761904762,
      "grad_norm": 2.9719654540369307,
      "learning_rate": 6.0164042529186075e-06,
      "loss": 0.6863,
      "step": 712
    },
    {
      "epoch": 1.358095238095238,
      "grad_norm": 2.547070736691609,
      "learning_rate": 6.006330079869577e-06,
      "loss": 0.6067,
      "step": 713
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 3.514053199524337,
      "learning_rate": 5.996251647287692e-06,
      "loss": 0.9087,
      "step": 714
    },
    {
      "epoch": 1.361904761904762,
      "grad_norm": 2.6237497788605113,
      "learning_rate": 5.98616899783233e-06,
      "loss": 0.5744,
      "step": 715
    },
    {
      "epoch": 1.3638095238095238,
      "grad_norm": 3.041927925551627,
      "learning_rate": 5.97608217418072e-06,
      "loss": 0.7331,
      "step": 716
    },
    {
      "epoch": 1.3657142857142857,
      "grad_norm": 3.915936199058181,
      "learning_rate": 5.965991219027753e-06,
      "loss": 0.6955,
      "step": 717
    },
    {
      "epoch": 1.3676190476190477,
      "grad_norm": 3.5420297523927675,
      "learning_rate": 5.9558961750858114e-06,
      "loss": 0.7517,
      "step": 718
    },
    {
      "epoch": 1.3695238095238096,
      "grad_norm": 2.9957096252738973,
      "learning_rate": 5.945797085084587e-06,
      "loss": 0.6347,
      "step": 719
    },
    {
      "epoch": 1.3714285714285714,
      "grad_norm": 2.2230170081613587,
      "learning_rate": 5.9356939917708945e-06,
      "loss": 0.6186,
      "step": 720
    },
    {
      "epoch": 1.3733333333333333,
      "grad_norm": 3.042777312528671,
      "learning_rate": 5.925586937908491e-06,
      "loss": 0.6627,
      "step": 721
    },
    {
      "epoch": 1.3752380952380951,
      "grad_norm": 2.8848699148590016,
      "learning_rate": 5.915475966277901e-06,
      "loss": 0.6201,
      "step": 722
    },
    {
      "epoch": 1.3771428571428572,
      "grad_norm": 2.5897752694710703,
      "learning_rate": 5.905361119676234e-06,
      "loss": 0.5931,
      "step": 723
    },
    {
      "epoch": 1.379047619047619,
      "grad_norm": 2.5972834657518744,
      "learning_rate": 5.895242440916996e-06,
      "loss": 0.6679,
      "step": 724
    },
    {
      "epoch": 1.380952380952381,
      "grad_norm": 2.395772503877728,
      "learning_rate": 5.885119972829915e-06,
      "loss": 0.7798,
      "step": 725
    },
    {
      "epoch": 1.3828571428571428,
      "grad_norm": 3.001203393028994,
      "learning_rate": 5.874993758260762e-06,
      "loss": 0.6708,
      "step": 726
    },
    {
      "epoch": 1.3847619047619046,
      "grad_norm": 2.602232348223643,
      "learning_rate": 5.864863840071162e-06,
      "loss": 0.7032,
      "step": 727
    },
    {
      "epoch": 1.3866666666666667,
      "grad_norm": 2.7432460005389565,
      "learning_rate": 5.854730261138418e-06,
      "loss": 0.8601,
      "step": 728
    },
    {
      "epoch": 1.3885714285714286,
      "grad_norm": 4.492597819123363,
      "learning_rate": 5.844593064355327e-06,
      "loss": 0.7868,
      "step": 729
    },
    {
      "epoch": 1.3904761904761904,
      "grad_norm": 3.174561526439116,
      "learning_rate": 5.834452292629999e-06,
      "loss": 0.6974,
      "step": 730
    },
    {
      "epoch": 1.3923809523809525,
      "grad_norm": 2.7381709410077377,
      "learning_rate": 5.8243079888856806e-06,
      "loss": 0.7537,
      "step": 731
    },
    {
      "epoch": 1.3942857142857144,
      "grad_norm": 2.814631088171071,
      "learning_rate": 5.814160196060559e-06,
      "loss": 0.6452,
      "step": 732
    },
    {
      "epoch": 1.3961904761904762,
      "grad_norm": 2.4498587166561796,
      "learning_rate": 5.8040089571075995e-06,
      "loss": 0.7442,
      "step": 733
    },
    {
      "epoch": 1.398095238095238,
      "grad_norm": 3.2496619835380596,
      "learning_rate": 5.7938543149943495e-06,
      "loss": 0.6019,
      "step": 734
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.032238546119449,
      "learning_rate": 5.78369631270276e-06,
      "loss": 0.6778,
      "step": 735
    },
    {
      "epoch": 1.401904761904762,
      "grad_norm": 3.140057835205741,
      "learning_rate": 5.773534993229007e-06,
      "loss": 0.7595,
      "step": 736
    },
    {
      "epoch": 1.4038095238095238,
      "grad_norm": 2.871132731659619,
      "learning_rate": 5.763370399583307e-06,
      "loss": 0.5589,
      "step": 737
    },
    {
      "epoch": 1.4057142857142857,
      "grad_norm": 3.1300745174835147,
      "learning_rate": 5.753202574789735e-06,
      "loss": 0.7086,
      "step": 738
    },
    {
      "epoch": 1.4076190476190475,
      "grad_norm": 2.79437180770245,
      "learning_rate": 5.743031561886039e-06,
      "loss": 0.6284,
      "step": 739
    },
    {
      "epoch": 1.4095238095238094,
      "grad_norm": 2.8627209838146364,
      "learning_rate": 5.7328574039234696e-06,
      "loss": 0.7121,
      "step": 740
    },
    {
      "epoch": 1.4114285714285715,
      "grad_norm": 3.7859419380269035,
      "learning_rate": 5.722680143966582e-06,
      "loss": 0.5724,
      "step": 741
    },
    {
      "epoch": 1.4133333333333333,
      "grad_norm": 2.479898353947677,
      "learning_rate": 5.7124998250930665e-06,
      "loss": 0.6805,
      "step": 742
    },
    {
      "epoch": 1.4152380952380952,
      "grad_norm": 3.3140708190612567,
      "learning_rate": 5.702316490393556e-06,
      "loss": 0.6668,
      "step": 743
    },
    {
      "epoch": 1.4171428571428573,
      "grad_norm": 3.0177328733621747,
      "learning_rate": 5.6921301829714525e-06,
      "loss": 0.7792,
      "step": 744
    },
    {
      "epoch": 1.4190476190476191,
      "grad_norm": 3.0086287709904362,
      "learning_rate": 5.681940945942739e-06,
      "loss": 0.5872,
      "step": 745
    },
    {
      "epoch": 1.420952380952381,
      "grad_norm": 3.519233924180684,
      "learning_rate": 5.6717488224358e-06,
      "loss": 0.6653,
      "step": 746
    },
    {
      "epoch": 1.4228571428571428,
      "grad_norm": 4.638840193856848,
      "learning_rate": 5.661553855591237e-06,
      "loss": 0.7945,
      "step": 747
    },
    {
      "epoch": 1.4247619047619047,
      "grad_norm": 2.9267759360744137,
      "learning_rate": 5.651356088561685e-06,
      "loss": 0.6684,
      "step": 748
    },
    {
      "epoch": 1.4266666666666667,
      "grad_norm": 3.046637456908023,
      "learning_rate": 5.641155564511634e-06,
      "loss": 0.7486,
      "step": 749
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 3.016922422419726,
      "learning_rate": 5.630952326617243e-06,
      "loss": 0.7607,
      "step": 750
    },
    {
      "epoch": 1.4304761904761905,
      "grad_norm": 2.077081545887961,
      "learning_rate": 5.620746418066158e-06,
      "loss": 0.6368,
      "step": 751
    },
    {
      "epoch": 1.4323809523809523,
      "grad_norm": 2.222786815953259,
      "learning_rate": 5.610537882057329e-06,
      "loss": 0.7579,
      "step": 752
    },
    {
      "epoch": 1.4342857142857142,
      "grad_norm": 2.8408310035328674,
      "learning_rate": 5.600326761800826e-06,
      "loss": 0.6567,
      "step": 753
    },
    {
      "epoch": 1.4361904761904762,
      "grad_norm": 3.620772954265577,
      "learning_rate": 5.59011310051766e-06,
      "loss": 0.8617,
      "step": 754
    },
    {
      "epoch": 1.438095238095238,
      "grad_norm": 3.4873578497643583,
      "learning_rate": 5.579896941439596e-06,
      "loss": 0.7156,
      "step": 755
    },
    {
      "epoch": 1.44,
      "grad_norm": 3.361434073720694,
      "learning_rate": 5.5696783278089724e-06,
      "loss": 0.6971,
      "step": 756
    },
    {
      "epoch": 1.441904761904762,
      "grad_norm": 2.412579114731131,
      "learning_rate": 5.559457302878517e-06,
      "loss": 0.5188,
      "step": 757
    },
    {
      "epoch": 1.4438095238095239,
      "grad_norm": 2.813791279931654,
      "learning_rate": 5.54923390991116e-06,
      "loss": 0.7952,
      "step": 758
    },
    {
      "epoch": 1.4457142857142857,
      "grad_norm": 3.2034583719282215,
      "learning_rate": 5.539008192179862e-06,
      "loss": 0.5124,
      "step": 759
    },
    {
      "epoch": 1.4476190476190476,
      "grad_norm": 2.5104234596805384,
      "learning_rate": 5.528780192967419e-06,
      "loss": 0.6554,
      "step": 760
    },
    {
      "epoch": 1.4495238095238094,
      "grad_norm": 4.1105396484207075,
      "learning_rate": 5.5185499555662845e-06,
      "loss": 0.7394,
      "step": 761
    },
    {
      "epoch": 1.4514285714285715,
      "grad_norm": 4.273975163273313,
      "learning_rate": 5.508317523278386e-06,
      "loss": 0.7807,
      "step": 762
    },
    {
      "epoch": 1.4533333333333334,
      "grad_norm": 2.946980421500577,
      "learning_rate": 5.498082939414943e-06,
      "loss": 0.6459,
      "step": 763
    },
    {
      "epoch": 1.4552380952380952,
      "grad_norm": 3.7878732338823076,
      "learning_rate": 5.487846247296278e-06,
      "loss": 0.7925,
      "step": 764
    },
    {
      "epoch": 1.457142857142857,
      "grad_norm": 2.8106610260560903,
      "learning_rate": 5.477607490251642e-06,
      "loss": 0.648,
      "step": 765
    },
    {
      "epoch": 1.459047619047619,
      "grad_norm": 1.986311245912668,
      "learning_rate": 5.467366711619024e-06,
      "loss": 0.6146,
      "step": 766
    },
    {
      "epoch": 1.460952380952381,
      "grad_norm": 2.8887232997693886,
      "learning_rate": 5.457123954744967e-06,
      "loss": 0.9805,
      "step": 767
    },
    {
      "epoch": 1.4628571428571429,
      "grad_norm": 3.2023413077412823,
      "learning_rate": 5.4468792629843944e-06,
      "loss": 0.5809,
      "step": 768
    },
    {
      "epoch": 1.4647619047619047,
      "grad_norm": 3.419380543553438,
      "learning_rate": 5.436632679700414e-06,
      "loss": 0.7441,
      "step": 769
    },
    {
      "epoch": 1.4666666666666668,
      "grad_norm": 3.562089726232499,
      "learning_rate": 5.42638424826414e-06,
      "loss": 0.7045,
      "step": 770
    },
    {
      "epoch": 1.4685714285714286,
      "grad_norm": 2.4486399824970544,
      "learning_rate": 5.416134012054512e-06,
      "loss": 0.6379,
      "step": 771
    },
    {
      "epoch": 1.4704761904761905,
      "grad_norm": 4.0111382244553,
      "learning_rate": 5.405882014458108e-06,
      "loss": 0.7735,
      "step": 772
    },
    {
      "epoch": 1.4723809523809523,
      "grad_norm": 3.1008837360887123,
      "learning_rate": 5.395628298868959e-06,
      "loss": 0.6142,
      "step": 773
    },
    {
      "epoch": 1.4742857142857142,
      "grad_norm": 3.076309986647131,
      "learning_rate": 5.385372908688371e-06,
      "loss": 0.6175,
      "step": 774
    },
    {
      "epoch": 1.4761904761904763,
      "grad_norm": 3.5369815120040684,
      "learning_rate": 5.375115887324736e-06,
      "loss": 0.7672,
      "step": 775
    },
    {
      "epoch": 1.4780952380952381,
      "grad_norm": 2.9278367481439895,
      "learning_rate": 5.364857278193352e-06,
      "loss": 0.6826,
      "step": 776
    },
    {
      "epoch": 1.48,
      "grad_norm": 2.4330230002175353,
      "learning_rate": 5.3545971247162355e-06,
      "loss": 0.5997,
      "step": 777
    },
    {
      "epoch": 1.4819047619047618,
      "grad_norm": 3.3558247094866593,
      "learning_rate": 5.344335470321943e-06,
      "loss": 0.72,
      "step": 778
    },
    {
      "epoch": 1.4838095238095237,
      "grad_norm": 2.645281645076092,
      "learning_rate": 5.33407235844538e-06,
      "loss": 0.584,
      "step": 779
    },
    {
      "epoch": 1.4857142857142858,
      "grad_norm": 2.549284753692442,
      "learning_rate": 5.3238078325276255e-06,
      "loss": 0.636,
      "step": 780
    },
    {
      "epoch": 1.4876190476190476,
      "grad_norm": 2.8779634095757567,
      "learning_rate": 5.313541936015738e-06,
      "loss": 0.7058,
      "step": 781
    },
    {
      "epoch": 1.4895238095238095,
      "grad_norm": 2.8052343478617017,
      "learning_rate": 5.303274712362585e-06,
      "loss": 0.6137,
      "step": 782
    },
    {
      "epoch": 1.4914285714285715,
      "grad_norm": 3.1443738009443525,
      "learning_rate": 5.293006205026646e-06,
      "loss": 0.7201,
      "step": 783
    },
    {
      "epoch": 1.4933333333333334,
      "grad_norm": 3.489118791079987,
      "learning_rate": 5.2827364574718345e-06,
      "loss": 0.73,
      "step": 784
    },
    {
      "epoch": 1.4952380952380953,
      "grad_norm": 2.7852702479871,
      "learning_rate": 5.272465513167314e-06,
      "loss": 0.5916,
      "step": 785
    },
    {
      "epoch": 1.497142857142857,
      "grad_norm": 2.957697607444805,
      "learning_rate": 5.2621934155873146e-06,
      "loss": 0.5232,
      "step": 786
    },
    {
      "epoch": 1.499047619047619,
      "grad_norm": 3.8701519060947946,
      "learning_rate": 5.2519202082109486e-06,
      "loss": 0.6723,
      "step": 787
    },
    {
      "epoch": 1.500952380952381,
      "grad_norm": 3.4906011764517864,
      "learning_rate": 5.24164593452202e-06,
      "loss": 0.7324,
      "step": 788
    },
    {
      "epoch": 1.502857142857143,
      "grad_norm": 3.023484572517283,
      "learning_rate": 5.231370638008856e-06,
      "loss": 0.5537,
      "step": 789
    },
    {
      "epoch": 1.5047619047619047,
      "grad_norm": 3.099329266333818,
      "learning_rate": 5.221094362164103e-06,
      "loss": 0.6879,
      "step": 790
    },
    {
      "epoch": 1.5066666666666668,
      "grad_norm": 3.1294355431433334,
      "learning_rate": 5.210817150484562e-06,
      "loss": 0.6488,
      "step": 791
    },
    {
      "epoch": 1.5085714285714285,
      "grad_norm": 2.7660308401375917,
      "learning_rate": 5.200539046470986e-06,
      "loss": 0.6598,
      "step": 792
    },
    {
      "epoch": 1.5104761904761905,
      "grad_norm": 2.7023455096788016,
      "learning_rate": 5.190260093627912e-06,
      "loss": 0.6454,
      "step": 793
    },
    {
      "epoch": 1.5123809523809524,
      "grad_norm": 2.5508349609780994,
      "learning_rate": 5.179980335463467e-06,
      "loss": 0.682,
      "step": 794
    },
    {
      "epoch": 1.5142857142857142,
      "grad_norm": 3.2152014721060604,
      "learning_rate": 5.169699815489187e-06,
      "loss": 0.7798,
      "step": 795
    },
    {
      "epoch": 1.5161904761904763,
      "grad_norm": 3.1630300513674756,
      "learning_rate": 5.159418577219832e-06,
      "loss": 0.5927,
      "step": 796
    },
    {
      "epoch": 1.518095238095238,
      "grad_norm": 3.5036069767173257,
      "learning_rate": 5.149136664173205e-06,
      "loss": 0.8095,
      "step": 797
    },
    {
      "epoch": 1.52,
      "grad_norm": 2.718578606579851,
      "learning_rate": 5.138854119869962e-06,
      "loss": 0.6012,
      "step": 798
    },
    {
      "epoch": 1.5219047619047619,
      "grad_norm": 3.8660886936224443,
      "learning_rate": 5.128570987833433e-06,
      "loss": 0.9095,
      "step": 799
    },
    {
      "epoch": 1.5238095238095237,
      "grad_norm": 2.8226941473058877,
      "learning_rate": 5.118287311589435e-06,
      "loss": 0.5466,
      "step": 800
    },
    {
      "epoch": 1.5257142857142858,
      "grad_norm": 3.5164132340577336,
      "learning_rate": 5.108003134666085e-06,
      "loss": 0.6947,
      "step": 801
    },
    {
      "epoch": 1.5276190476190477,
      "grad_norm": 3.446538166929701,
      "learning_rate": 5.0977185005936245e-06,
      "loss": 0.6454,
      "step": 802
    },
    {
      "epoch": 1.5295238095238095,
      "grad_norm": 2.8804223530054833,
      "learning_rate": 5.08743345290423e-06,
      "loss": 0.6789,
      "step": 803
    },
    {
      "epoch": 1.5314285714285716,
      "grad_norm": 3.728491017598619,
      "learning_rate": 5.0771480351318245e-06,
      "loss": 0.6611,
      "step": 804
    },
    {
      "epoch": 1.5333333333333332,
      "grad_norm": 2.683746756487477,
      "learning_rate": 5.066862290811901e-06,
      "loss": 0.628,
      "step": 805
    },
    {
      "epoch": 1.5352380952380953,
      "grad_norm": 2.826122839600481,
      "learning_rate": 5.056576263481333e-06,
      "loss": 0.6233,
      "step": 806
    },
    {
      "epoch": 1.5371428571428571,
      "grad_norm": 2.4424899804917395,
      "learning_rate": 5.046289996678193e-06,
      "loss": 0.6067,
      "step": 807
    },
    {
      "epoch": 1.539047619047619,
      "grad_norm": 3.315304635544952,
      "learning_rate": 5.036003533941566e-06,
      "loss": 0.5374,
      "step": 808
    },
    {
      "epoch": 1.540952380952381,
      "grad_norm": 2.5365866522887788,
      "learning_rate": 5.025716918811365e-06,
      "loss": 0.6411,
      "step": 809
    },
    {
      "epoch": 1.5428571428571427,
      "grad_norm": 3.0745905352339853,
      "learning_rate": 5.015430194828153e-06,
      "loss": 0.6221,
      "step": 810
    },
    {
      "epoch": 1.5447619047619048,
      "grad_norm": 3.395566166882911,
      "learning_rate": 5.005143405532949e-06,
      "loss": 0.7,
      "step": 811
    },
    {
      "epoch": 1.5466666666666666,
      "grad_norm": 3.177105463980441,
      "learning_rate": 4.994856594467052e-06,
      "loss": 0.7141,
      "step": 812
    },
    {
      "epoch": 1.5485714285714285,
      "grad_norm": 2.869780556824362,
      "learning_rate": 4.984569805171848e-06,
      "loss": 0.6307,
      "step": 813
    },
    {
      "epoch": 1.5504761904761906,
      "grad_norm": 2.781756018163063,
      "learning_rate": 4.974283081188637e-06,
      "loss": 0.6423,
      "step": 814
    },
    {
      "epoch": 1.5523809523809524,
      "grad_norm": 2.6451418188447886,
      "learning_rate": 4.963996466058437e-06,
      "loss": 0.6308,
      "step": 815
    },
    {
      "epoch": 1.5542857142857143,
      "grad_norm": 2.600168166749201,
      "learning_rate": 4.9537100033218095e-06,
      "loss": 0.7036,
      "step": 816
    },
    {
      "epoch": 1.5561904761904763,
      "grad_norm": 3.7102416903278344,
      "learning_rate": 4.9434237365186684e-06,
      "loss": 0.7789,
      "step": 817
    },
    {
      "epoch": 1.558095238095238,
      "grad_norm": 2.7509058272754063,
      "learning_rate": 4.933137709188101e-06,
      "loss": 0.6555,
      "step": 818
    },
    {
      "epoch": 1.56,
      "grad_norm": 3.693036365858434,
      "learning_rate": 4.9228519648681755e-06,
      "loss": 0.7627,
      "step": 819
    },
    {
      "epoch": 1.561904761904762,
      "grad_norm": 2.509385243026165,
      "learning_rate": 4.912566547095771e-06,
      "loss": 0.6935,
      "step": 820
    },
    {
      "epoch": 1.5638095238095238,
      "grad_norm": 2.835926938956802,
      "learning_rate": 4.9022814994063755e-06,
      "loss": 0.7063,
      "step": 821
    },
    {
      "epoch": 1.5657142857142858,
      "grad_norm": 4.228622983137872,
      "learning_rate": 4.891996865333916e-06,
      "loss": 0.8022,
      "step": 822
    },
    {
      "epoch": 1.5676190476190475,
      "grad_norm": 2.375014267818986,
      "learning_rate": 4.881712688410567e-06,
      "loss": 0.7894,
      "step": 823
    },
    {
      "epoch": 1.5695238095238095,
      "grad_norm": 2.9474220653843735,
      "learning_rate": 4.8714290121665685e-06,
      "loss": 0.7133,
      "step": 824
    },
    {
      "epoch": 1.5714285714285714,
      "grad_norm": 2.7906613623942857,
      "learning_rate": 4.8611458801300385e-06,
      "loss": 0.6856,
      "step": 825
    },
    {
      "epoch": 1.5733333333333333,
      "grad_norm": 2.70156780088855,
      "learning_rate": 4.850863335826796e-06,
      "loss": 0.7329,
      "step": 826
    },
    {
      "epoch": 1.5752380952380953,
      "grad_norm": 2.9695588772186587,
      "learning_rate": 4.840581422780169e-06,
      "loss": 0.7742,
      "step": 827
    },
    {
      "epoch": 1.5771428571428572,
      "grad_norm": 2.719873968817656,
      "learning_rate": 4.830300184510815e-06,
      "loss": 0.6049,
      "step": 828
    },
    {
      "epoch": 1.579047619047619,
      "grad_norm": 2.812046412806861,
      "learning_rate": 4.820019664536535e-06,
      "loss": 0.6597,
      "step": 829
    },
    {
      "epoch": 1.580952380952381,
      "grad_norm": 2.1229560914014574,
      "learning_rate": 4.809739906372091e-06,
      "loss": 0.6903,
      "step": 830
    },
    {
      "epoch": 1.5828571428571427,
      "grad_norm": 2.4094540674992806,
      "learning_rate": 4.799460953529017e-06,
      "loss": 0.7547,
      "step": 831
    },
    {
      "epoch": 1.5847619047619048,
      "grad_norm": 2.6862956326258485,
      "learning_rate": 4.789182849515441e-06,
      "loss": 0.5746,
      "step": 832
    },
    {
      "epoch": 1.5866666666666667,
      "grad_norm": 2.9202442761476366,
      "learning_rate": 4.778905637835896e-06,
      "loss": 0.6531,
      "step": 833
    },
    {
      "epoch": 1.5885714285714285,
      "grad_norm": 3.1979912363852487,
      "learning_rate": 4.768629361991145e-06,
      "loss": 0.6923,
      "step": 834
    },
    {
      "epoch": 1.5904761904761906,
      "grad_norm": 3.7943376500290826,
      "learning_rate": 4.75835406547798e-06,
      "loss": 0.7664,
      "step": 835
    },
    {
      "epoch": 1.5923809523809522,
      "grad_norm": 3.3082326577429453,
      "learning_rate": 4.748079791789053e-06,
      "loss": 0.7764,
      "step": 836
    },
    {
      "epoch": 1.5942857142857143,
      "grad_norm": 3.048345940965857,
      "learning_rate": 4.737806584412687e-06,
      "loss": 0.6646,
      "step": 837
    },
    {
      "epoch": 1.5961904761904762,
      "grad_norm": 2.918550148394755,
      "learning_rate": 4.727534486832688e-06,
      "loss": 0.6275,
      "step": 838
    },
    {
      "epoch": 1.598095238095238,
      "grad_norm": 2.674098077658351,
      "learning_rate": 4.717263542528168e-06,
      "loss": 0.738,
      "step": 839
    },
    {
      "epoch": 1.6,
      "grad_norm": 2.9715429623860703,
      "learning_rate": 4.706993794973355e-06,
      "loss": 0.6091,
      "step": 840
    },
    {
      "epoch": 1.601904761904762,
      "grad_norm": 2.847112291191056,
      "learning_rate": 4.696725287637416e-06,
      "loss": 0.6481,
      "step": 841
    },
    {
      "epoch": 1.6038095238095238,
      "grad_norm": 2.42127343761098,
      "learning_rate": 4.686458063984262e-06,
      "loss": 0.6373,
      "step": 842
    },
    {
      "epoch": 1.6057142857142859,
      "grad_norm": 3.289201855822001,
      "learning_rate": 4.676192167472377e-06,
      "loss": 0.6502,
      "step": 843
    },
    {
      "epoch": 1.6076190476190475,
      "grad_norm": 3.3588879451203244,
      "learning_rate": 4.665927641554622e-06,
      "loss": 0.7436,
      "step": 844
    },
    {
      "epoch": 1.6095238095238096,
      "grad_norm": 3.402550196707435,
      "learning_rate": 4.65566452967806e-06,
      "loss": 0.5958,
      "step": 845
    },
    {
      "epoch": 1.6114285714285714,
      "grad_norm": 3.0956646090286575,
      "learning_rate": 4.645402875283766e-06,
      "loss": 0.6677,
      "step": 846
    },
    {
      "epoch": 1.6133333333333333,
      "grad_norm": 2.7562886994563383,
      "learning_rate": 4.635142721806648e-06,
      "loss": 0.7779,
      "step": 847
    },
    {
      "epoch": 1.6152380952380954,
      "grad_norm": 2.0483098437098137,
      "learning_rate": 4.624884112675264e-06,
      "loss": 0.7145,
      "step": 848
    },
    {
      "epoch": 1.617142857142857,
      "grad_norm": 3.175126134568553,
      "learning_rate": 4.61462709131163e-06,
      "loss": 0.7021,
      "step": 849
    },
    {
      "epoch": 1.619047619047619,
      "grad_norm": 3.104049491157547,
      "learning_rate": 4.604371701131042e-06,
      "loss": 0.6795,
      "step": 850
    },
    {
      "epoch": 1.620952380952381,
      "grad_norm": 3.0202728359535747,
      "learning_rate": 4.594117985541894e-06,
      "loss": 0.6878,
      "step": 851
    },
    {
      "epoch": 1.6228571428571428,
      "grad_norm": 4.1585216294967875,
      "learning_rate": 4.5838659879454885e-06,
      "loss": 0.6134,
      "step": 852
    },
    {
      "epoch": 1.6247619047619049,
      "grad_norm": 2.4888343626712275,
      "learning_rate": 4.5736157517358605e-06,
      "loss": 0.5989,
      "step": 853
    },
    {
      "epoch": 1.6266666666666667,
      "grad_norm": 2.71491706666814,
      "learning_rate": 4.563367320299587e-06,
      "loss": 0.6745,
      "step": 854
    },
    {
      "epoch": 1.6285714285714286,
      "grad_norm": 3.4916761720558505,
      "learning_rate": 4.553120737015606e-06,
      "loss": 0.6381,
      "step": 855
    },
    {
      "epoch": 1.6304761904761906,
      "grad_norm": 2.8324096551846454,
      "learning_rate": 4.542876045255034e-06,
      "loss": 0.6739,
      "step": 856
    },
    {
      "epoch": 1.6323809523809523,
      "grad_norm": 2.2204855147124754,
      "learning_rate": 4.5326332883809786e-06,
      "loss": 0.6818,
      "step": 857
    },
    {
      "epoch": 1.6342857142857143,
      "grad_norm": 2.9580703323692688,
      "learning_rate": 4.52239250974836e-06,
      "loss": 0.5558,
      "step": 858
    },
    {
      "epoch": 1.6361904761904762,
      "grad_norm": 2.5163725808744504,
      "learning_rate": 4.512153752703724e-06,
      "loss": 0.7532,
      "step": 859
    },
    {
      "epoch": 1.638095238095238,
      "grad_norm": 2.7245795479033994,
      "learning_rate": 4.50191706058506e-06,
      "loss": 0.5273,
      "step": 860
    },
    {
      "epoch": 1.6400000000000001,
      "grad_norm": 2.837670571626123,
      "learning_rate": 4.491682476721614e-06,
      "loss": 0.6603,
      "step": 861
    },
    {
      "epoch": 1.6419047619047618,
      "grad_norm": 2.9029427498443545,
      "learning_rate": 4.481450044433716e-06,
      "loss": 0.702,
      "step": 862
    },
    {
      "epoch": 1.6438095238095238,
      "grad_norm": 3.6327705075800547,
      "learning_rate": 4.471219807032582e-06,
      "loss": 0.7007,
      "step": 863
    },
    {
      "epoch": 1.6457142857142857,
      "grad_norm": 2.9520614396209863,
      "learning_rate": 4.4609918078201386e-06,
      "loss": 0.6278,
      "step": 864
    },
    {
      "epoch": 1.6476190476190475,
      "grad_norm": 2.9406888144314323,
      "learning_rate": 4.4507660900888405e-06,
      "loss": 0.7473,
      "step": 865
    },
    {
      "epoch": 1.6495238095238096,
      "grad_norm": 2.589423462875807,
      "learning_rate": 4.440542697121485e-06,
      "loss": 0.6206,
      "step": 866
    },
    {
      "epoch": 1.6514285714285715,
      "grad_norm": 3.7177152584023814,
      "learning_rate": 4.430321672191028e-06,
      "loss": 0.6792,
      "step": 867
    },
    {
      "epoch": 1.6533333333333333,
      "grad_norm": 3.5635519844294645,
      "learning_rate": 4.420103058560405e-06,
      "loss": 0.792,
      "step": 868
    },
    {
      "epoch": 1.6552380952380954,
      "grad_norm": 2.5931410583648846,
      "learning_rate": 4.409886899482341e-06,
      "loss": 0.6266,
      "step": 869
    },
    {
      "epoch": 1.657142857142857,
      "grad_norm": 3.2482541957748174,
      "learning_rate": 4.399673238199176e-06,
      "loss": 0.6542,
      "step": 870
    },
    {
      "epoch": 1.659047619047619,
      "grad_norm": 2.7941492631087548,
      "learning_rate": 4.389462117942673e-06,
      "loss": 0.5911,
      "step": 871
    },
    {
      "epoch": 1.660952380952381,
      "grad_norm": 3.198917735948789,
      "learning_rate": 4.379253581933844e-06,
      "loss": 0.7099,
      "step": 872
    },
    {
      "epoch": 1.6628571428571428,
      "grad_norm": 3.5179040731056395,
      "learning_rate": 4.369047673382759e-06,
      "loss": 0.6747,
      "step": 873
    },
    {
      "epoch": 1.6647619047619049,
      "grad_norm": 2.4504790181751965,
      "learning_rate": 4.358844435488367e-06,
      "loss": 0.5693,
      "step": 874
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 2.9500759019933214,
      "learning_rate": 4.3486439114383174e-06,
      "loss": 0.6927,
      "step": 875
    },
    {
      "epoch": 1.6685714285714286,
      "grad_norm": 2.935824994705467,
      "learning_rate": 4.338446144408766e-06,
      "loss": 0.6949,
      "step": 876
    },
    {
      "epoch": 1.6704761904761904,
      "grad_norm": 2.581849917285868,
      "learning_rate": 4.328251177564202e-06,
      "loss": 0.8297,
      "step": 877
    },
    {
      "epoch": 1.6723809523809523,
      "grad_norm": 3.649626741114842,
      "learning_rate": 4.318059054057263e-06,
      "loss": 0.7397,
      "step": 878
    },
    {
      "epoch": 1.6742857142857144,
      "grad_norm": 3.5574483603650466,
      "learning_rate": 4.307869817028549e-06,
      "loss": 0.6174,
      "step": 879
    },
    {
      "epoch": 1.6761904761904762,
      "grad_norm": 3.278158882599443,
      "learning_rate": 4.297683509606446e-06,
      "loss": 0.6851,
      "step": 880
    },
    {
      "epoch": 1.678095238095238,
      "grad_norm": 3.539550696262606,
      "learning_rate": 4.287500174906935e-06,
      "loss": 0.8195,
      "step": 881
    },
    {
      "epoch": 1.6800000000000002,
      "grad_norm": 3.3932565080563886,
      "learning_rate": 4.2773198560334186e-06,
      "loss": 0.7482,
      "step": 882
    },
    {
      "epoch": 1.6819047619047618,
      "grad_norm": 2.7348422029401833,
      "learning_rate": 4.267142596076532e-06,
      "loss": 0.4363,
      "step": 883
    },
    {
      "epoch": 1.6838095238095239,
      "grad_norm": 2.6544372954428352,
      "learning_rate": 4.256968438113962e-06,
      "loss": 0.7329,
      "step": 884
    },
    {
      "epoch": 1.6857142857142857,
      "grad_norm": 3.4679395965649276,
      "learning_rate": 4.246797425210268e-06,
      "loss": 0.7275,
      "step": 885
    },
    {
      "epoch": 1.6876190476190476,
      "grad_norm": 2.7341568318252447,
      "learning_rate": 4.236629600416695e-06,
      "loss": 0.6893,
      "step": 886
    },
    {
      "epoch": 1.6895238095238096,
      "grad_norm": 2.655452784384076,
      "learning_rate": 4.226465006770995e-06,
      "loss": 0.6781,
      "step": 887
    },
    {
      "epoch": 1.6914285714285713,
      "grad_norm": 3.166798197497907,
      "learning_rate": 4.216303687297242e-06,
      "loss": 0.6028,
      "step": 888
    },
    {
      "epoch": 1.6933333333333334,
      "grad_norm": 2.6579456682906164,
      "learning_rate": 4.206145685005653e-06,
      "loss": 0.6088,
      "step": 889
    },
    {
      "epoch": 1.6952380952380952,
      "grad_norm": 2.978773460797544,
      "learning_rate": 4.195991042892401e-06,
      "loss": 0.5404,
      "step": 890
    },
    {
      "epoch": 1.697142857142857,
      "grad_norm": 3.5451760661992346,
      "learning_rate": 4.185839803939442e-06,
      "loss": 0.7008,
      "step": 891
    },
    {
      "epoch": 1.6990476190476191,
      "grad_norm": 3.219380575339356,
      "learning_rate": 4.175692011114322e-06,
      "loss": 0.7501,
      "step": 892
    },
    {
      "epoch": 1.700952380952381,
      "grad_norm": 7.776458381128656,
      "learning_rate": 4.1655477073700015e-06,
      "loss": 0.7529,
      "step": 893
    },
    {
      "epoch": 1.7028571428571428,
      "grad_norm": 3.3465572345350076,
      "learning_rate": 4.1554069356446746e-06,
      "loss": 0.7325,
      "step": 894
    },
    {
      "epoch": 1.704761904761905,
      "grad_norm": 2.450860718244936,
      "learning_rate": 4.145269738861583e-06,
      "loss": 0.581,
      "step": 895
    },
    {
      "epoch": 1.7066666666666666,
      "grad_norm": 2.7750989329792257,
      "learning_rate": 4.1351361599288385e-06,
      "loss": 0.5982,
      "step": 896
    },
    {
      "epoch": 1.7085714285714286,
      "grad_norm": 2.6798522757105503,
      "learning_rate": 4.125006241739239e-06,
      "loss": 0.7034,
      "step": 897
    },
    {
      "epoch": 1.7104761904761905,
      "grad_norm": 2.973051275506407,
      "learning_rate": 4.114880027170086e-06,
      "loss": 0.7019,
      "step": 898
    },
    {
      "epoch": 1.7123809523809523,
      "grad_norm": 2.764933346748833,
      "learning_rate": 4.104757559083007e-06,
      "loss": 0.7546,
      "step": 899
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 3.026951164354996,
      "learning_rate": 4.094638880323768e-06,
      "loss": 0.5869,
      "step": 900
    },
    {
      "epoch": 1.716190476190476,
      "grad_norm": 3.0933296947341544,
      "learning_rate": 4.0845240337221005e-06,
      "loss": 0.639,
      "step": 901
    },
    {
      "epoch": 1.7180952380952381,
      "grad_norm": 3.186168970973745,
      "learning_rate": 4.0744130620915105e-06,
      "loss": 0.6596,
      "step": 902
    },
    {
      "epoch": 1.72,
      "grad_norm": 3.5668548195343077,
      "learning_rate": 4.064306008229107e-06,
      "loss": 0.7901,
      "step": 903
    },
    {
      "epoch": 1.7219047619047618,
      "grad_norm": 3.2842572628708475,
      "learning_rate": 4.054202914915414e-06,
      "loss": 0.6883,
      "step": 904
    },
    {
      "epoch": 1.723809523809524,
      "grad_norm": 4.67391971111033,
      "learning_rate": 4.044103824914189e-06,
      "loss": 0.7509,
      "step": 905
    },
    {
      "epoch": 1.7257142857142858,
      "grad_norm": 3.5070012137829973,
      "learning_rate": 4.03400878097225e-06,
      "loss": 0.7955,
      "step": 906
    },
    {
      "epoch": 1.7276190476190476,
      "grad_norm": 2.6875978886234226,
      "learning_rate": 4.023917825819283e-06,
      "loss": 0.6201,
      "step": 907
    },
    {
      "epoch": 1.7295238095238097,
      "grad_norm": 3.100726327946747,
      "learning_rate": 4.013831002167671e-06,
      "loss": 0.7032,
      "step": 908
    },
    {
      "epoch": 1.7314285714285713,
      "grad_norm": 3.631607759969553,
      "learning_rate": 4.00374835271231e-06,
      "loss": 0.7222,
      "step": 909
    },
    {
      "epoch": 1.7333333333333334,
      "grad_norm": 2.73818249684235,
      "learning_rate": 3.993669920130425e-06,
      "loss": 0.509,
      "step": 910
    },
    {
      "epoch": 1.7352380952380952,
      "grad_norm": 2.9059609345818407,
      "learning_rate": 3.983595747081394e-06,
      "loss": 0.6586,
      "step": 911
    },
    {
      "epoch": 1.737142857142857,
      "grad_norm": 3.360713371381218,
      "learning_rate": 3.9735258762065655e-06,
      "loss": 0.7986,
      "step": 912
    },
    {
      "epoch": 1.7390476190476192,
      "grad_norm": 3.774601486556661,
      "learning_rate": 3.963460350129076e-06,
      "loss": 0.7997,
      "step": 913
    },
    {
      "epoch": 1.7409523809523808,
      "grad_norm": 2.8127772449564747,
      "learning_rate": 3.953399211453674e-06,
      "loss": 0.6729,
      "step": 914
    },
    {
      "epoch": 1.7428571428571429,
      "grad_norm": 2.4749517512555514,
      "learning_rate": 3.943342502766536e-06,
      "loss": 0.6299,
      "step": 915
    },
    {
      "epoch": 1.7447619047619047,
      "grad_norm": 3.817257079363956,
      "learning_rate": 3.933290266635088e-06,
      "loss": 0.6733,
      "step": 916
    },
    {
      "epoch": 1.7466666666666666,
      "grad_norm": 2.5918427988265598,
      "learning_rate": 3.9232425456078245e-06,
      "loss": 0.896,
      "step": 917
    },
    {
      "epoch": 1.7485714285714287,
      "grad_norm": 2.4734952674166664,
      "learning_rate": 3.9131993822141305e-06,
      "loss": 0.6836,
      "step": 918
    },
    {
      "epoch": 1.7504761904761905,
      "grad_norm": 2.804256649930396,
      "learning_rate": 3.9031608189640956e-06,
      "loss": 0.5622,
      "step": 919
    },
    {
      "epoch": 1.7523809523809524,
      "grad_norm": 2.6027738683533346,
      "learning_rate": 3.893126898348343e-06,
      "loss": 0.806,
      "step": 920
    },
    {
      "epoch": 1.7542857142857144,
      "grad_norm": 3.235966019007431,
      "learning_rate": 3.8830976628378404e-06,
      "loss": 0.6003,
      "step": 921
    },
    {
      "epoch": 1.756190476190476,
      "grad_norm": 3.0276903817709138,
      "learning_rate": 3.873073154883729e-06,
      "loss": 0.5287,
      "step": 922
    },
    {
      "epoch": 1.7580952380952382,
      "grad_norm": 3.4626169524862935,
      "learning_rate": 3.863053416917136e-06,
      "loss": 0.539,
      "step": 923
    },
    {
      "epoch": 1.76,
      "grad_norm": 2.4458581267907276,
      "learning_rate": 3.853038491349e-06,
      "loss": 0.6652,
      "step": 924
    },
    {
      "epoch": 1.7619047619047619,
      "grad_norm": 2.404546447757066,
      "learning_rate": 3.84302842056989e-06,
      "loss": 0.6102,
      "step": 925
    },
    {
      "epoch": 1.763809523809524,
      "grad_norm": 3.764939456234988,
      "learning_rate": 3.8330232469498266e-06,
      "loss": 0.6831,
      "step": 926
    },
    {
      "epoch": 1.7657142857142856,
      "grad_norm": 2.1595889230545358,
      "learning_rate": 3.823023012838099e-06,
      "loss": 0.7069,
      "step": 927
    },
    {
      "epoch": 1.7676190476190476,
      "grad_norm": 3.7038396355014136,
      "learning_rate": 3.8130277605630928e-06,
      "loss": 0.6964,
      "step": 928
    },
    {
      "epoch": 1.7695238095238095,
      "grad_norm": 2.736883383483487,
      "learning_rate": 3.8030375324321037e-06,
      "loss": 0.6855,
      "step": 929
    },
    {
      "epoch": 1.7714285714285714,
      "grad_norm": 3.7513936330334143,
      "learning_rate": 3.793052370731163e-06,
      "loss": 0.7708,
      "step": 930
    },
    {
      "epoch": 1.7733333333333334,
      "grad_norm": 2.8367499337672832,
      "learning_rate": 3.7830723177248574e-06,
      "loss": 0.65,
      "step": 931
    },
    {
      "epoch": 1.7752380952380953,
      "grad_norm": 3.8323012751674588,
      "learning_rate": 3.7730974156561494e-06,
      "loss": 0.6391,
      "step": 932
    },
    {
      "epoch": 1.7771428571428571,
      "grad_norm": 2.7479238941827777,
      "learning_rate": 3.763127706746198e-06,
      "loss": 0.6623,
      "step": 933
    },
    {
      "epoch": 1.7790476190476192,
      "grad_norm": 3.1677388425003787,
      "learning_rate": 3.7531632331941826e-06,
      "loss": 0.7287,
      "step": 934
    },
    {
      "epoch": 1.7809523809523808,
      "grad_norm": 3.3814399371999277,
      "learning_rate": 3.74320403717712e-06,
      "loss": 0.6708,
      "step": 935
    },
    {
      "epoch": 1.782857142857143,
      "grad_norm": 3.596369055589814,
      "learning_rate": 3.7332501608496917e-06,
      "loss": 0.6385,
      "step": 936
    },
    {
      "epoch": 1.7847619047619048,
      "grad_norm": 2.6366411572094997,
      "learning_rate": 3.7233016463440612e-06,
      "loss": 0.6257,
      "step": 937
    },
    {
      "epoch": 1.7866666666666666,
      "grad_norm": 2.836274956577337,
      "learning_rate": 3.7133585357696954e-06,
      "loss": 0.5953,
      "step": 938
    },
    {
      "epoch": 1.7885714285714287,
      "grad_norm": 3.007747276794172,
      "learning_rate": 3.70342087121319e-06,
      "loss": 0.8086,
      "step": 939
    },
    {
      "epoch": 1.7904761904761903,
      "grad_norm": 4.314502666779546,
      "learning_rate": 3.693488694738089e-06,
      "loss": 0.8031,
      "step": 940
    },
    {
      "epoch": 1.7923809523809524,
      "grad_norm": 3.165669381401534,
      "learning_rate": 3.683562048384703e-06,
      "loss": 0.6624,
      "step": 941
    },
    {
      "epoch": 1.7942857142857143,
      "grad_norm": 3.283653265862736,
      "learning_rate": 3.67364097416994e-06,
      "loss": 0.6585,
      "step": 942
    },
    {
      "epoch": 1.7961904761904761,
      "grad_norm": 3.196551561721693,
      "learning_rate": 3.6637255140871196e-06,
      "loss": 0.8229,
      "step": 943
    },
    {
      "epoch": 1.7980952380952382,
      "grad_norm": 3.7514672905574966,
      "learning_rate": 3.653815710105799e-06,
      "loss": 0.5775,
      "step": 944
    },
    {
      "epoch": 1.8,
      "grad_norm": 3.407540122692509,
      "learning_rate": 3.643911604171596e-06,
      "loss": 0.6983,
      "step": 945
    },
    {
      "epoch": 1.801904761904762,
      "grad_norm": 3.2162916453910912,
      "learning_rate": 3.6340132382060084e-06,
      "loss": 0.824,
      "step": 946
    },
    {
      "epoch": 1.803809523809524,
      "grad_norm": 3.2014625669277588,
      "learning_rate": 3.6241206541062378e-06,
      "loss": 0.6648,
      "step": 947
    },
    {
      "epoch": 1.8057142857142856,
      "grad_norm": 4.000883770798457,
      "learning_rate": 3.6142338937450143e-06,
      "loss": 0.6828,
      "step": 948
    },
    {
      "epoch": 1.8076190476190477,
      "grad_norm": 3.0455307406818206,
      "learning_rate": 3.604352998970416e-06,
      "loss": 0.8198,
      "step": 949
    },
    {
      "epoch": 1.8095238095238095,
      "grad_norm": 2.7766421105119043,
      "learning_rate": 3.5944780116056955e-06,
      "loss": 0.6788,
      "step": 950
    },
    {
      "epoch": 1.8114285714285714,
      "grad_norm": 2.1483977793142843,
      "learning_rate": 3.5846089734490983e-06,
      "loss": 0.6736,
      "step": 951
    },
    {
      "epoch": 1.8133333333333335,
      "grad_norm": 2.9116355673970644,
      "learning_rate": 3.5747459262736905e-06,
      "loss": 0.6319,
      "step": 952
    },
    {
      "epoch": 1.815238095238095,
      "grad_norm": 2.292376967765798,
      "learning_rate": 3.564888911827179e-06,
      "loss": 0.7351,
      "step": 953
    },
    {
      "epoch": 1.8171428571428572,
      "grad_norm": 3.7201113111301654,
      "learning_rate": 3.5550379718317364e-06,
      "loss": 0.6455,
      "step": 954
    },
    {
      "epoch": 1.819047619047619,
      "grad_norm": 3.143821334076954,
      "learning_rate": 3.5451931479838226e-06,
      "loss": 0.6841,
      "step": 955
    },
    {
      "epoch": 1.8209523809523809,
      "grad_norm": 2.8356800264594373,
      "learning_rate": 3.5353544819540107e-06,
      "loss": 0.6358,
      "step": 956
    },
    {
      "epoch": 1.822857142857143,
      "grad_norm": 2.455253693710853,
      "learning_rate": 3.5255220153868093e-06,
      "loss": 0.6615,
      "step": 957
    },
    {
      "epoch": 1.8247619047619048,
      "grad_norm": 3.0476554863606578,
      "learning_rate": 3.515695789900484e-06,
      "loss": 0.7617,
      "step": 958
    },
    {
      "epoch": 1.8266666666666667,
      "grad_norm": 3.395291262485949,
      "learning_rate": 3.505875847086887e-06,
      "loss": 0.6318,
      "step": 959
    },
    {
      "epoch": 1.8285714285714287,
      "grad_norm": 2.661352086088814,
      "learning_rate": 3.4960622285112768e-06,
      "loss": 0.7293,
      "step": 960
    },
    {
      "epoch": 1.8304761904761904,
      "grad_norm": 2.7634313031316866,
      "learning_rate": 3.48625497571214e-06,
      "loss": 0.683,
      "step": 961
    },
    {
      "epoch": 1.8323809523809524,
      "grad_norm": 3.4979611469618046,
      "learning_rate": 3.4764541302010224e-06,
      "loss": 0.6342,
      "step": 962
    },
    {
      "epoch": 1.8342857142857143,
      "grad_norm": 3.8328189313722234,
      "learning_rate": 3.4666597334623463e-06,
      "loss": 0.8184,
      "step": 963
    },
    {
      "epoch": 1.8361904761904762,
      "grad_norm": 3.9301064995000257,
      "learning_rate": 3.4568718269532407e-06,
      "loss": 0.6645,
      "step": 964
    },
    {
      "epoch": 1.8380952380952382,
      "grad_norm": 3.2556643627852475,
      "learning_rate": 3.447090452103361e-06,
      "loss": 0.6603,
      "step": 965
    },
    {
      "epoch": 1.8399999999999999,
      "grad_norm": 3.0439338169853785,
      "learning_rate": 3.4373156503147175e-06,
      "loss": 0.8471,
      "step": 966
    },
    {
      "epoch": 1.841904761904762,
      "grad_norm": 2.551604789353069,
      "learning_rate": 3.4275474629614974e-06,
      "loss": 0.7113,
      "step": 967
    },
    {
      "epoch": 1.8438095238095238,
      "grad_norm": 2.8233013118021035,
      "learning_rate": 3.417785931389891e-06,
      "loss": 0.8102,
      "step": 968
    },
    {
      "epoch": 1.8457142857142856,
      "grad_norm": 3.4451629055832087,
      "learning_rate": 3.408031096917915e-06,
      "loss": 0.8296,
      "step": 969
    },
    {
      "epoch": 1.8476190476190477,
      "grad_norm": 2.590423919061846,
      "learning_rate": 3.3982830008352426e-06,
      "loss": 0.7947,
      "step": 970
    },
    {
      "epoch": 1.8495238095238096,
      "grad_norm": 3.2163996343351213,
      "learning_rate": 3.3885416844030228e-06,
      "loss": 0.7455,
      "step": 971
    },
    {
      "epoch": 1.8514285714285714,
      "grad_norm": 3.7606931096655893,
      "learning_rate": 3.3788071888537046e-06,
      "loss": 0.7031,
      "step": 972
    },
    {
      "epoch": 1.8533333333333335,
      "grad_norm": 2.737429417892746,
      "learning_rate": 3.369079555390874e-06,
      "loss": 0.6771,
      "step": 973
    },
    {
      "epoch": 1.8552380952380951,
      "grad_norm": 2.9544444010544204,
      "learning_rate": 3.3593588251890654e-06,
      "loss": 0.7257,
      "step": 974
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 3.705886648160575,
      "learning_rate": 3.3496450393935964e-06,
      "loss": 0.6716,
      "step": 975
    },
    {
      "epoch": 1.859047619047619,
      "grad_norm": 2.66903969876622,
      "learning_rate": 3.3399382391203883e-06,
      "loss": 0.6443,
      "step": 976
    },
    {
      "epoch": 1.860952380952381,
      "grad_norm": 2.44918155965145,
      "learning_rate": 3.330238465455796e-06,
      "loss": 0.6651,
      "step": 977
    },
    {
      "epoch": 1.862857142857143,
      "grad_norm": 3.7074574220776153,
      "learning_rate": 3.3205457594564337e-06,
      "loss": 0.7179,
      "step": 978
    },
    {
      "epoch": 1.8647619047619046,
      "grad_norm": 3.46726522210293,
      "learning_rate": 3.310860162148998e-06,
      "loss": 0.6445,
      "step": 979
    },
    {
      "epoch": 1.8666666666666667,
      "grad_norm": 3.2617819343089125,
      "learning_rate": 3.301181714530097e-06,
      "loss": 0.6553,
      "step": 980
    },
    {
      "epoch": 1.8685714285714285,
      "grad_norm": 3.040505154545407,
      "learning_rate": 3.2915104575660762e-06,
      "loss": 0.6299,
      "step": 981
    },
    {
      "epoch": 1.8704761904761904,
      "grad_norm": 2.3678634776746805,
      "learning_rate": 3.281846432192844e-06,
      "loss": 0.6782,
      "step": 982
    },
    {
      "epoch": 1.8723809523809525,
      "grad_norm": 3.9216230243119417,
      "learning_rate": 3.272189679315701e-06,
      "loss": 0.5438,
      "step": 983
    },
    {
      "epoch": 1.8742857142857143,
      "grad_norm": 3.3827666136105083,
      "learning_rate": 3.2625402398091656e-06,
      "loss": 0.6743,
      "step": 984
    },
    {
      "epoch": 1.8761904761904762,
      "grad_norm": 2.66060656768813,
      "learning_rate": 3.252898154516797e-06,
      "loss": 0.7815,
      "step": 985
    },
    {
      "epoch": 1.878095238095238,
      "grad_norm": 3.614194893411707,
      "learning_rate": 3.2432634642510262e-06,
      "loss": 0.6835,
      "step": 986
    },
    {
      "epoch": 1.88,
      "grad_norm": 3.171225832381452,
      "learning_rate": 3.233636209792992e-06,
      "loss": 0.7357,
      "step": 987
    },
    {
      "epoch": 1.881904761904762,
      "grad_norm": 2.632051058171287,
      "learning_rate": 3.224016431892347e-06,
      "loss": 0.6752,
      "step": 988
    },
    {
      "epoch": 1.8838095238095238,
      "grad_norm": 4.15848478612645,
      "learning_rate": 3.214404171267106e-06,
      "loss": 0.7246,
      "step": 989
    },
    {
      "epoch": 1.8857142857142857,
      "grad_norm": 2.5402689839432124,
      "learning_rate": 3.20479946860346e-06,
      "loss": 0.6542,
      "step": 990
    },
    {
      "epoch": 1.8876190476190478,
      "grad_norm": 2.8793422010297243,
      "learning_rate": 3.1952023645556126e-06,
      "loss": 0.5687,
      "step": 991
    },
    {
      "epoch": 1.8895238095238094,
      "grad_norm": 2.485960071524638,
      "learning_rate": 3.1856128997456015e-06,
      "loss": 0.6152,
      "step": 992
    },
    {
      "epoch": 1.8914285714285715,
      "grad_norm": 2.675451120665098,
      "learning_rate": 3.176031114763133e-06,
      "loss": 0.779,
      "step": 993
    },
    {
      "epoch": 1.8933333333333333,
      "grad_norm": 2.708103339026693,
      "learning_rate": 3.1664570501654026e-06,
      "loss": 0.6607,
      "step": 994
    },
    {
      "epoch": 1.8952380952380952,
      "grad_norm": 4.097481901284391,
      "learning_rate": 3.1568907464769307e-06,
      "loss": 0.6029,
      "step": 995
    },
    {
      "epoch": 1.8971428571428572,
      "grad_norm": 3.1323265468145745,
      "learning_rate": 3.1473322441893885e-06,
      "loss": 0.5538,
      "step": 996
    },
    {
      "epoch": 1.899047619047619,
      "grad_norm": 2.433413823618976,
      "learning_rate": 3.13778158376142e-06,
      "loss": 0.6347,
      "step": 997
    },
    {
      "epoch": 1.900952380952381,
      "grad_norm": 2.4950458606931467,
      "learning_rate": 3.128238805618483e-06,
      "loss": 0.6297,
      "step": 998
    },
    {
      "epoch": 1.9028571428571428,
      "grad_norm": 3.490653170933105,
      "learning_rate": 3.118703950152667e-06,
      "loss": 0.8749,
      "step": 999
    },
    {
      "epoch": 1.9047619047619047,
      "grad_norm": 2.5501564144705413,
      "learning_rate": 3.1091770577225343e-06,
      "loss": 0.5847,
      "step": 1000
    },
    {
      "epoch": 1.9066666666666667,
      "grad_norm": 3.678700551498614,
      "learning_rate": 3.0996581686529337e-06,
      "loss": 0.8202,
      "step": 1001
    },
    {
      "epoch": 1.9085714285714286,
      "grad_norm": 3.0464766067692466,
      "learning_rate": 3.09014732323484e-06,
      "loss": 0.815,
      "step": 1002
    },
    {
      "epoch": 1.9104761904761904,
      "grad_norm": 2.801546039768304,
      "learning_rate": 3.0806445617251834e-06,
      "loss": 0.6446,
      "step": 1003
    },
    {
      "epoch": 1.9123809523809525,
      "grad_norm": 3.45647373721546,
      "learning_rate": 3.0711499243466757e-06,
      "loss": 0.7771,
      "step": 1004
    },
    {
      "epoch": 1.9142857142857141,
      "grad_norm": 2.699534712808293,
      "learning_rate": 3.061663451287641e-06,
      "loss": 0.6534,
      "step": 1005
    },
    {
      "epoch": 1.9161904761904762,
      "grad_norm": 2.7654829962859617,
      "learning_rate": 3.052185182701847e-06,
      "loss": 0.5684,
      "step": 1006
    },
    {
      "epoch": 1.918095238095238,
      "grad_norm": 3.421459596071975,
      "learning_rate": 3.042715158708332e-06,
      "loss": 0.6075,
      "step": 1007
    },
    {
      "epoch": 1.92,
      "grad_norm": 3.5066328280690393,
      "learning_rate": 3.03325341939124e-06,
      "loss": 0.5306,
      "step": 1008
    },
    {
      "epoch": 1.921904761904762,
      "grad_norm": 3.1911134834340853,
      "learning_rate": 3.0238000047996476e-06,
      "loss": 0.5241,
      "step": 1009
    },
    {
      "epoch": 1.9238095238095239,
      "grad_norm": 2.501138139258464,
      "learning_rate": 3.0143549549473904e-06,
      "loss": 0.7524,
      "step": 1010
    },
    {
      "epoch": 1.9257142857142857,
      "grad_norm": 3.647850545631917,
      "learning_rate": 3.0049183098129048e-06,
      "loss": 0.658,
      "step": 1011
    },
    {
      "epoch": 1.9276190476190476,
      "grad_norm": 2.744037371013075,
      "learning_rate": 2.9954901093390486e-06,
      "loss": 0.6452,
      "step": 1012
    },
    {
      "epoch": 1.9295238095238094,
      "grad_norm": 2.8659802053956613,
      "learning_rate": 2.986070393432934e-06,
      "loss": 0.7747,
      "step": 1013
    },
    {
      "epoch": 1.9314285714285715,
      "grad_norm": 2.8519051436590312,
      "learning_rate": 2.9766592019657666e-06,
      "loss": 0.6887,
      "step": 1014
    },
    {
      "epoch": 1.9333333333333333,
      "grad_norm": 2.7105018106037138,
      "learning_rate": 2.967256574772664e-06,
      "loss": 0.746,
      "step": 1015
    },
    {
      "epoch": 1.9352380952380952,
      "grad_norm": 3.5703490064192396,
      "learning_rate": 2.957862551652496e-06,
      "loss": 0.7283,
      "step": 1016
    },
    {
      "epoch": 1.9371428571428573,
      "grad_norm": 3.279438652392567,
      "learning_rate": 2.948477172367713e-06,
      "loss": 0.7389,
      "step": 1017
    },
    {
      "epoch": 1.939047619047619,
      "grad_norm": 2.541490937895215,
      "learning_rate": 2.9391004766441787e-06,
      "loss": 0.6688,
      "step": 1018
    },
    {
      "epoch": 1.940952380952381,
      "grad_norm": 3.616846672630028,
      "learning_rate": 2.9297325041710014e-06,
      "loss": 0.7739,
      "step": 1019
    },
    {
      "epoch": 1.9428571428571428,
      "grad_norm": 2.995718018703318,
      "learning_rate": 2.920373294600366e-06,
      "loss": 0.6873,
      "step": 1020
    },
    {
      "epoch": 1.9447619047619047,
      "grad_norm": 3.611165661455842,
      "learning_rate": 2.9110228875473674e-06,
      "loss": 0.6595,
      "step": 1021
    },
    {
      "epoch": 1.9466666666666668,
      "grad_norm": 2.4968443513012404,
      "learning_rate": 2.9016813225898376e-06,
      "loss": 0.7015,
      "step": 1022
    },
    {
      "epoch": 1.9485714285714286,
      "grad_norm": 3.2779763545269454,
      "learning_rate": 2.89234863926819e-06,
      "loss": 0.764,
      "step": 1023
    },
    {
      "epoch": 1.9504761904761905,
      "grad_norm": 2.5324476225016426,
      "learning_rate": 2.8830248770852353e-06,
      "loss": 0.5953,
      "step": 1024
    },
    {
      "epoch": 1.9523809523809523,
      "grad_norm": 2.7416031392621445,
      "learning_rate": 2.8737100755060322e-06,
      "loss": 0.8657,
      "step": 1025
    },
    {
      "epoch": 1.9542857142857142,
      "grad_norm": 2.4743787009536593,
      "learning_rate": 2.8644042739577027e-06,
      "loss": 0.662,
      "step": 1026
    },
    {
      "epoch": 1.9561904761904763,
      "grad_norm": 2.3127230950622866,
      "learning_rate": 2.8551075118292815e-06,
      "loss": 0.4841,
      "step": 1027
    },
    {
      "epoch": 1.958095238095238,
      "grad_norm": 2.8095548290062053,
      "learning_rate": 2.84581982847154e-06,
      "loss": 0.6555,
      "step": 1028
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.955659645310816,
      "learning_rate": 2.836541263196817e-06,
      "loss": 0.6848,
      "step": 1029
    },
    {
      "epoch": 1.961904761904762,
      "grad_norm": 3.2003943913507573,
      "learning_rate": 2.8272718552788632e-06,
      "loss": 0.5073,
      "step": 1030
    },
    {
      "epoch": 1.9638095238095237,
      "grad_norm": 2.9190579627459377,
      "learning_rate": 2.818011643952662e-06,
      "loss": 0.7645,
      "step": 1031
    },
    {
      "epoch": 1.9657142857142857,
      "grad_norm": 3.519051828501603,
      "learning_rate": 2.808760668414278e-06,
      "loss": 0.7284,
      "step": 1032
    },
    {
      "epoch": 1.9676190476190476,
      "grad_norm": 3.5969210561054075,
      "learning_rate": 2.7995189678206745e-06,
      "loss": 0.6626,
      "step": 1033
    },
    {
      "epoch": 1.9695238095238095,
      "grad_norm": 2.560460669410691,
      "learning_rate": 2.7902865812895643e-06,
      "loss": 0.6835,
      "step": 1034
    },
    {
      "epoch": 1.9714285714285715,
      "grad_norm": 3.279912639171773,
      "learning_rate": 2.7810635478992288e-06,
      "loss": 0.559,
      "step": 1035
    },
    {
      "epoch": 1.9733333333333334,
      "grad_norm": 3.12364393300968,
      "learning_rate": 2.7718499066883674e-06,
      "loss": 0.6219,
      "step": 1036
    },
    {
      "epoch": 1.9752380952380952,
      "grad_norm": 2.522241057158423,
      "learning_rate": 2.762645696655918e-06,
      "loss": 0.6325,
      "step": 1037
    },
    {
      "epoch": 1.977142857142857,
      "grad_norm": 3.139250680280131,
      "learning_rate": 2.753450956760904e-06,
      "loss": 0.5558,
      "step": 1038
    },
    {
      "epoch": 1.979047619047619,
      "grad_norm": 3.6694684757927476,
      "learning_rate": 2.74426572592226e-06,
      "loss": 0.737,
      "step": 1039
    },
    {
      "epoch": 1.980952380952381,
      "grad_norm": 2.344084256099442,
      "learning_rate": 2.7350900430186765e-06,
      "loss": 0.5458,
      "step": 1040
    },
    {
      "epoch": 1.9828571428571429,
      "grad_norm": 2.367817415842074,
      "learning_rate": 2.7259239468884226e-06,
      "loss": 0.6367,
      "step": 1041
    },
    {
      "epoch": 1.9847619047619047,
      "grad_norm": 2.4794917294433096,
      "learning_rate": 2.716767476329196e-06,
      "loss": 0.8844,
      "step": 1042
    },
    {
      "epoch": 1.9866666666666668,
      "grad_norm": 3.35916805510129,
      "learning_rate": 2.7076206700979513e-06,
      "loss": 0.7125,
      "step": 1043
    },
    {
      "epoch": 1.9885714285714284,
      "grad_norm": 2.8537894720704355,
      "learning_rate": 2.6984835669107306e-06,
      "loss": 0.6735,
      "step": 1044
    },
    {
      "epoch": 1.9904761904761905,
      "grad_norm": 3.0004375782493473,
      "learning_rate": 2.6893562054425128e-06,
      "loss": 0.802,
      "step": 1045
    },
    {
      "epoch": 1.9923809523809524,
      "grad_norm": 3.5562355351148645,
      "learning_rate": 2.680238624327035e-06,
      "loss": 0.7207,
      "step": 1046
    },
    {
      "epoch": 1.9942857142857142,
      "grad_norm": 2.883149790302853,
      "learning_rate": 2.671130862156646e-06,
      "loss": 0.6391,
      "step": 1047
    },
    {
      "epoch": 1.9961904761904763,
      "grad_norm": 2.996491262752851,
      "learning_rate": 2.662032957482124e-06,
      "loss": 0.7328,
      "step": 1048
    },
    {
      "epoch": 1.9980952380952381,
      "grad_norm": 3.081309281423561,
      "learning_rate": 2.6529449488125312e-06,
      "loss": 0.8187,
      "step": 1049
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.851740093939044,
      "learning_rate": 2.6438668746150354e-06,
      "loss": 0.5597,
      "step": 1050
    },
    {
      "epoch": 2.001904761904762,
      "grad_norm": 3.637068489397177,
      "learning_rate": 2.6347987733147607e-06,
      "loss": 0.385,
      "step": 1051
    },
    {
      "epoch": 2.0038095238095237,
      "grad_norm": 3.2375252146572073,
      "learning_rate": 2.625740683294613e-06,
      "loss": 0.3336,
      "step": 1052
    },
    {
      "epoch": 2.005714285714286,
      "grad_norm": 3.4134579978259496,
      "learning_rate": 2.616692642895129e-06,
      "loss": 0.4053,
      "step": 1053
    },
    {
      "epoch": 2.0076190476190474,
      "grad_norm": 3.199313958815869,
      "learning_rate": 2.6076546904143005e-06,
      "loss": 0.3686,
      "step": 1054
    },
    {
      "epoch": 2.0095238095238095,
      "grad_norm": 2.844795300700537,
      "learning_rate": 2.5986268641074263e-06,
      "loss": 0.5282,
      "step": 1055
    },
    {
      "epoch": 2.0114285714285716,
      "grad_norm": 4.06889303615382,
      "learning_rate": 2.589609202186943e-06,
      "loss": 0.4347,
      "step": 1056
    },
    {
      "epoch": 2.013333333333333,
      "grad_norm": 4.07097786992572,
      "learning_rate": 2.5806017428222586e-06,
      "loss": 0.4167,
      "step": 1057
    },
    {
      "epoch": 2.0152380952380953,
      "grad_norm": 5.0547059965679635,
      "learning_rate": 2.571604524139604e-06,
      "loss": 0.439,
      "step": 1058
    },
    {
      "epoch": 2.0171428571428573,
      "grad_norm": 3.892356662771511,
      "learning_rate": 2.562617584221857e-06,
      "loss": 0.3594,
      "step": 1059
    },
    {
      "epoch": 2.019047619047619,
      "grad_norm": 4.006910134325121,
      "learning_rate": 2.5536409611083938e-06,
      "loss": 0.4314,
      "step": 1060
    },
    {
      "epoch": 2.020952380952381,
      "grad_norm": 3.2198974596850913,
      "learning_rate": 2.5446746927949168e-06,
      "loss": 0.3748,
      "step": 1061
    },
    {
      "epoch": 2.0228571428571427,
      "grad_norm": 3.0382467482233215,
      "learning_rate": 2.5357188172333047e-06,
      "loss": 0.3801,
      "step": 1062
    },
    {
      "epoch": 2.0247619047619048,
      "grad_norm": 3.0279668403638955,
      "learning_rate": 2.5267733723314415e-06,
      "loss": 0.4278,
      "step": 1063
    },
    {
      "epoch": 2.026666666666667,
      "grad_norm": 3.666895705374802,
      "learning_rate": 2.517838395953066e-06,
      "loss": 0.3257,
      "step": 1064
    },
    {
      "epoch": 2.0285714285714285,
      "grad_norm": 2.882130718796121,
      "learning_rate": 2.5089139259175995e-06,
      "loss": 0.3724,
      "step": 1065
    },
    {
      "epoch": 2.0304761904761905,
      "grad_norm": 3.885836296926442,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.3703,
      "step": 1066
    },
    {
      "epoch": 2.032380952380952,
      "grad_norm": 3.3787482844896823,
      "learning_rate": 2.4910966559305915e-06,
      "loss": 0.4495,
      "step": 1067
    },
    {
      "epoch": 2.0342857142857143,
      "grad_norm": 2.6043378408223106,
      "learning_rate": 2.4822039313949085e-06,
      "loss": 0.4384,
      "step": 1068
    },
    {
      "epoch": 2.0361904761904763,
      "grad_norm": 3.4165299550075163,
      "learning_rate": 2.4733218640335326e-06,
      "loss": 0.461,
      "step": 1069
    },
    {
      "epoch": 2.038095238095238,
      "grad_norm": 3.7409614105375035,
      "learning_rate": 2.4644504914419415e-06,
      "loss": 0.4579,
      "step": 1070
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.7901050347361758,
      "learning_rate": 2.4555898511703442e-06,
      "loss": 0.3874,
      "step": 1071
    },
    {
      "epoch": 2.041904761904762,
      "grad_norm": 3.551625845594265,
      "learning_rate": 2.446739980723516e-06,
      "loss": 0.3443,
      "step": 1072
    },
    {
      "epoch": 2.0438095238095237,
      "grad_norm": 3.634716148388194,
      "learning_rate": 2.437900917560656e-06,
      "loss": 0.4921,
      "step": 1073
    },
    {
      "epoch": 2.045714285714286,
      "grad_norm": 3.9765704091420613,
      "learning_rate": 2.4290726990952102e-06,
      "loss": 0.4454,
      "step": 1074
    },
    {
      "epoch": 2.0476190476190474,
      "grad_norm": 4.402686833130976,
      "learning_rate": 2.4202553626947284e-06,
      "loss": 0.4732,
      "step": 1075
    },
    {
      "epoch": 2.0495238095238095,
      "grad_norm": 4.361861112667196,
      "learning_rate": 2.4114489456806946e-06,
      "loss": 0.4168,
      "step": 1076
    },
    {
      "epoch": 2.0514285714285716,
      "grad_norm": 2.929300835921894,
      "learning_rate": 2.4026534853283778e-06,
      "loss": 0.3405,
      "step": 1077
    },
    {
      "epoch": 2.0533333333333332,
      "grad_norm": 2.991802014109771,
      "learning_rate": 2.3938690188666664e-06,
      "loss": 0.344,
      "step": 1078
    },
    {
      "epoch": 2.0552380952380953,
      "grad_norm": 2.6690057769651254,
      "learning_rate": 2.3850955834779193e-06,
      "loss": 0.3568,
      "step": 1079
    },
    {
      "epoch": 2.057142857142857,
      "grad_norm": 2.578287566633522,
      "learning_rate": 2.376333216297798e-06,
      "loss": 0.3228,
      "step": 1080
    },
    {
      "epoch": 2.059047619047619,
      "grad_norm": 3.8783638362187647,
      "learning_rate": 2.367581954415122e-06,
      "loss": 0.3392,
      "step": 1081
    },
    {
      "epoch": 2.060952380952381,
      "grad_norm": 3.6678426369720216,
      "learning_rate": 2.3588418348716987e-06,
      "loss": 0.4087,
      "step": 1082
    },
    {
      "epoch": 2.0628571428571427,
      "grad_norm": 2.946493462555248,
      "learning_rate": 2.3501128946621715e-06,
      "loss": 0.466,
      "step": 1083
    },
    {
      "epoch": 2.064761904761905,
      "grad_norm": 2.6534360876858525,
      "learning_rate": 2.3413951707338767e-06,
      "loss": 0.3933,
      "step": 1084
    },
    {
      "epoch": 2.066666666666667,
      "grad_norm": 2.834606080607265,
      "learning_rate": 2.3326886999866603e-06,
      "loss": 0.347,
      "step": 1085
    },
    {
      "epoch": 2.0685714285714285,
      "grad_norm": 2.686933628880203,
      "learning_rate": 2.323993519272748e-06,
      "loss": 0.3219,
      "step": 1086
    },
    {
      "epoch": 2.0704761904761906,
      "grad_norm": 3.4857052478354826,
      "learning_rate": 2.315309665396568e-06,
      "loss": 0.4491,
      "step": 1087
    },
    {
      "epoch": 2.072380952380952,
      "grad_norm": 5.058133692422829,
      "learning_rate": 2.3066371751146132e-06,
      "loss": 0.3538,
      "step": 1088
    },
    {
      "epoch": 2.0742857142857143,
      "grad_norm": 2.4866568973729075,
      "learning_rate": 2.297976085135271e-06,
      "loss": 0.2771,
      "step": 1089
    },
    {
      "epoch": 2.0761904761904764,
      "grad_norm": 4.351501615409117,
      "learning_rate": 2.2893264321186803e-06,
      "loss": 0.4655,
      "step": 1090
    },
    {
      "epoch": 2.078095238095238,
      "grad_norm": 4.257936336090124,
      "learning_rate": 2.2806882526765635e-06,
      "loss": 0.6581,
      "step": 1091
    },
    {
      "epoch": 2.08,
      "grad_norm": 3.0426623147432075,
      "learning_rate": 2.2720615833720862e-06,
      "loss": 0.3481,
      "step": 1092
    },
    {
      "epoch": 2.0819047619047617,
      "grad_norm": 2.693294807397359,
      "learning_rate": 2.2634464607196855e-06,
      "loss": 0.3785,
      "step": 1093
    },
    {
      "epoch": 2.083809523809524,
      "grad_norm": 2.6583578382142274,
      "learning_rate": 2.2548429211849337e-06,
      "loss": 0.2755,
      "step": 1094
    },
    {
      "epoch": 2.085714285714286,
      "grad_norm": 2.9621505212145784,
      "learning_rate": 2.246251001184369e-06,
      "loss": 0.3734,
      "step": 1095
    },
    {
      "epoch": 2.0876190476190475,
      "grad_norm": 3.66829937704388,
      "learning_rate": 2.2376707370853473e-06,
      "loss": 0.4639,
      "step": 1096
    },
    {
      "epoch": 2.0895238095238096,
      "grad_norm": 3.8231149601138017,
      "learning_rate": 2.2291021652058897e-06,
      "loss": 0.4128,
      "step": 1097
    },
    {
      "epoch": 2.0914285714285716,
      "grad_norm": 2.7921788181447202,
      "learning_rate": 2.2205453218145286e-06,
      "loss": 0.3763,
      "step": 1098
    },
    {
      "epoch": 2.0933333333333333,
      "grad_norm": 3.9865751397877496,
      "learning_rate": 2.212000243130151e-06,
      "loss": 0.3469,
      "step": 1099
    },
    {
      "epoch": 2.0952380952380953,
      "grad_norm": 3.0954143664768914,
      "learning_rate": 2.203466965321846e-06,
      "loss": 0.4177,
      "step": 1100
    },
    {
      "epoch": 2.097142857142857,
      "grad_norm": 3.7135269645904363,
      "learning_rate": 2.1949455245087554e-06,
      "loss": 0.4683,
      "step": 1101
    },
    {
      "epoch": 2.099047619047619,
      "grad_norm": 3.957808295416001,
      "learning_rate": 2.186435956759913e-06,
      "loss": 0.5077,
      "step": 1102
    },
    {
      "epoch": 2.100952380952381,
      "grad_norm": 3.5007930514656644,
      "learning_rate": 2.1779382980941042e-06,
      "loss": 0.458,
      "step": 1103
    },
    {
      "epoch": 2.1028571428571428,
      "grad_norm": 2.9406873637767132,
      "learning_rate": 2.1694525844797e-06,
      "loss": 0.4427,
      "step": 1104
    },
    {
      "epoch": 2.104761904761905,
      "grad_norm": 3.9966184949988306,
      "learning_rate": 2.160978851834516e-06,
      "loss": 0.4219,
      "step": 1105
    },
    {
      "epoch": 2.1066666666666665,
      "grad_norm": 2.514429690188998,
      "learning_rate": 2.15251713602565e-06,
      "loss": 0.2744,
      "step": 1106
    },
    {
      "epoch": 2.1085714285714285,
      "grad_norm": 3.602114295244378,
      "learning_rate": 2.144067472869342e-06,
      "loss": 0.5359,
      "step": 1107
    },
    {
      "epoch": 2.1104761904761906,
      "grad_norm": 3.7462220837495566,
      "learning_rate": 2.1356298981308126e-06,
      "loss": 0.4043,
      "step": 1108
    },
    {
      "epoch": 2.1123809523809522,
      "grad_norm": 3.9329481396503154,
      "learning_rate": 2.1272044475241134e-06,
      "loss": 0.5052,
      "step": 1109
    },
    {
      "epoch": 2.1142857142857143,
      "grad_norm": 2.3815082299779666,
      "learning_rate": 2.1187911567119812e-06,
      "loss": 0.3828,
      "step": 1110
    },
    {
      "epoch": 2.1161904761904764,
      "grad_norm": 2.84877978554825,
      "learning_rate": 2.110390061305683e-06,
      "loss": 0.401,
      "step": 1111
    },
    {
      "epoch": 2.118095238095238,
      "grad_norm": 4.320354472675589,
      "learning_rate": 2.1020011968648673e-06,
      "loss": 0.4373,
      "step": 1112
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.701440585490528,
      "learning_rate": 2.0936245988974062e-06,
      "loss": 0.4058,
      "step": 1113
    },
    {
      "epoch": 2.1219047619047617,
      "grad_norm": 3.3798638958857636,
      "learning_rate": 2.085260302859258e-06,
      "loss": 0.5092,
      "step": 1114
    },
    {
      "epoch": 2.123809523809524,
      "grad_norm": 3.130502295225098,
      "learning_rate": 2.0769083441543046e-06,
      "loss": 0.4058,
      "step": 1115
    },
    {
      "epoch": 2.125714285714286,
      "grad_norm": 3.090050846066777,
      "learning_rate": 2.0685687581342127e-06,
      "loss": 0.3739,
      "step": 1116
    },
    {
      "epoch": 2.1276190476190475,
      "grad_norm": 3.2830486598830486,
      "learning_rate": 2.060241580098272e-06,
      "loss": 0.4216,
      "step": 1117
    },
    {
      "epoch": 2.1295238095238096,
      "grad_norm": 3.3343154112824767,
      "learning_rate": 2.05192684529326e-06,
      "loss": 0.3602,
      "step": 1118
    },
    {
      "epoch": 2.1314285714285712,
      "grad_norm": 3.5671379639458602,
      "learning_rate": 2.0436245889132765e-06,
      "loss": 0.4054,
      "step": 1119
    },
    {
      "epoch": 2.1333333333333333,
      "grad_norm": 3.333639073382267,
      "learning_rate": 2.035334846099613e-06,
      "loss": 0.4402,
      "step": 1120
    },
    {
      "epoch": 2.1352380952380954,
      "grad_norm": 3.3805831725285542,
      "learning_rate": 2.027057651940587e-06,
      "loss": 0.3703,
      "step": 1121
    },
    {
      "epoch": 2.137142857142857,
      "grad_norm": 3.03027582099334,
      "learning_rate": 2.0187930414714023e-06,
      "loss": 0.5583,
      "step": 1122
    },
    {
      "epoch": 2.139047619047619,
      "grad_norm": 2.7024682625681673,
      "learning_rate": 2.010541049674003e-06,
      "loss": 0.4717,
      "step": 1123
    },
    {
      "epoch": 2.140952380952381,
      "grad_norm": 3.2578378613917622,
      "learning_rate": 2.0023017114769133e-06,
      "loss": 0.4811,
      "step": 1124
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 3.3474506754750997,
      "learning_rate": 1.994075061755112e-06,
      "loss": 0.4186,
      "step": 1125
    },
    {
      "epoch": 2.144761904761905,
      "grad_norm": 4.22012112067756,
      "learning_rate": 1.9858611353298563e-06,
      "loss": 0.4412,
      "step": 1126
    },
    {
      "epoch": 2.1466666666666665,
      "grad_norm": 2.4212155137342624,
      "learning_rate": 1.977659966968558e-06,
      "loss": 0.4716,
      "step": 1127
    },
    {
      "epoch": 2.1485714285714286,
      "grad_norm": 4.64405232980952,
      "learning_rate": 1.9694715913846228e-06,
      "loss": 0.3662,
      "step": 1128
    },
    {
      "epoch": 2.1504761904761907,
      "grad_norm": 2.7658128060068705,
      "learning_rate": 1.961296043237312e-06,
      "loss": 0.4806,
      "step": 1129
    },
    {
      "epoch": 2.1523809523809523,
      "grad_norm": 3.540213415127392,
      "learning_rate": 1.953133357131586e-06,
      "loss": 0.5283,
      "step": 1130
    },
    {
      "epoch": 2.1542857142857144,
      "grad_norm": 2.969408319601373,
      "learning_rate": 1.94498356761797e-06,
      "loss": 0.3982,
      "step": 1131
    },
    {
      "epoch": 2.156190476190476,
      "grad_norm": 2.872092505122507,
      "learning_rate": 1.9368467091923978e-06,
      "loss": 0.3288,
      "step": 1132
    },
    {
      "epoch": 2.158095238095238,
      "grad_norm": 3.4439979488246877,
      "learning_rate": 1.928722816296066e-06,
      "loss": 0.381,
      "step": 1133
    },
    {
      "epoch": 2.16,
      "grad_norm": 3.4309509220122645,
      "learning_rate": 1.9206119233152996e-06,
      "loss": 0.3921,
      "step": 1134
    },
    {
      "epoch": 2.1619047619047618,
      "grad_norm": 3.1835470436040088,
      "learning_rate": 1.9125140645813904e-06,
      "loss": 0.372,
      "step": 1135
    },
    {
      "epoch": 2.163809523809524,
      "grad_norm": 2.3932633149660867,
      "learning_rate": 1.9044292743704673e-06,
      "loss": 0.3702,
      "step": 1136
    },
    {
      "epoch": 2.1657142857142855,
      "grad_norm": 3.1030703304533698,
      "learning_rate": 1.8963575869033346e-06,
      "loss": 0.3868,
      "step": 1137
    },
    {
      "epoch": 2.1676190476190476,
      "grad_norm": 2.521589719276709,
      "learning_rate": 1.888299036345343e-06,
      "loss": 0.3888,
      "step": 1138
    },
    {
      "epoch": 2.1695238095238096,
      "grad_norm": 4.825811235043827,
      "learning_rate": 1.8802536568062363e-06,
      "loss": 0.3248,
      "step": 1139
    },
    {
      "epoch": 2.1714285714285713,
      "grad_norm": 3.3612064619949007,
      "learning_rate": 1.8722214823400103e-06,
      "loss": 0.4765,
      "step": 1140
    },
    {
      "epoch": 2.1733333333333333,
      "grad_norm": 2.886250636366377,
      "learning_rate": 1.864202546944761e-06,
      "loss": 0.2861,
      "step": 1141
    },
    {
      "epoch": 2.1752380952380954,
      "grad_norm": 2.2837320689744387,
      "learning_rate": 1.8561968845625555e-06,
      "loss": 0.428,
      "step": 1142
    },
    {
      "epoch": 2.177142857142857,
      "grad_norm": 2.8346529566093013,
      "learning_rate": 1.848204529079272e-06,
      "loss": 0.4237,
      "step": 1143
    },
    {
      "epoch": 2.179047619047619,
      "grad_norm": 3.219090745045557,
      "learning_rate": 1.8402255143244701e-06,
      "loss": 0.3996,
      "step": 1144
    },
    {
      "epoch": 2.1809523809523808,
      "grad_norm": 3.192895426645066,
      "learning_rate": 1.8322598740712384e-06,
      "loss": 0.512,
      "step": 1145
    },
    {
      "epoch": 2.182857142857143,
      "grad_norm": 3.629443687291866,
      "learning_rate": 1.8243076420360523e-06,
      "loss": 0.4467,
      "step": 1146
    },
    {
      "epoch": 2.184761904761905,
      "grad_norm": 3.625294884298241,
      "learning_rate": 1.816368851878641e-06,
      "loss": 0.3876,
      "step": 1147
    },
    {
      "epoch": 2.1866666666666665,
      "grad_norm": 4.019880271073956,
      "learning_rate": 1.8084435372018295e-06,
      "loss": 0.4984,
      "step": 1148
    },
    {
      "epoch": 2.1885714285714286,
      "grad_norm": 2.6062029177907773,
      "learning_rate": 1.8005317315514114e-06,
      "loss": 0.3756,
      "step": 1149
    },
    {
      "epoch": 2.1904761904761907,
      "grad_norm": 2.566241646404076,
      "learning_rate": 1.792633468415995e-06,
      "loss": 0.2642,
      "step": 1150
    },
    {
      "epoch": 2.1923809523809523,
      "grad_norm": 3.4474740159911166,
      "learning_rate": 1.7847487812268721e-06,
      "loss": 0.4034,
      "step": 1151
    },
    {
      "epoch": 2.1942857142857144,
      "grad_norm": 2.670620051299626,
      "learning_rate": 1.776877703357862e-06,
      "loss": 0.3992,
      "step": 1152
    },
    {
      "epoch": 2.196190476190476,
      "grad_norm": 3.2849299047974,
      "learning_rate": 1.7690202681251927e-06,
      "loss": 0.4459,
      "step": 1153
    },
    {
      "epoch": 2.198095238095238,
      "grad_norm": 3.66880510649691,
      "learning_rate": 1.7611765087873333e-06,
      "loss": 0.4983,
      "step": 1154
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.2016559419107278,
      "learning_rate": 1.7533464585448755e-06,
      "loss": 0.3522,
      "step": 1155
    },
    {
      "epoch": 2.201904761904762,
      "grad_norm": 3.415633656263071,
      "learning_rate": 1.7455301505403771e-06,
      "loss": 0.3607,
      "step": 1156
    },
    {
      "epoch": 2.203809523809524,
      "grad_norm": 3.511986918826307,
      "learning_rate": 1.7377276178582358e-06,
      "loss": 0.3274,
      "step": 1157
    },
    {
      "epoch": 2.2057142857142855,
      "grad_norm": 3.0218179740679294,
      "learning_rate": 1.7299388935245358e-06,
      "loss": 0.3865,
      "step": 1158
    },
    {
      "epoch": 2.2076190476190476,
      "grad_norm": 3.453936345865144,
      "learning_rate": 1.7221640105069154e-06,
      "loss": 0.3142,
      "step": 1159
    },
    {
      "epoch": 2.2095238095238097,
      "grad_norm": 3.2074621573328543,
      "learning_rate": 1.7144030017144304e-06,
      "loss": 0.471,
      "step": 1160
    },
    {
      "epoch": 2.2114285714285713,
      "grad_norm": 3.661789164085878,
      "learning_rate": 1.7066558999974043e-06,
      "loss": 0.4124,
      "step": 1161
    },
    {
      "epoch": 2.2133333333333334,
      "grad_norm": 3.0883965500037367,
      "learning_rate": 1.6989227381473016e-06,
      "loss": 0.3642,
      "step": 1162
    },
    {
      "epoch": 2.215238095238095,
      "grad_norm": 3.4525578562259582,
      "learning_rate": 1.6912035488965778e-06,
      "loss": 0.3326,
      "step": 1163
    },
    {
      "epoch": 2.217142857142857,
      "grad_norm": 2.930457124153742,
      "learning_rate": 1.683498364918551e-06,
      "loss": 0.3066,
      "step": 1164
    },
    {
      "epoch": 2.219047619047619,
      "grad_norm": 2.5723752034293614,
      "learning_rate": 1.6758072188272529e-06,
      "loss": 0.3661,
      "step": 1165
    },
    {
      "epoch": 2.220952380952381,
      "grad_norm": 4.0659896157792765,
      "learning_rate": 1.668130143177301e-06,
      "loss": 0.3379,
      "step": 1166
    },
    {
      "epoch": 2.222857142857143,
      "grad_norm": 3.737920467163103,
      "learning_rate": 1.6604671704637564e-06,
      "loss": 0.3717,
      "step": 1167
    },
    {
      "epoch": 2.224761904761905,
      "grad_norm": 3.035081661043054,
      "learning_rate": 1.6528183331219844e-06,
      "loss": 0.3519,
      "step": 1168
    },
    {
      "epoch": 2.2266666666666666,
      "grad_norm": 2.841489894580811,
      "learning_rate": 1.645183663527517e-06,
      "loss": 0.3696,
      "step": 1169
    },
    {
      "epoch": 2.2285714285714286,
      "grad_norm": 2.7676880519534763,
      "learning_rate": 1.6375631939959235e-06,
      "loss": 0.2515,
      "step": 1170
    },
    {
      "epoch": 2.2304761904761903,
      "grad_norm": 3.6765967083017186,
      "learning_rate": 1.6299569567826622e-06,
      "loss": 0.4866,
      "step": 1171
    },
    {
      "epoch": 2.2323809523809524,
      "grad_norm": 3.149916698208449,
      "learning_rate": 1.622364984082951e-06,
      "loss": 0.403,
      "step": 1172
    },
    {
      "epoch": 2.2342857142857144,
      "grad_norm": 2.9585551383344866,
      "learning_rate": 1.614787308031634e-06,
      "loss": 0.3332,
      "step": 1173
    },
    {
      "epoch": 2.236190476190476,
      "grad_norm": 2.7192716574200326,
      "learning_rate": 1.607223960703035e-06,
      "loss": 0.3447,
      "step": 1174
    },
    {
      "epoch": 2.238095238095238,
      "grad_norm": 2.682523524597222,
      "learning_rate": 1.5996749741108347e-06,
      "loss": 0.3914,
      "step": 1175
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.8753179641552764,
      "learning_rate": 1.5921403802079215e-06,
      "loss": 0.3362,
      "step": 1176
    },
    {
      "epoch": 2.241904761904762,
      "grad_norm": 3.1454625399575704,
      "learning_rate": 1.5846202108862708e-06,
      "loss": 0.4496,
      "step": 1177
    },
    {
      "epoch": 2.243809523809524,
      "grad_norm": 2.614532609803121,
      "learning_rate": 1.5771144979767949e-06,
      "loss": 0.3388,
      "step": 1178
    },
    {
      "epoch": 2.2457142857142856,
      "grad_norm": 2.6261149744473937,
      "learning_rate": 1.5696232732492227e-06,
      "loss": 0.4328,
      "step": 1179
    },
    {
      "epoch": 2.2476190476190476,
      "grad_norm": 2.670501922851883,
      "learning_rate": 1.5621465684119513e-06,
      "loss": 0.43,
      "step": 1180
    },
    {
      "epoch": 2.2495238095238097,
      "grad_norm": 2.7287552818929766,
      "learning_rate": 1.5546844151119289e-06,
      "loss": 0.4136,
      "step": 1181
    },
    {
      "epoch": 2.2514285714285713,
      "grad_norm": 2.9070167886277685,
      "learning_rate": 1.547236844934501e-06,
      "loss": 0.4597,
      "step": 1182
    },
    {
      "epoch": 2.2533333333333334,
      "grad_norm": 3.283685902232836,
      "learning_rate": 1.5398038894032903e-06,
      "loss": 0.4166,
      "step": 1183
    },
    {
      "epoch": 2.255238095238095,
      "grad_norm": 4.230049290604123,
      "learning_rate": 1.5323855799800614e-06,
      "loss": 0.3438,
      "step": 1184
    },
    {
      "epoch": 2.257142857142857,
      "grad_norm": 3.406372929627461,
      "learning_rate": 1.5249819480645811e-06,
      "loss": 0.3863,
      "step": 1185
    },
    {
      "epoch": 2.259047619047619,
      "grad_norm": 4.1159605855140855,
      "learning_rate": 1.517593024994497e-06,
      "loss": 0.3358,
      "step": 1186
    },
    {
      "epoch": 2.260952380952381,
      "grad_norm": 3.051894376903517,
      "learning_rate": 1.5102188420451898e-06,
      "loss": 0.373,
      "step": 1187
    },
    {
      "epoch": 2.262857142857143,
      "grad_norm": 3.4516579564846133,
      "learning_rate": 1.5028594304296573e-06,
      "loss": 0.4894,
      "step": 1188
    },
    {
      "epoch": 2.2647619047619045,
      "grad_norm": 3.2107967908714405,
      "learning_rate": 1.495514821298366e-06,
      "loss": 0.5129,
      "step": 1189
    },
    {
      "epoch": 2.2666666666666666,
      "grad_norm": 2.338776131432317,
      "learning_rate": 1.488185045739135e-06,
      "loss": 0.3153,
      "step": 1190
    },
    {
      "epoch": 2.2685714285714287,
      "grad_norm": 2.7746390696586265,
      "learning_rate": 1.4808701347769905e-06,
      "loss": 0.3746,
      "step": 1191
    },
    {
      "epoch": 2.2704761904761903,
      "grad_norm": 3.130401139553232,
      "learning_rate": 1.4735701193740465e-06,
      "loss": 0.3716,
      "step": 1192
    },
    {
      "epoch": 2.2723809523809524,
      "grad_norm": 3.6994407092758466,
      "learning_rate": 1.4662850304293607e-06,
      "loss": 0.394,
      "step": 1193
    },
    {
      "epoch": 2.2742857142857145,
      "grad_norm": 5.126454460234536,
      "learning_rate": 1.4590148987788184e-06,
      "loss": 0.4805,
      "step": 1194
    },
    {
      "epoch": 2.276190476190476,
      "grad_norm": 3.0360236806265193,
      "learning_rate": 1.4517597551949913e-06,
      "loss": 0.3368,
      "step": 1195
    },
    {
      "epoch": 2.278095238095238,
      "grad_norm": 3.6170511722106684,
      "learning_rate": 1.444519630387009e-06,
      "loss": 0.3136,
      "step": 1196
    },
    {
      "epoch": 2.2800000000000002,
      "grad_norm": 2.300799446548077,
      "learning_rate": 1.4372945550004341e-06,
      "loss": 0.3717,
      "step": 1197
    },
    {
      "epoch": 2.281904761904762,
      "grad_norm": 3.572842146297453,
      "learning_rate": 1.4300845596171248e-06,
      "loss": 0.3624,
      "step": 1198
    },
    {
      "epoch": 2.283809523809524,
      "grad_norm": 2.952165089843252,
      "learning_rate": 1.4228896747551136e-06,
      "loss": 0.3673,
      "step": 1199
    },
    {
      "epoch": 2.2857142857142856,
      "grad_norm": 3.0322786600825014,
      "learning_rate": 1.4157099308684702e-06,
      "loss": 0.3715,
      "step": 1200
    },
    {
      "epoch": 2.2876190476190477,
      "grad_norm": 3.882708101241063,
      "learning_rate": 1.4085453583471814e-06,
      "loss": 0.4364,
      "step": 1201
    },
    {
      "epoch": 2.2895238095238097,
      "grad_norm": 3.196731937051594,
      "learning_rate": 1.401395987517012e-06,
      "loss": 0.3987,
      "step": 1202
    },
    {
      "epoch": 2.2914285714285714,
      "grad_norm": 3.385773170082968,
      "learning_rate": 1.3942618486393867e-06,
      "loss": 0.4438,
      "step": 1203
    },
    {
      "epoch": 2.2933333333333334,
      "grad_norm": 4.0973836344009245,
      "learning_rate": 1.3871429719112528e-06,
      "loss": 0.3532,
      "step": 1204
    },
    {
      "epoch": 2.295238095238095,
      "grad_norm": 3.219180576999784,
      "learning_rate": 1.380039387464963e-06,
      "loss": 0.2788,
      "step": 1205
    },
    {
      "epoch": 2.297142857142857,
      "grad_norm": 3.0657965191326544,
      "learning_rate": 1.3729511253681344e-06,
      "loss": 0.3791,
      "step": 1206
    },
    {
      "epoch": 2.2990476190476192,
      "grad_norm": 2.464035187586816,
      "learning_rate": 1.365878215623536e-06,
      "loss": 0.5005,
      "step": 1207
    },
    {
      "epoch": 2.300952380952381,
      "grad_norm": 2.5835905340795855,
      "learning_rate": 1.3588206881689476e-06,
      "loss": 0.3999,
      "step": 1208
    },
    {
      "epoch": 2.302857142857143,
      "grad_norm": 3.6957918375657735,
      "learning_rate": 1.3517785728770432e-06,
      "loss": 0.3608,
      "step": 1209
    },
    {
      "epoch": 2.3047619047619046,
      "grad_norm": 3.135304637867425,
      "learning_rate": 1.3447518995552643e-06,
      "loss": 0.4139,
      "step": 1210
    },
    {
      "epoch": 2.3066666666666666,
      "grad_norm": 3.2007051101269792,
      "learning_rate": 1.3377406979456825e-06,
      "loss": 0.3954,
      "step": 1211
    },
    {
      "epoch": 2.3085714285714287,
      "grad_norm": 2.780756756792339,
      "learning_rate": 1.3307449977248898e-06,
      "loss": 0.3687,
      "step": 1212
    },
    {
      "epoch": 2.3104761904761904,
      "grad_norm": 3.1272455954467167,
      "learning_rate": 1.323764828503858e-06,
      "loss": 0.3496,
      "step": 1213
    },
    {
      "epoch": 2.3123809523809524,
      "grad_norm": 2.830798158375888,
      "learning_rate": 1.3168002198278268e-06,
      "loss": 0.377,
      "step": 1214
    },
    {
      "epoch": 2.314285714285714,
      "grad_norm": 4.102361171482372,
      "learning_rate": 1.309851201176166e-06,
      "loss": 0.4986,
      "step": 1215
    },
    {
      "epoch": 2.316190476190476,
      "grad_norm": 2.9593571901729616,
      "learning_rate": 1.3029178019622624e-06,
      "loss": 0.5163,
      "step": 1216
    },
    {
      "epoch": 2.318095238095238,
      "grad_norm": 3.2051134620417785,
      "learning_rate": 1.2960000515333843e-06,
      "loss": 0.4319,
      "step": 1217
    },
    {
      "epoch": 2.32,
      "grad_norm": 3.5559167967252807,
      "learning_rate": 1.2890979791705688e-06,
      "loss": 0.4566,
      "step": 1218
    },
    {
      "epoch": 2.321904761904762,
      "grad_norm": 3.025199221927989,
      "learning_rate": 1.2822116140884855e-06,
      "loss": 0.3216,
      "step": 1219
    },
    {
      "epoch": 2.323809523809524,
      "grad_norm": 2.579715437235281,
      "learning_rate": 1.2753409854353254e-06,
      "loss": 0.4014,
      "step": 1220
    },
    {
      "epoch": 2.3257142857142856,
      "grad_norm": 2.4851024809676825,
      "learning_rate": 1.2684861222926654e-06,
      "loss": 0.2797,
      "step": 1221
    },
    {
      "epoch": 2.3276190476190477,
      "grad_norm": 2.7059064982076895,
      "learning_rate": 1.2616470536753555e-06,
      "loss": 0.435,
      "step": 1222
    },
    {
      "epoch": 2.3295238095238093,
      "grad_norm": 4.3042464570143375,
      "learning_rate": 1.2548238085313912e-06,
      "loss": 0.4358,
      "step": 1223
    },
    {
      "epoch": 2.3314285714285714,
      "grad_norm": 3.1448520627806036,
      "learning_rate": 1.2480164157417873e-06,
      "loss": 0.4141,
      "step": 1224
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 3.449490552266584,
      "learning_rate": 1.2412249041204654e-06,
      "loss": 0.3241,
      "step": 1225
    },
    {
      "epoch": 2.335238095238095,
      "grad_norm": 2.9052101577150333,
      "learning_rate": 1.2344493024141213e-06,
      "loss": 0.3339,
      "step": 1226
    },
    {
      "epoch": 2.337142857142857,
      "grad_norm": 3.785793650108638,
      "learning_rate": 1.227689639302113e-06,
      "loss": 0.3466,
      "step": 1227
    },
    {
      "epoch": 2.3390476190476193,
      "grad_norm": 3.7439990838976636,
      "learning_rate": 1.220945943396329e-06,
      "loss": 0.296,
      "step": 1228
    },
    {
      "epoch": 2.340952380952381,
      "grad_norm": 3.4275093920414337,
      "learning_rate": 1.2142182432410782e-06,
      "loss": 0.3064,
      "step": 1229
    },
    {
      "epoch": 2.342857142857143,
      "grad_norm": 3.0659322263681092,
      "learning_rate": 1.20750656731296e-06,
      "loss": 0.3069,
      "step": 1230
    },
    {
      "epoch": 2.3447619047619046,
      "grad_norm": 4.064174570719443,
      "learning_rate": 1.2008109440207499e-06,
      "loss": 0.3406,
      "step": 1231
    },
    {
      "epoch": 2.3466666666666667,
      "grad_norm": 4.077566287734659,
      "learning_rate": 1.1941314017052735e-06,
      "loss": 0.3993,
      "step": 1232
    },
    {
      "epoch": 2.3485714285714288,
      "grad_norm": 4.797164051684954,
      "learning_rate": 1.1874679686392942e-06,
      "loss": 0.4089,
      "step": 1233
    },
    {
      "epoch": 2.3504761904761904,
      "grad_norm": 3.596740884749646,
      "learning_rate": 1.1808206730273853e-06,
      "loss": 0.3792,
      "step": 1234
    },
    {
      "epoch": 2.3523809523809525,
      "grad_norm": 4.233433309827213,
      "learning_rate": 1.174189543005816e-06,
      "loss": 0.3625,
      "step": 1235
    },
    {
      "epoch": 2.354285714285714,
      "grad_norm": 3.3148035157580704,
      "learning_rate": 1.1675746066424332e-06,
      "loss": 0.3794,
      "step": 1236
    },
    {
      "epoch": 2.356190476190476,
      "grad_norm": 5.458566216130069,
      "learning_rate": 1.1609758919365343e-06,
      "loss": 0.4712,
      "step": 1237
    },
    {
      "epoch": 2.3580952380952382,
      "grad_norm": 3.1954593518316985,
      "learning_rate": 1.1543934268187617e-06,
      "loss": 0.3712,
      "step": 1238
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.9362257307670765,
      "learning_rate": 1.147827239150971e-06,
      "loss": 0.4358,
      "step": 1239
    },
    {
      "epoch": 2.361904761904762,
      "grad_norm": 2.5507383849093395,
      "learning_rate": 1.141277356726125e-06,
      "loss": 0.3633,
      "step": 1240
    },
    {
      "epoch": 2.3638095238095236,
      "grad_norm": 3.930161843172662,
      "learning_rate": 1.1347438072681654e-06,
      "loss": 0.4992,
      "step": 1241
    },
    {
      "epoch": 2.3657142857142857,
      "grad_norm": 3.6442977513860497,
      "learning_rate": 1.1282266184319074e-06,
      "loss": 0.431,
      "step": 1242
    },
    {
      "epoch": 2.3676190476190477,
      "grad_norm": 2.88509525496193,
      "learning_rate": 1.1217258178029088e-06,
      "loss": 0.3725,
      "step": 1243
    },
    {
      "epoch": 2.3695238095238094,
      "grad_norm": 3.2848467649880493,
      "learning_rate": 1.1152414328973665e-06,
      "loss": 0.402,
      "step": 1244
    },
    {
      "epoch": 2.3714285714285714,
      "grad_norm": 3.815313038557439,
      "learning_rate": 1.1087734911619884e-06,
      "loss": 0.4339,
      "step": 1245
    },
    {
      "epoch": 2.3733333333333335,
      "grad_norm": 3.6309379264671815,
      "learning_rate": 1.1023220199738882e-06,
      "loss": 0.3348,
      "step": 1246
    },
    {
      "epoch": 2.375238095238095,
      "grad_norm": 3.1841497089404665,
      "learning_rate": 1.0958870466404587e-06,
      "loss": 0.3103,
      "step": 1247
    },
    {
      "epoch": 2.3771428571428572,
      "grad_norm": 2.3865502853297635,
      "learning_rate": 1.089468598399267e-06,
      "loss": 0.36,
      "step": 1248
    },
    {
      "epoch": 2.379047619047619,
      "grad_norm": 2.4753079239449973,
      "learning_rate": 1.0830667024179286e-06,
      "loss": 0.4125,
      "step": 1249
    },
    {
      "epoch": 2.380952380952381,
      "grad_norm": 3.637229151331397,
      "learning_rate": 1.0766813857940012e-06,
      "loss": 0.3054,
      "step": 1250
    },
    {
      "epoch": 2.382857142857143,
      "grad_norm": 3.5240887449222065,
      "learning_rate": 1.0703126755548666e-06,
      "loss": 0.3594,
      "step": 1251
    },
    {
      "epoch": 2.3847619047619046,
      "grad_norm": 2.8892907297195505,
      "learning_rate": 1.0639605986576117e-06,
      "loss": 0.3579,
      "step": 1252
    },
    {
      "epoch": 2.3866666666666667,
      "grad_norm": 2.7937069485868085,
      "learning_rate": 1.0576251819889256e-06,
      "loss": 0.3601,
      "step": 1253
    },
    {
      "epoch": 2.388571428571429,
      "grad_norm": 2.6891544024815306,
      "learning_rate": 1.0513064523649718e-06,
      "loss": 0.4678,
      "step": 1254
    },
    {
      "epoch": 2.3904761904761904,
      "grad_norm": 3.267206768884853,
      "learning_rate": 1.0450044365312878e-06,
      "loss": 0.3913,
      "step": 1255
    },
    {
      "epoch": 2.3923809523809525,
      "grad_norm": 2.9980974451733062,
      "learning_rate": 1.0387191611626613e-06,
      "loss": 0.3417,
      "step": 1256
    },
    {
      "epoch": 2.394285714285714,
      "grad_norm": 4.961973510439375,
      "learning_rate": 1.0324506528630275e-06,
      "loss": 0.57,
      "step": 1257
    },
    {
      "epoch": 2.396190476190476,
      "grad_norm": 2.2876925610079,
      "learning_rate": 1.0261989381653436e-06,
      "loss": 0.4377,
      "step": 1258
    },
    {
      "epoch": 2.3980952380952383,
      "grad_norm": 4.19103461984004,
      "learning_rate": 1.0199640435314912e-06,
      "loss": 0.5733,
      "step": 1259
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.2220638450965318,
      "learning_rate": 1.0137459953521516e-06,
      "loss": 0.3965,
      "step": 1260
    },
    {
      "epoch": 2.401904761904762,
      "grad_norm": 2.7274877458539626,
      "learning_rate": 1.0075448199467037e-06,
      "loss": 0.3123,
      "step": 1261
    },
    {
      "epoch": 2.4038095238095236,
      "grad_norm": 2.6098326894467747,
      "learning_rate": 1.0013605435631035e-06,
      "loss": 0.4197,
      "step": 1262
    },
    {
      "epoch": 2.4057142857142857,
      "grad_norm": 3.1376814358846947,
      "learning_rate": 9.951931923777825e-07,
      "loss": 0.3685,
      "step": 1263
    },
    {
      "epoch": 2.4076190476190478,
      "grad_norm": 3.892536895474269,
      "learning_rate": 9.890427924955304e-07,
      "loss": 0.4284,
      "step": 1264
    },
    {
      "epoch": 2.4095238095238094,
      "grad_norm": 3.752672645105414,
      "learning_rate": 9.829093699493847e-07,
      "loss": 0.4295,
      "step": 1265
    },
    {
      "epoch": 2.4114285714285715,
      "grad_norm": 3.4976663359860223,
      "learning_rate": 9.767929507005263e-07,
      "loss": 0.2839,
      "step": 1266
    },
    {
      "epoch": 2.413333333333333,
      "grad_norm": 3.4083427867856915,
      "learning_rate": 9.706935606381606e-07,
      "loss": 0.4173,
      "step": 1267
    },
    {
      "epoch": 2.415238095238095,
      "grad_norm": 2.7603969390144716,
      "learning_rate": 9.646112255794182e-07,
      "loss": 0.4539,
      "step": 1268
    },
    {
      "epoch": 2.4171428571428573,
      "grad_norm": 2.86185545317862,
      "learning_rate": 9.585459712692347e-07,
      "loss": 0.4081,
      "step": 1269
    },
    {
      "epoch": 2.419047619047619,
      "grad_norm": 3.5080078245583177,
      "learning_rate": 9.524978233802529e-07,
      "loss": 0.3905,
      "step": 1270
    },
    {
      "epoch": 2.420952380952381,
      "grad_norm": 3.1485365755593784,
      "learning_rate": 9.464668075127032e-07,
      "loss": 0.5148,
      "step": 1271
    },
    {
      "epoch": 2.422857142857143,
      "grad_norm": 3.259521974462558,
      "learning_rate": 9.40452949194306e-07,
      "loss": 0.3295,
      "step": 1272
    },
    {
      "epoch": 2.4247619047619047,
      "grad_norm": 2.647219117161924,
      "learning_rate": 9.344562738801532e-07,
      "loss": 0.4309,
      "step": 1273
    },
    {
      "epoch": 2.4266666666666667,
      "grad_norm": 2.819943962692332,
      "learning_rate": 9.284768069526106e-07,
      "loss": 0.3978,
      "step": 1274
    },
    {
      "epoch": 2.4285714285714284,
      "grad_norm": 3.37110171891748,
      "learning_rate": 9.225145737211999e-07,
      "loss": 0.3497,
      "step": 1275
    },
    {
      "epoch": 2.4304761904761905,
      "grad_norm": 3.2343737650750284,
      "learning_rate": 9.165695994225027e-07,
      "loss": 0.484,
      "step": 1276
    },
    {
      "epoch": 2.4323809523809525,
      "grad_norm": 2.4102341743748266,
      "learning_rate": 9.106419092200419e-07,
      "loss": 0.3887,
      "step": 1277
    },
    {
      "epoch": 2.434285714285714,
      "grad_norm": 2.881380818665366,
      "learning_rate": 9.047315282041868e-07,
      "loss": 0.3833,
      "step": 1278
    },
    {
      "epoch": 2.4361904761904762,
      "grad_norm": 3.586289356745323,
      "learning_rate": 8.988384813920403e-07,
      "loss": 0.3013,
      "step": 1279
    },
    {
      "epoch": 2.4380952380952383,
      "grad_norm": 2.76177799494823,
      "learning_rate": 8.929627937273306e-07,
      "loss": 0.384,
      "step": 1280
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.7208978965938915,
      "learning_rate": 8.871044900803139e-07,
      "loss": 0.3112,
      "step": 1281
    },
    {
      "epoch": 2.441904761904762,
      "grad_norm": 3.8741608271506425,
      "learning_rate": 8.812635952476595e-07,
      "loss": 0.4467,
      "step": 1282
    },
    {
      "epoch": 2.4438095238095237,
      "grad_norm": 2.547711345625711,
      "learning_rate": 8.754401339523555e-07,
      "loss": 0.371,
      "step": 1283
    },
    {
      "epoch": 2.4457142857142857,
      "grad_norm": 3.1248629401369876,
      "learning_rate": 8.696341308435918e-07,
      "loss": 0.6343,
      "step": 1284
    },
    {
      "epoch": 2.447619047619048,
      "grad_norm": 4.534808822420833,
      "learning_rate": 8.638456104966697e-07,
      "loss": 0.4842,
      "step": 1285
    },
    {
      "epoch": 2.4495238095238094,
      "grad_norm": 2.483158317099844,
      "learning_rate": 8.580745974128829e-07,
      "loss": 0.501,
      "step": 1286
    },
    {
      "epoch": 2.4514285714285715,
      "grad_norm": 3.1165792064698055,
      "learning_rate": 8.523211160194278e-07,
      "loss": 0.3496,
      "step": 1287
    },
    {
      "epoch": 2.453333333333333,
      "grad_norm": 2.9019588757409323,
      "learning_rate": 8.46585190669289e-07,
      "loss": 0.3665,
      "step": 1288
    },
    {
      "epoch": 2.455238095238095,
      "grad_norm": 3.585524995120755,
      "learning_rate": 8.408668456411456e-07,
      "loss": 0.5054,
      "step": 1289
    },
    {
      "epoch": 2.4571428571428573,
      "grad_norm": 3.017357271562797,
      "learning_rate": 8.351661051392584e-07,
      "loss": 0.3167,
      "step": 1290
    },
    {
      "epoch": 2.459047619047619,
      "grad_norm": 3.8636051850719144,
      "learning_rate": 8.294829932933767e-07,
      "loss": 0.4196,
      "step": 1291
    },
    {
      "epoch": 2.460952380952381,
      "grad_norm": 4.981089118580047,
      "learning_rate": 8.238175341586325e-07,
      "loss": 0.3626,
      "step": 1292
    },
    {
      "epoch": 2.4628571428571426,
      "grad_norm": 2.5514861598971197,
      "learning_rate": 8.181697517154352e-07,
      "loss": 0.4561,
      "step": 1293
    },
    {
      "epoch": 2.4647619047619047,
      "grad_norm": 2.8188578033727176,
      "learning_rate": 8.125396698693777e-07,
      "loss": 0.2887,
      "step": 1294
    },
    {
      "epoch": 2.466666666666667,
      "grad_norm": 2.478880902222213,
      "learning_rate": 8.069273124511256e-07,
      "loss": 0.3192,
      "step": 1295
    },
    {
      "epoch": 2.4685714285714284,
      "grad_norm": 4.658508362356186,
      "learning_rate": 8.013327032163277e-07,
      "loss": 0.4468,
      "step": 1296
    },
    {
      "epoch": 2.4704761904761905,
      "grad_norm": 2.7622379592389965,
      "learning_rate": 7.957558658455028e-07,
      "loss": 0.4443,
      "step": 1297
    },
    {
      "epoch": 2.4723809523809526,
      "grad_norm": 3.153774592176001,
      "learning_rate": 7.901968239439528e-07,
      "loss": 0.3512,
      "step": 1298
    },
    {
      "epoch": 2.474285714285714,
      "grad_norm": 4.3831048273406035,
      "learning_rate": 7.846556010416506e-07,
      "loss": 0.3387,
      "step": 1299
    },
    {
      "epoch": 2.4761904761904763,
      "grad_norm": 2.531799103099177,
      "learning_rate": 7.791322205931495e-07,
      "loss": 0.4368,
      "step": 1300
    },
    {
      "epoch": 2.478095238095238,
      "grad_norm": 2.6820930155282143,
      "learning_rate": 7.736267059774765e-07,
      "loss": 0.4298,
      "step": 1301
    },
    {
      "epoch": 2.48,
      "grad_norm": 3.6156272420973075,
      "learning_rate": 7.681390804980427e-07,
      "loss": 0.2211,
      "step": 1302
    },
    {
      "epoch": 2.481904761904762,
      "grad_norm": 2.5307331169149045,
      "learning_rate": 7.626693673825331e-07,
      "loss": 0.3678,
      "step": 1303
    },
    {
      "epoch": 2.4838095238095237,
      "grad_norm": 4.493697410671703,
      "learning_rate": 7.572175897828194e-07,
      "loss": 0.3768,
      "step": 1304
    },
    {
      "epoch": 2.4857142857142858,
      "grad_norm": 2.641598284220468,
      "learning_rate": 7.517837707748532e-07,
      "loss": 0.3967,
      "step": 1305
    },
    {
      "epoch": 2.487619047619048,
      "grad_norm": 3.4005298819475467,
      "learning_rate": 7.463679333585738e-07,
      "loss": 0.3599,
      "step": 1306
    },
    {
      "epoch": 2.4895238095238095,
      "grad_norm": 2.253509952531745,
      "learning_rate": 7.409701004578107e-07,
      "loss": 0.3859,
      "step": 1307
    },
    {
      "epoch": 2.4914285714285715,
      "grad_norm": 2.649069855195501,
      "learning_rate": 7.355902949201815e-07,
      "loss": 0.3645,
      "step": 1308
    },
    {
      "epoch": 2.493333333333333,
      "grad_norm": 2.985186218107175,
      "learning_rate": 7.302285395170012e-07,
      "loss": 0.3504,
      "step": 1309
    },
    {
      "epoch": 2.4952380952380953,
      "grad_norm": 3.688843440211025,
      "learning_rate": 7.248848569431815e-07,
      "loss": 0.3659,
      "step": 1310
    },
    {
      "epoch": 2.4971428571428573,
      "grad_norm": 4.05327133401772,
      "learning_rate": 7.195592698171394e-07,
      "loss": 0.3845,
      "step": 1311
    },
    {
      "epoch": 2.499047619047619,
      "grad_norm": 2.3142681128860776,
      "learning_rate": 7.142518006806936e-07,
      "loss": 0.3501,
      "step": 1312
    },
    {
      "epoch": 2.500952380952381,
      "grad_norm": 3.4686533405931126,
      "learning_rate": 7.089624719989807e-07,
      "loss": 0.3473,
      "step": 1313
    },
    {
      "epoch": 2.5028571428571427,
      "grad_norm": 2.9404207774730122,
      "learning_rate": 7.036913061603473e-07,
      "loss": 0.3238,
      "step": 1314
    },
    {
      "epoch": 2.5047619047619047,
      "grad_norm": 4.15485613539608,
      "learning_rate": 6.984383254762661e-07,
      "loss": 0.4574,
      "step": 1315
    },
    {
      "epoch": 2.506666666666667,
      "grad_norm": 2.7557936823792604,
      "learning_rate": 6.932035521812336e-07,
      "loss": 0.3371,
      "step": 1316
    },
    {
      "epoch": 2.5085714285714285,
      "grad_norm": 2.599832908064384,
      "learning_rate": 6.879870084326795e-07,
      "loss": 0.3433,
      "step": 1317
    },
    {
      "epoch": 2.5104761904761905,
      "grad_norm": 3.5663635213742815,
      "learning_rate": 6.827887163108754e-07,
      "loss": 0.5258,
      "step": 1318
    },
    {
      "epoch": 2.512380952380952,
      "grad_norm": 2.843365900601389,
      "learning_rate": 6.776086978188334e-07,
      "loss": 0.241,
      "step": 1319
    },
    {
      "epoch": 2.5142857142857142,
      "grad_norm": 2.536016449353789,
      "learning_rate": 6.724469748822254e-07,
      "loss": 0.512,
      "step": 1320
    },
    {
      "epoch": 2.5161904761904763,
      "grad_norm": 3.5526772582835204,
      "learning_rate": 6.673035693492758e-07,
      "loss": 0.3119,
      "step": 1321
    },
    {
      "epoch": 2.518095238095238,
      "grad_norm": 2.5327402916683863,
      "learning_rate": 6.621785029906813e-07,
      "loss": 0.3636,
      "step": 1322
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.381014927773133,
      "learning_rate": 6.570717974995094e-07,
      "loss": 0.479,
      "step": 1323
    },
    {
      "epoch": 2.5219047619047616,
      "grad_norm": 2.3551915060559887,
      "learning_rate": 6.519834744911146e-07,
      "loss": 0.308,
      "step": 1324
    },
    {
      "epoch": 2.5238095238095237,
      "grad_norm": 3.1540737828322234,
      "learning_rate": 6.469135555030403e-07,
      "loss": 0.4138,
      "step": 1325
    },
    {
      "epoch": 2.525714285714286,
      "grad_norm": 2.6493262273934115,
      "learning_rate": 6.418620619949334e-07,
      "loss": 0.3728,
      "step": 1326
    },
    {
      "epoch": 2.527619047619048,
      "grad_norm": 2.8419165745822736,
      "learning_rate": 6.368290153484469e-07,
      "loss": 0.4861,
      "step": 1327
    },
    {
      "epoch": 2.5295238095238095,
      "grad_norm": 3.8726166413097607,
      "learning_rate": 6.31814436867157e-07,
      "loss": 0.4817,
      "step": 1328
    },
    {
      "epoch": 2.5314285714285716,
      "grad_norm": 3.2432496298782083,
      "learning_rate": 6.268183477764655e-07,
      "loss": 0.4095,
      "step": 1329
    },
    {
      "epoch": 2.533333333333333,
      "grad_norm": 4.032195981982127,
      "learning_rate": 6.218407692235152e-07,
      "loss": 0.3235,
      "step": 1330
    },
    {
      "epoch": 2.5352380952380953,
      "grad_norm": 3.5709363772617597,
      "learning_rate": 6.168817222770989e-07,
      "loss": 0.4395,
      "step": 1331
    },
    {
      "epoch": 2.5371428571428574,
      "grad_norm": 3.5346555515961247,
      "learning_rate": 6.119412279275683e-07,
      "loss": 0.3615,
      "step": 1332
    },
    {
      "epoch": 2.539047619047619,
      "grad_norm": 3.3497917270237303,
      "learning_rate": 6.070193070867486e-07,
      "loss": 0.4471,
      "step": 1333
    },
    {
      "epoch": 2.540952380952381,
      "grad_norm": 2.2466358209673634,
      "learning_rate": 6.021159805878474e-07,
      "loss": 0.5183,
      "step": 1334
    },
    {
      "epoch": 2.5428571428571427,
      "grad_norm": 3.596964073046631,
      "learning_rate": 5.972312691853688e-07,
      "loss": 0.4359,
      "step": 1335
    },
    {
      "epoch": 2.544761904761905,
      "grad_norm": 2.7079095099532067,
      "learning_rate": 5.923651935550195e-07,
      "loss": 0.3068,
      "step": 1336
    },
    {
      "epoch": 2.546666666666667,
      "grad_norm": 2.7033135992266817,
      "learning_rate": 5.875177742936322e-07,
      "loss": 0.3709,
      "step": 1337
    },
    {
      "epoch": 2.5485714285714285,
      "grad_norm": 2.652968881037819,
      "learning_rate": 5.826890319190665e-07,
      "loss": 0.4036,
      "step": 1338
    },
    {
      "epoch": 2.5504761904761906,
      "grad_norm": 2.6371934100581336,
      "learning_rate": 5.778789868701312e-07,
      "loss": 0.4442,
      "step": 1339
    },
    {
      "epoch": 2.552380952380952,
      "grad_norm": 2.8582048423887576,
      "learning_rate": 5.730876595064916e-07,
      "loss": 0.5423,
      "step": 1340
    },
    {
      "epoch": 2.5542857142857143,
      "grad_norm": 3.1528069659987383,
      "learning_rate": 5.683150701085887e-07,
      "loss": 0.3365,
      "step": 1341
    },
    {
      "epoch": 2.5561904761904763,
      "grad_norm": 3.930743161228964,
      "learning_rate": 5.635612388775486e-07,
      "loss": 0.5023,
      "step": 1342
    },
    {
      "epoch": 2.558095238095238,
      "grad_norm": 3.756322935202279,
      "learning_rate": 5.588261859350985e-07,
      "loss": 0.3707,
      "step": 1343
    },
    {
      "epoch": 2.56,
      "grad_norm": 3.405875755916657,
      "learning_rate": 5.541099313234844e-07,
      "loss": 0.4591,
      "step": 1344
    },
    {
      "epoch": 2.5619047619047617,
      "grad_norm": 4.1807164501016825,
      "learning_rate": 5.494124950053803e-07,
      "loss": 0.5218,
      "step": 1345
    },
    {
      "epoch": 2.5638095238095238,
      "grad_norm": 3.458898909701398,
      "learning_rate": 5.447338968638116e-07,
      "loss": 0.4553,
      "step": 1346
    },
    {
      "epoch": 2.565714285714286,
      "grad_norm": 6.058750231115991,
      "learning_rate": 5.400741567020634e-07,
      "loss": 0.334,
      "step": 1347
    },
    {
      "epoch": 2.5676190476190475,
      "grad_norm": 2.6103683471853616,
      "learning_rate": 5.354332942436019e-07,
      "loss": 0.3754,
      "step": 1348
    },
    {
      "epoch": 2.5695238095238095,
      "grad_norm": 2.3986875583973655,
      "learning_rate": 5.308113291319867e-07,
      "loss": 0.4138,
      "step": 1349
    },
    {
      "epoch": 2.571428571428571,
      "grad_norm": 3.4869072096598175,
      "learning_rate": 5.262082809307922e-07,
      "loss": 0.4711,
      "step": 1350
    },
    {
      "epoch": 2.5733333333333333,
      "grad_norm": 2.9784366132074536,
      "learning_rate": 5.216241691235197e-07,
      "loss": 0.3354,
      "step": 1351
    },
    {
      "epoch": 2.5752380952380953,
      "grad_norm": 3.180436288983747,
      "learning_rate": 5.170590131135211e-07,
      "loss": 0.3609,
      "step": 1352
    },
    {
      "epoch": 2.5771428571428574,
      "grad_norm": 3.8361573002155396,
      "learning_rate": 5.125128322239098e-07,
      "loss": 0.396,
      "step": 1353
    },
    {
      "epoch": 2.579047619047619,
      "grad_norm": 2.6090330681331313,
      "learning_rate": 5.079856456974857e-07,
      "loss": 0.3995,
      "step": 1354
    },
    {
      "epoch": 2.580952380952381,
      "grad_norm": 3.732794756755359,
      "learning_rate": 5.034774726966484e-07,
      "loss": 0.2656,
      "step": 1355
    },
    {
      "epoch": 2.5828571428571427,
      "grad_norm": 3.5102566381961995,
      "learning_rate": 4.989883323033184e-07,
      "loss": 0.4213,
      "step": 1356
    },
    {
      "epoch": 2.584761904761905,
      "grad_norm": 2.7909195722765356,
      "learning_rate": 4.945182435188578e-07,
      "loss": 0.394,
      "step": 1357
    },
    {
      "epoch": 2.586666666666667,
      "grad_norm": 2.7819914481638848,
      "learning_rate": 4.900672252639849e-07,
      "loss": 0.3475,
      "step": 1358
    },
    {
      "epoch": 2.5885714285714285,
      "grad_norm": 2.6724145817445946,
      "learning_rate": 4.85635296378702e-07,
      "loss": 0.4638,
      "step": 1359
    },
    {
      "epoch": 2.5904761904761906,
      "grad_norm": 3.5006447584276623,
      "learning_rate": 4.81222475622205e-07,
      "loss": 0.4069,
      "step": 1360
    },
    {
      "epoch": 2.5923809523809522,
      "grad_norm": 3.0844175410211525,
      "learning_rate": 4.7682878167281843e-07,
      "loss": 0.3402,
      "step": 1361
    },
    {
      "epoch": 2.5942857142857143,
      "grad_norm": 3.9376326123065835,
      "learning_rate": 4.724542331278992e-07,
      "loss": 0.3444,
      "step": 1362
    },
    {
      "epoch": 2.5961904761904764,
      "grad_norm": 2.823796348027554,
      "learning_rate": 4.680988485037735e-07,
      "loss": 0.2914,
      "step": 1363
    },
    {
      "epoch": 2.598095238095238,
      "grad_norm": 2.4702037734496,
      "learning_rate": 4.6376264623564704e-07,
      "loss": 0.413,
      "step": 1364
    },
    {
      "epoch": 2.6,
      "grad_norm": 3.849088204366855,
      "learning_rate": 4.594456446775347e-07,
      "loss": 0.3225,
      "step": 1365
    },
    {
      "epoch": 2.6019047619047617,
      "grad_norm": 2.51525711757472,
      "learning_rate": 4.5514786210217766e-07,
      "loss": 0.4084,
      "step": 1366
    },
    {
      "epoch": 2.603809523809524,
      "grad_norm": 2.642093902899198,
      "learning_rate": 4.5086931670096865e-07,
      "loss": 0.4074,
      "step": 1367
    },
    {
      "epoch": 2.605714285714286,
      "grad_norm": 2.9969418485889556,
      "learning_rate": 4.4661002658387685e-07,
      "loss": 0.414,
      "step": 1368
    },
    {
      "epoch": 2.6076190476190475,
      "grad_norm": 3.0151416444159054,
      "learning_rate": 4.4237000977936374e-07,
      "loss": 0.3344,
      "step": 1369
    },
    {
      "epoch": 2.6095238095238096,
      "grad_norm": 3.798310147755509,
      "learning_rate": 4.3814928423431826e-07,
      "loss": 0.5015,
      "step": 1370
    },
    {
      "epoch": 2.611428571428571,
      "grad_norm": 2.7051732198925413,
      "learning_rate": 4.339478678139686e-07,
      "loss": 0.3423,
      "step": 1371
    },
    {
      "epoch": 2.6133333333333333,
      "grad_norm": 3.5429724013247696,
      "learning_rate": 4.2976577830181663e-07,
      "loss": 0.3774,
      "step": 1372
    },
    {
      "epoch": 2.6152380952380954,
      "grad_norm": 3.9592191869302726,
      "learning_rate": 4.256030333995559e-07,
      "loss": 0.3812,
      "step": 1373
    },
    {
      "epoch": 2.617142857142857,
      "grad_norm": 2.282944362851788,
      "learning_rate": 4.2145965072700077e-07,
      "loss": 0.4928,
      "step": 1374
    },
    {
      "epoch": 2.619047619047619,
      "grad_norm": 2.8351895207824356,
      "learning_rate": 4.1733564782200973e-07,
      "loss": 0.3266,
      "step": 1375
    },
    {
      "epoch": 2.6209523809523807,
      "grad_norm": 2.779981276567339,
      "learning_rate": 4.132310421404123e-07,
      "loss": 0.3837,
      "step": 1376
    },
    {
      "epoch": 2.6228571428571428,
      "grad_norm": 2.845737076699493,
      "learning_rate": 4.091458510559332e-07,
      "loss": 0.4606,
      "step": 1377
    },
    {
      "epoch": 2.624761904761905,
      "grad_norm": 2.828519446651989,
      "learning_rate": 4.0508009186012277e-07,
      "loss": 0.2951,
      "step": 1378
    },
    {
      "epoch": 2.626666666666667,
      "grad_norm": 3.3949772921852968,
      "learning_rate": 4.0103378176227834e-07,
      "loss": 0.3903,
      "step": 1379
    },
    {
      "epoch": 2.6285714285714286,
      "grad_norm": 6.074886731771548,
      "learning_rate": 3.9700693788937537e-07,
      "loss": 0.436,
      "step": 1380
    },
    {
      "epoch": 2.6304761904761906,
      "grad_norm": 4.499566828369018,
      "learning_rate": 3.9299957728599516e-07,
      "loss": 0.3919,
      "step": 1381
    },
    {
      "epoch": 2.6323809523809523,
      "grad_norm": 2.9742452007328217,
      "learning_rate": 3.8901171691424876e-07,
      "loss": 0.507,
      "step": 1382
    },
    {
      "epoch": 2.6342857142857143,
      "grad_norm": 3.7095646212873645,
      "learning_rate": 3.8504337365371125e-07,
      "loss": 0.314,
      "step": 1383
    },
    {
      "epoch": 2.6361904761904764,
      "grad_norm": 3.3148733787507125,
      "learning_rate": 3.8109456430134364e-07,
      "loss": 0.3758,
      "step": 1384
    },
    {
      "epoch": 2.638095238095238,
      "grad_norm": 2.917740867278345,
      "learning_rate": 3.7716530557142695e-07,
      "loss": 0.3709,
      "step": 1385
    },
    {
      "epoch": 2.64,
      "grad_norm": 3.5169704814332583,
      "learning_rate": 3.732556140954885e-07,
      "loss": 0.4023,
      "step": 1386
    },
    {
      "epoch": 2.6419047619047618,
      "grad_norm": 4.205817670058243,
      "learning_rate": 3.69365506422234e-07,
      "loss": 0.3434,
      "step": 1387
    },
    {
      "epoch": 2.643809523809524,
      "grad_norm": 3.397887260874195,
      "learning_rate": 3.6549499901747276e-07,
      "loss": 0.3697,
      "step": 1388
    },
    {
      "epoch": 2.645714285714286,
      "grad_norm": 2.649276590805659,
      "learning_rate": 3.6164410826405704e-07,
      "loss": 0.3382,
      "step": 1389
    },
    {
      "epoch": 2.6476190476190475,
      "grad_norm": 3.6697458488411936,
      "learning_rate": 3.5781285046180005e-07,
      "loss": 0.4605,
      "step": 1390
    },
    {
      "epoch": 2.6495238095238096,
      "grad_norm": 4.573104325720787,
      "learning_rate": 3.5400124182741913e-07,
      "loss": 0.454,
      "step": 1391
    },
    {
      "epoch": 2.6514285714285712,
      "grad_norm": 2.742805647411334,
      "learning_rate": 3.502092984944594e-07,
      "loss": 0.3325,
      "step": 1392
    },
    {
      "epoch": 2.6533333333333333,
      "grad_norm": 2.9476397630311078,
      "learning_rate": 3.4643703651322745e-07,
      "loss": 0.4399,
      "step": 1393
    },
    {
      "epoch": 2.6552380952380954,
      "grad_norm": 3.55869789840011,
      "learning_rate": 3.426844718507261e-07,
      "loss": 0.4367,
      "step": 1394
    },
    {
      "epoch": 2.657142857142857,
      "grad_norm": 3.850299601922021,
      "learning_rate": 3.389516203905818e-07,
      "loss": 0.4826,
      "step": 1395
    },
    {
      "epoch": 2.659047619047619,
      "grad_norm": 2.830147432712001,
      "learning_rate": 3.352384979329826e-07,
      "loss": 0.374,
      "step": 1396
    },
    {
      "epoch": 2.6609523809523807,
      "grad_norm": 2.590774281488631,
      "learning_rate": 3.3154512019460736e-07,
      "loss": 0.3244,
      "step": 1397
    },
    {
      "epoch": 2.662857142857143,
      "grad_norm": 2.867865871729688,
      "learning_rate": 3.2787150280856106e-07,
      "loss": 0.3995,
      "step": 1398
    },
    {
      "epoch": 2.664761904761905,
      "grad_norm": 2.612773831567773,
      "learning_rate": 3.242176613243081e-07,
      "loss": 0.3918,
      "step": 1399
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 3.2587468213415027,
      "learning_rate": 3.205836112076077e-07,
      "loss": 0.3468,
      "step": 1400
    },
    {
      "epoch": 2.6685714285714286,
      "grad_norm": 3.106198659399771,
      "learning_rate": 3.169693678404456e-07,
      "loss": 0.4179,
      "step": 1401
    },
    {
      "epoch": 2.6704761904761902,
      "grad_norm": 3.6793058346398184,
      "learning_rate": 3.133749465209718e-07,
      "loss": 0.3862,
      "step": 1402
    },
    {
      "epoch": 2.6723809523809523,
      "grad_norm": 2.457921563894658,
      "learning_rate": 3.0980036246343435e-07,
      "loss": 0.4651,
      "step": 1403
    },
    {
      "epoch": 2.6742857142857144,
      "grad_norm": 3.561474305675107,
      "learning_rate": 3.0624563079811707e-07,
      "loss": 0.3603,
      "step": 1404
    },
    {
      "epoch": 2.6761904761904765,
      "grad_norm": 3.3406922223285735,
      "learning_rate": 3.0271076657127143e-07,
      "loss": 0.4338,
      "step": 1405
    },
    {
      "epoch": 2.678095238095238,
      "grad_norm": 3.209473700097369,
      "learning_rate": 2.991957847450544e-07,
      "loss": 0.4381,
      "step": 1406
    },
    {
      "epoch": 2.68,
      "grad_norm": 3.546067059971776,
      "learning_rate": 2.957007001974704e-07,
      "loss": 0.4188,
      "step": 1407
    },
    {
      "epoch": 2.681904761904762,
      "grad_norm": 3.96011605011883,
      "learning_rate": 2.922255277222991e-07,
      "loss": 0.3798,
      "step": 1408
    },
    {
      "epoch": 2.683809523809524,
      "grad_norm": 3.8059135549293424,
      "learning_rate": 2.8877028202904113e-07,
      "loss": 0.274,
      "step": 1409
    },
    {
      "epoch": 2.685714285714286,
      "grad_norm": 4.560997007280084,
      "learning_rate": 2.8533497774284824e-07,
      "loss": 0.4536,
      "step": 1410
    },
    {
      "epoch": 2.6876190476190476,
      "grad_norm": 2.7252847235298567,
      "learning_rate": 2.8191962940447073e-07,
      "loss": 0.4049,
      "step": 1411
    },
    {
      "epoch": 2.6895238095238096,
      "grad_norm": 2.7343016725312945,
      "learning_rate": 2.785242514701847e-07,
      "loss": 0.3729,
      "step": 1412
    },
    {
      "epoch": 2.6914285714285713,
      "grad_norm": 2.9167279858792137,
      "learning_rate": 2.7514885831174286e-07,
      "loss": 0.3592,
      "step": 1413
    },
    {
      "epoch": 2.6933333333333334,
      "grad_norm": 2.3358141872441576,
      "learning_rate": 2.7179346421630113e-07,
      "loss": 0.4178,
      "step": 1414
    },
    {
      "epoch": 2.6952380952380954,
      "grad_norm": 3.9378285249680385,
      "learning_rate": 2.684580833863709e-07,
      "loss": 0.3633,
      "step": 1415
    },
    {
      "epoch": 2.697142857142857,
      "grad_norm": 3.246397261454276,
      "learning_rate": 2.651427299397469e-07,
      "loss": 0.4293,
      "step": 1416
    },
    {
      "epoch": 2.699047619047619,
      "grad_norm": 2.6996868196263137,
      "learning_rate": 2.6184741790945767e-07,
      "loss": 0.3808,
      "step": 1417
    },
    {
      "epoch": 2.7009523809523808,
      "grad_norm": 3.3285473725669785,
      "learning_rate": 2.5857216124369977e-07,
      "loss": 0.3566,
      "step": 1418
    },
    {
      "epoch": 2.702857142857143,
      "grad_norm": 3.0290915678180266,
      "learning_rate": 2.553169738057798e-07,
      "loss": 0.4717,
      "step": 1419
    },
    {
      "epoch": 2.704761904761905,
      "grad_norm": 2.743377770077459,
      "learning_rate": 2.5208186937405845e-07,
      "loss": 0.3877,
      "step": 1420
    },
    {
      "epoch": 2.7066666666666666,
      "grad_norm": 3.499598182556431,
      "learning_rate": 2.488668616418888e-07,
      "loss": 0.3816,
      "step": 1421
    },
    {
      "epoch": 2.7085714285714286,
      "grad_norm": 3.3128107331165704,
      "learning_rate": 2.4567196421756034e-07,
      "loss": 0.4121,
      "step": 1422
    },
    {
      "epoch": 2.7104761904761903,
      "grad_norm": 2.825163170393998,
      "learning_rate": 2.424971906242407e-07,
      "loss": 0.3842,
      "step": 1423
    },
    {
      "epoch": 2.7123809523809523,
      "grad_norm": 2.6359801000422767,
      "learning_rate": 2.3934255429991946e-07,
      "loss": 0.3833,
      "step": 1424
    },
    {
      "epoch": 2.7142857142857144,
      "grad_norm": 2.7877973861490966,
      "learning_rate": 2.3620806859734892e-07,
      "loss": 0.335,
      "step": 1425
    },
    {
      "epoch": 2.716190476190476,
      "grad_norm": 2.325442852658568,
      "learning_rate": 2.330937467839911e-07,
      "loss": 0.4041,
      "step": 1426
    },
    {
      "epoch": 2.718095238095238,
      "grad_norm": 3.6700218613653393,
      "learning_rate": 2.2999960204195593e-07,
      "loss": 0.3112,
      "step": 1427
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 3.8014301204876633,
      "learning_rate": 2.2692564746795375e-07,
      "loss": 0.4377,
      "step": 1428
    },
    {
      "epoch": 2.721904761904762,
      "grad_norm": 2.4998097908967276,
      "learning_rate": 2.238718960732311e-07,
      "loss": 0.339,
      "step": 1429
    },
    {
      "epoch": 2.723809523809524,
      "grad_norm": 2.9387378035324563,
      "learning_rate": 2.2083836078352238e-07,
      "loss": 0.4116,
      "step": 1430
    },
    {
      "epoch": 2.725714285714286,
      "grad_norm": 3.1794317286718847,
      "learning_rate": 2.1782505443899205e-07,
      "loss": 0.4797,
      "step": 1431
    },
    {
      "epoch": 2.7276190476190476,
      "grad_norm": 2.632631500118067,
      "learning_rate": 2.1483198979417863e-07,
      "loss": 0.3901,
      "step": 1432
    },
    {
      "epoch": 2.7295238095238097,
      "grad_norm": 3.346673856710114,
      "learning_rate": 2.1185917951794644e-07,
      "loss": 0.5032,
      "step": 1433
    },
    {
      "epoch": 2.7314285714285713,
      "grad_norm": 2.8640736427366122,
      "learning_rate": 2.0890663619342387e-07,
      "loss": 0.2558,
      "step": 1434
    },
    {
      "epoch": 2.7333333333333334,
      "grad_norm": 2.363884429861665,
      "learning_rate": 2.059743723179597e-07,
      "loss": 0.4956,
      "step": 1435
    },
    {
      "epoch": 2.7352380952380955,
      "grad_norm": 2.6590632628301374,
      "learning_rate": 2.0306240030305956e-07,
      "loss": 0.4531,
      "step": 1436
    },
    {
      "epoch": 2.737142857142857,
      "grad_norm": 4.5851846282878315,
      "learning_rate": 2.001707324743446e-07,
      "loss": 0.3446,
      "step": 1437
    },
    {
      "epoch": 2.739047619047619,
      "grad_norm": 2.364896724892544,
      "learning_rate": 1.972993810714885e-07,
      "loss": 0.341,
      "step": 1438
    },
    {
      "epoch": 2.740952380952381,
      "grad_norm": 3.394643740963008,
      "learning_rate": 1.9444835824817488e-07,
      "loss": 0.4775,
      "step": 1439
    },
    {
      "epoch": 2.742857142857143,
      "grad_norm": 3.1497123888944936,
      "learning_rate": 1.916176760720384e-07,
      "loss": 0.4329,
      "step": 1440
    },
    {
      "epoch": 2.744761904761905,
      "grad_norm": 3.8255350611127605,
      "learning_rate": 1.8880734652461985e-07,
      "loss": 0.3892,
      "step": 1441
    },
    {
      "epoch": 2.7466666666666666,
      "grad_norm": 3.223193557047404,
      "learning_rate": 1.860173815013111e-07,
      "loss": 0.4634,
      "step": 1442
    },
    {
      "epoch": 2.7485714285714287,
      "grad_norm": 3.1081408487720728,
      "learning_rate": 1.8324779281130688e-07,
      "loss": 0.3539,
      "step": 1443
    },
    {
      "epoch": 2.7504761904761903,
      "grad_norm": 2.9818754324793275,
      "learning_rate": 1.8049859217755318e-07,
      "loss": 0.3294,
      "step": 1444
    },
    {
      "epoch": 2.7523809523809524,
      "grad_norm": 3.5304142104767005,
      "learning_rate": 1.7776979123670046e-07,
      "loss": 0.4255,
      "step": 1445
    },
    {
      "epoch": 2.7542857142857144,
      "grad_norm": 3.148823744356506,
      "learning_rate": 1.7506140153905172e-07,
      "loss": 0.3619,
      "step": 1446
    },
    {
      "epoch": 2.756190476190476,
      "grad_norm": 3.4208616640780805,
      "learning_rate": 1.723734345485134e-07,
      "loss": 0.4907,
      "step": 1447
    },
    {
      "epoch": 2.758095238095238,
      "grad_norm": 2.7842581607279127,
      "learning_rate": 1.6970590164255063e-07,
      "loss": 0.3696,
      "step": 1448
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.474949464107439,
      "learning_rate": 1.6705881411213266e-07,
      "loss": 0.3597,
      "step": 1449
    },
    {
      "epoch": 2.761904761904762,
      "grad_norm": 3.167833960819498,
      "learning_rate": 1.644321831616924e-07,
      "loss": 0.2895,
      "step": 1450
    },
    {
      "epoch": 2.763809523809524,
      "grad_norm": 4.186950007948689,
      "learning_rate": 1.618260199090732e-07,
      "loss": 0.5209,
      "step": 1451
    },
    {
      "epoch": 2.7657142857142856,
      "grad_norm": 3.9017836964297627,
      "learning_rate": 1.592403353854849e-07,
      "loss": 0.4623,
      "step": 1452
    },
    {
      "epoch": 2.7676190476190476,
      "grad_norm": 3.1034275458652743,
      "learning_rate": 1.5667514053545497e-07,
      "loss": 0.1982,
      "step": 1453
    },
    {
      "epoch": 2.7695238095238093,
      "grad_norm": 2.5896371066569643,
      "learning_rate": 1.5413044621678585e-07,
      "loss": 0.3693,
      "step": 1454
    },
    {
      "epoch": 2.7714285714285714,
      "grad_norm": 2.9423138679709706,
      "learning_rate": 1.5160626320050441e-07,
      "loss": 0.3612,
      "step": 1455
    },
    {
      "epoch": 2.7733333333333334,
      "grad_norm": 3.0077242605314054,
      "learning_rate": 1.4910260217081917e-07,
      "loss": 0.4115,
      "step": 1456
    },
    {
      "epoch": 2.7752380952380955,
      "grad_norm": 2.8821019011482774,
      "learning_rate": 1.4661947372507478e-07,
      "loss": 0.3463,
      "step": 1457
    },
    {
      "epoch": 2.777142857142857,
      "grad_norm": 2.791275624534033,
      "learning_rate": 1.4415688837370602e-07,
      "loss": 0.3416,
      "step": 1458
    },
    {
      "epoch": 2.779047619047619,
      "grad_norm": 2.8027088738155257,
      "learning_rate": 1.4171485654019545e-07,
      "loss": 0.3663,
      "step": 1459
    },
    {
      "epoch": 2.780952380952381,
      "grad_norm": 2.871962893395915,
      "learning_rate": 1.3929338856102648e-07,
      "loss": 0.2784,
      "step": 1460
    },
    {
      "epoch": 2.782857142857143,
      "grad_norm": 2.5917801560004246,
      "learning_rate": 1.3689249468564204e-07,
      "loss": 0.4239,
      "step": 1461
    },
    {
      "epoch": 2.784761904761905,
      "grad_norm": 3.0061953368794767,
      "learning_rate": 1.3451218507639919e-07,
      "loss": 0.351,
      "step": 1462
    },
    {
      "epoch": 2.7866666666666666,
      "grad_norm": 3.0760968984821844,
      "learning_rate": 1.321524698085297e-07,
      "loss": 0.3419,
      "step": 1463
    },
    {
      "epoch": 2.7885714285714287,
      "grad_norm": 4.888186011705799,
      "learning_rate": 1.2981335887009117e-07,
      "loss": 0.432,
      "step": 1464
    },
    {
      "epoch": 2.7904761904761903,
      "grad_norm": 3.2387190691494605,
      "learning_rate": 1.2749486216193096e-07,
      "loss": 0.2988,
      "step": 1465
    },
    {
      "epoch": 2.7923809523809524,
      "grad_norm": 2.471741045610758,
      "learning_rate": 1.2519698949764125e-07,
      "loss": 0.3658,
      "step": 1466
    },
    {
      "epoch": 2.7942857142857145,
      "grad_norm": 3.230907017656687,
      "learning_rate": 1.2291975060351845e-07,
      "loss": 0.2395,
      "step": 1467
    },
    {
      "epoch": 2.796190476190476,
      "grad_norm": 2.9442534154426894,
      "learning_rate": 1.2066315511852e-07,
      "loss": 0.3982,
      "step": 1468
    },
    {
      "epoch": 2.798095238095238,
      "grad_norm": 2.985865849547776,
      "learning_rate": 1.1842721259422762e-07,
      "loss": 0.4266,
      "step": 1469
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.4288167755474896,
      "learning_rate": 1.1621193249480134e-07,
      "loss": 0.3132,
      "step": 1470
    },
    {
      "epoch": 2.801904761904762,
      "grad_norm": 3.2452240068078386,
      "learning_rate": 1.14017324196945e-07,
      "loss": 0.4057,
      "step": 1471
    },
    {
      "epoch": 2.803809523809524,
      "grad_norm": 2.6775551540518143,
      "learning_rate": 1.1184339698986413e-07,
      "loss": 0.3664,
      "step": 1472
    },
    {
      "epoch": 2.8057142857142856,
      "grad_norm": 3.8951080894899546,
      "learning_rate": 1.0969016007522482e-07,
      "loss": 0.3875,
      "step": 1473
    },
    {
      "epoch": 2.8076190476190477,
      "grad_norm": 3.05342819085455,
      "learning_rate": 1.0755762256711821e-07,
      "loss": 0.3428,
      "step": 1474
    },
    {
      "epoch": 2.8095238095238093,
      "grad_norm": 2.830622206587528,
      "learning_rate": 1.0544579349201834e-07,
      "loss": 0.2576,
      "step": 1475
    },
    {
      "epoch": 2.8114285714285714,
      "grad_norm": 3.612325869758792,
      "learning_rate": 1.0335468178874875e-07,
      "loss": 0.4498,
      "step": 1476
    },
    {
      "epoch": 2.8133333333333335,
      "grad_norm": 2.251910206558631,
      "learning_rate": 1.0128429630843928e-07,
      "loss": 0.4282,
      "step": 1477
    },
    {
      "epoch": 2.815238095238095,
      "grad_norm": 2.3402149243566064,
      "learning_rate": 9.92346458144916e-08,
      "loss": 0.3128,
      "step": 1478
    },
    {
      "epoch": 2.817142857142857,
      "grad_norm": 3.0457760796060045,
      "learning_rate": 9.720573898254204e-08,
      "loss": 0.4435,
      "step": 1479
    },
    {
      "epoch": 2.819047619047619,
      "grad_norm": 3.2918956223155913,
      "learning_rate": 9.519758440042437e-08,
      "loss": 0.4222,
      "step": 1480
    },
    {
      "epoch": 2.820952380952381,
      "grad_norm": 4.799824062475176,
      "learning_rate": 9.321019056813319e-08,
      "loss": 0.3054,
      "step": 1481
    },
    {
      "epoch": 2.822857142857143,
      "grad_norm": 2.6879575793447126,
      "learning_rate": 9.124356589778894e-08,
      "loss": 0.432,
      "step": 1482
    },
    {
      "epoch": 2.824761904761905,
      "grad_norm": 3.7092648149545964,
      "learning_rate": 8.929771871360016e-08,
      "loss": 0.3283,
      "step": 1483
    },
    {
      "epoch": 2.8266666666666667,
      "grad_norm": 3.720362161985696,
      "learning_rate": 8.737265725183187e-08,
      "loss": 0.3941,
      "step": 1484
    },
    {
      "epoch": 2.8285714285714287,
      "grad_norm": 3.085587086580329,
      "learning_rate": 8.546838966076609e-08,
      "loss": 0.3258,
      "step": 1485
    },
    {
      "epoch": 2.8304761904761904,
      "grad_norm": 2.772409125190257,
      "learning_rate": 8.358492400067143e-08,
      "loss": 0.3166,
      "step": 1486
    },
    {
      "epoch": 2.8323809523809524,
      "grad_norm": 2.834852282659052,
      "learning_rate": 8.172226824376794e-08,
      "loss": 0.3456,
      "step": 1487
    },
    {
      "epoch": 2.8342857142857145,
      "grad_norm": 2.851141595395487,
      "learning_rate": 7.988043027419012e-08,
      "loss": 0.3203,
      "step": 1488
    },
    {
      "epoch": 2.836190476190476,
      "grad_norm": 3.1638348841918185,
      "learning_rate": 7.805941788795957e-08,
      "loss": 0.3634,
      "step": 1489
    },
    {
      "epoch": 2.8380952380952382,
      "grad_norm": 3.4988085681855936,
      "learning_rate": 7.625923879294562e-08,
      "loss": 0.3346,
      "step": 1490
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.932817774645209,
      "learning_rate": 7.447990060883813e-08,
      "loss": 0.4711,
      "step": 1491
    },
    {
      "epoch": 2.841904761904762,
      "grad_norm": 3.314782741051733,
      "learning_rate": 7.272141086711038e-08,
      "loss": 0.3805,
      "step": 1492
    },
    {
      "epoch": 2.843809523809524,
      "grad_norm": 2.6586503316878343,
      "learning_rate": 7.098377701099224e-08,
      "loss": 0.3988,
      "step": 1493
    },
    {
      "epoch": 2.8457142857142856,
      "grad_norm": 3.1274166789629567,
      "learning_rate": 6.926700639543426e-08,
      "loss": 0.2704,
      "step": 1494
    },
    {
      "epoch": 2.8476190476190477,
      "grad_norm": 4.335230393574005,
      "learning_rate": 6.757110628707875e-08,
      "loss": 0.3717,
      "step": 1495
    },
    {
      "epoch": 2.8495238095238093,
      "grad_norm": 3.9943236036288834,
      "learning_rate": 6.589608386422864e-08,
      "loss": 0.4777,
      "step": 1496
    },
    {
      "epoch": 2.8514285714285714,
      "grad_norm": 2.3847339579793982,
      "learning_rate": 6.424194621681757e-08,
      "loss": 0.338,
      "step": 1497
    },
    {
      "epoch": 2.8533333333333335,
      "grad_norm": 5.75275060131789,
      "learning_rate": 6.260870034637823e-08,
      "loss": 0.3713,
      "step": 1498
    },
    {
      "epoch": 2.855238095238095,
      "grad_norm": 2.6313275408387398,
      "learning_rate": 6.099635316601516e-08,
      "loss": 0.387,
      "step": 1499
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 2.6378410361597298,
      "learning_rate": 5.9404911500374195e-08,
      "loss": 0.3942,
      "step": 1500
    },
    {
      "epoch": 2.859047619047619,
      "grad_norm": 2.544117154140232,
      "learning_rate": 5.783438208561143e-08,
      "loss": 0.279,
      "step": 1501
    },
    {
      "epoch": 2.860952380952381,
      "grad_norm": 3.09851200759129,
      "learning_rate": 5.628477156936929e-08,
      "loss": 0.4811,
      "step": 1502
    },
    {
      "epoch": 2.862857142857143,
      "grad_norm": 3.3533025620016423,
      "learning_rate": 5.475608651074493e-08,
      "loss": 0.3734,
      "step": 1503
    },
    {
      "epoch": 2.8647619047619046,
      "grad_norm": 2.4971679525564454,
      "learning_rate": 5.324833338026414e-08,
      "loss": 0.2796,
      "step": 1504
    },
    {
      "epoch": 2.8666666666666667,
      "grad_norm": 2.8569636184413403,
      "learning_rate": 5.1761518559851895e-08,
      "loss": 0.396,
      "step": 1505
    },
    {
      "epoch": 2.8685714285714283,
      "grad_norm": 2.5957718140266084,
      "learning_rate": 5.029564834280854e-08,
      "loss": 0.3612,
      "step": 1506
    },
    {
      "epoch": 2.8704761904761904,
      "grad_norm": 2.5932946095313842,
      "learning_rate": 4.885072893377973e-08,
      "loss": 0.2873,
      "step": 1507
    },
    {
      "epoch": 2.8723809523809525,
      "grad_norm": 3.4386666381344955,
      "learning_rate": 4.742676644873323e-08,
      "loss": 0.3646,
      "step": 1508
    },
    {
      "epoch": 2.8742857142857146,
      "grad_norm": 2.7282812718294798,
      "learning_rate": 4.602376691493049e-08,
      "loss": 0.3819,
      "step": 1509
    },
    {
      "epoch": 2.876190476190476,
      "grad_norm": 3.502412241084784,
      "learning_rate": 4.46417362709034e-08,
      "loss": 0.4119,
      "step": 1510
    },
    {
      "epoch": 2.878095238095238,
      "grad_norm": 2.6490198698781304,
      "learning_rate": 4.328068036642652e-08,
      "loss": 0.3538,
      "step": 1511
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.5486078555380267,
      "learning_rate": 4.1940604962495414e-08,
      "loss": 0.4733,
      "step": 1512
    },
    {
      "epoch": 2.881904761904762,
      "grad_norm": 2.876476371352215,
      "learning_rate": 4.062151573129891e-08,
      "loss": 0.4285,
      "step": 1513
    },
    {
      "epoch": 2.883809523809524,
      "grad_norm": 2.9246670664314123,
      "learning_rate": 3.932341825619912e-08,
      "loss": 0.4738,
      "step": 1514
    },
    {
      "epoch": 2.8857142857142857,
      "grad_norm": 5.16920512330396,
      "learning_rate": 3.804631803170311e-08,
      "loss": 0.4935,
      "step": 1515
    },
    {
      "epoch": 2.8876190476190478,
      "grad_norm": 3.555521994274648,
      "learning_rate": 3.6790220463443494e-08,
      "loss": 0.4549,
      "step": 1516
    },
    {
      "epoch": 2.8895238095238094,
      "grad_norm": 3.7071302903675396,
      "learning_rate": 3.555513086815398e-08,
      "loss": 0.4049,
      "step": 1517
    },
    {
      "epoch": 2.8914285714285715,
      "grad_norm": 3.9377888354070283,
      "learning_rate": 3.434105447364666e-08,
      "loss": 0.4843,
      "step": 1518
    },
    {
      "epoch": 2.8933333333333335,
      "grad_norm": 3.6947499162878907,
      "learning_rate": 3.314799641879029e-08,
      "loss": 0.4104,
      "step": 1519
    },
    {
      "epoch": 2.895238095238095,
      "grad_norm": 2.5344718134461997,
      "learning_rate": 3.197596175348872e-08,
      "loss": 0.3859,
      "step": 1520
    },
    {
      "epoch": 2.8971428571428572,
      "grad_norm": 2.968032795769075,
      "learning_rate": 3.0824955438659155e-08,
      "loss": 0.6132,
      "step": 1521
    },
    {
      "epoch": 2.899047619047619,
      "grad_norm": 3.4079549983296635,
      "learning_rate": 2.9694982346211155e-08,
      "loss": 0.3593,
      "step": 1522
    },
    {
      "epoch": 2.900952380952381,
      "grad_norm": 3.5174081383743983,
      "learning_rate": 2.8586047259026027e-08,
      "loss": 0.4305,
      "step": 1523
    },
    {
      "epoch": 2.902857142857143,
      "grad_norm": 2.823236769842876,
      "learning_rate": 2.7498154870936877e-08,
      "loss": 0.3946,
      "step": 1524
    },
    {
      "epoch": 2.9047619047619047,
      "grad_norm": 3.301918701278776,
      "learning_rate": 2.643130978670916e-08,
      "loss": 0.3012,
      "step": 1525
    },
    {
      "epoch": 2.9066666666666667,
      "grad_norm": 4.008765745850195,
      "learning_rate": 2.5385516522019038e-08,
      "loss": 0.3526,
      "step": 1526
    },
    {
      "epoch": 2.9085714285714284,
      "grad_norm": 2.647771121600037,
      "learning_rate": 2.4360779503436737e-08,
      "loss": 0.3276,
      "step": 1527
    },
    {
      "epoch": 2.9104761904761904,
      "grad_norm": 2.8371789333816144,
      "learning_rate": 2.3357103068408216e-08,
      "loss": 0.2851,
      "step": 1528
    },
    {
      "epoch": 2.9123809523809525,
      "grad_norm": 2.9382849895020655,
      "learning_rate": 2.2374491465232962e-08,
      "loss": 0.3664,
      "step": 1529
    },
    {
      "epoch": 2.914285714285714,
      "grad_norm": 3.877471654577267,
      "learning_rate": 2.1412948853050118e-08,
      "loss": 0.3694,
      "step": 1530
    },
    {
      "epoch": 2.916190476190476,
      "grad_norm": 3.0412428757357093,
      "learning_rate": 2.047247930181795e-08,
      "loss": 0.313,
      "step": 1531
    },
    {
      "epoch": 2.918095238095238,
      "grad_norm": 2.610434076947357,
      "learning_rate": 1.95530867923005e-08,
      "loss": 0.3205,
      "step": 1532
    },
    {
      "epoch": 2.92,
      "grad_norm": 2.872911037706269,
      "learning_rate": 1.8654775216045416e-08,
      "loss": 0.422,
      "step": 1533
    },
    {
      "epoch": 2.921904761904762,
      "grad_norm": 2.4915798889589142,
      "learning_rate": 1.777754837537171e-08,
      "loss": 0.4449,
      "step": 1534
    },
    {
      "epoch": 2.923809523809524,
      "grad_norm": 4.022478647232431,
      "learning_rate": 1.6921409983352012e-08,
      "loss": 0.3899,
      "step": 1535
    },
    {
      "epoch": 2.9257142857142857,
      "grad_norm": 2.493714838512802,
      "learning_rate": 1.6086363663797012e-08,
      "loss": 0.2982,
      "step": 1536
    },
    {
      "epoch": 2.9276190476190473,
      "grad_norm": 2.9697157354072816,
      "learning_rate": 1.527241295123938e-08,
      "loss": 0.3651,
      "step": 1537
    },
    {
      "epoch": 2.9295238095238094,
      "grad_norm": 2.7197768930952306,
      "learning_rate": 1.447956129092043e-08,
      "loss": 0.3514,
      "step": 1538
    },
    {
      "epoch": 2.9314285714285715,
      "grad_norm": 3.5781090251386924,
      "learning_rate": 1.3707812038775692e-08,
      "loss": 0.5047,
      "step": 1539
    },
    {
      "epoch": 2.9333333333333336,
      "grad_norm": 3.2730755264424953,
      "learning_rate": 1.2957168461416591e-08,
      "loss": 0.4054,
      "step": 1540
    },
    {
      "epoch": 2.935238095238095,
      "grad_norm": 2.528277959234961,
      "learning_rate": 1.2227633736123235e-08,
      "loss": 0.4211,
      "step": 1541
    },
    {
      "epoch": 2.9371428571428573,
      "grad_norm": 3.4226473693644284,
      "learning_rate": 1.1519210950825532e-08,
      "loss": 0.4452,
      "step": 1542
    },
    {
      "epoch": 2.939047619047619,
      "grad_norm": 2.6766217811106596,
      "learning_rate": 1.08319031040921e-08,
      "loss": 0.4472,
      "step": 1543
    },
    {
      "epoch": 2.940952380952381,
      "grad_norm": 3.4616430384994206,
      "learning_rate": 1.0165713105118047e-08,
      "loss": 0.4327,
      "step": 1544
    },
    {
      "epoch": 2.942857142857143,
      "grad_norm": 4.305382509125797,
      "learning_rate": 9.520643773712202e-09,
      "loss": 0.3988,
      "step": 1545
    },
    {
      "epoch": 2.9447619047619047,
      "grad_norm": 2.896555187828505,
      "learning_rate": 8.896697840284352e-09,
      "loss": 0.3547,
      "step": 1546
    },
    {
      "epoch": 2.9466666666666668,
      "grad_norm": 4.480592353258285,
      "learning_rate": 8.293877945835804e-09,
      "loss": 0.3365,
      "step": 1547
    },
    {
      "epoch": 2.9485714285714284,
      "grad_norm": 3.5724025531135903,
      "learning_rate": 7.712186641944952e-09,
      "loss": 0.3971,
      "step": 1548
    },
    {
      "epoch": 2.9504761904761905,
      "grad_norm": 3.3096316069641993,
      "learning_rate": 7.151626390760058e-09,
      "loss": 0.3749,
      "step": 1549
    },
    {
      "epoch": 2.9523809523809526,
      "grad_norm": 2.559914247262096,
      "learning_rate": 6.6121995649859326e-09,
      "loss": 0.3817,
      "step": 1550
    },
    {
      "epoch": 2.954285714285714,
      "grad_norm": 2.997375555550545,
      "learning_rate": 6.093908447876162e-09,
      "loss": 0.382,
      "step": 1551
    },
    {
      "epoch": 2.9561904761904763,
      "grad_norm": 3.108071839021827,
      "learning_rate": 5.596755233222561e-09,
      "loss": 0.4033,
      "step": 1552
    },
    {
      "epoch": 2.958095238095238,
      "grad_norm": 2.300321365538265,
      "learning_rate": 5.120742025343517e-09,
      "loss": 0.4264,
      "step": 1553
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.6958337209009158,
      "learning_rate": 4.6658708390801e-09,
      "loss": 0.4062,
      "step": 1554
    },
    {
      "epoch": 2.961904761904762,
      "grad_norm": 3.5919435563455657,
      "learning_rate": 4.232143599782746e-09,
      "loss": 0.4033,
      "step": 1555
    },
    {
      "epoch": 2.9638095238095237,
      "grad_norm": 2.6611038770764903,
      "learning_rate": 3.819562143306255e-09,
      "loss": 0.3345,
      "step": 1556
    },
    {
      "epoch": 2.9657142857142857,
      "grad_norm": 3.4092163151172388,
      "learning_rate": 3.4281282160003594e-09,
      "loss": 0.4142,
      "step": 1557
    },
    {
      "epoch": 2.9676190476190474,
      "grad_norm": 3.6347390497508036,
      "learning_rate": 3.0578434747025042e-09,
      "loss": 0.4654,
      "step": 1558
    },
    {
      "epoch": 2.9695238095238095,
      "grad_norm": 2.4751005519357836,
      "learning_rate": 2.7087094867311868e-09,
      "loss": 0.4332,
      "step": 1559
    },
    {
      "epoch": 2.9714285714285715,
      "grad_norm": 3.022900699813934,
      "learning_rate": 2.3807277298804053e-09,
      "loss": 0.3647,
      "step": 1560
    },
    {
      "epoch": 2.9733333333333336,
      "grad_norm": 3.746902190262692,
      "learning_rate": 2.0738995924107774e-09,
      "loss": 0.4073,
      "step": 1561
    },
    {
      "epoch": 2.9752380952380952,
      "grad_norm": 3.1209308325275513,
      "learning_rate": 1.7882263730462091e-09,
      "loss": 0.2681,
      "step": 1562
    },
    {
      "epoch": 2.977142857142857,
      "grad_norm": 4.6229541934062,
      "learning_rate": 1.5237092809666786e-09,
      "loss": 0.4218,
      "step": 1563
    },
    {
      "epoch": 2.979047619047619,
      "grad_norm": 2.461331500764114,
      "learning_rate": 1.280349435803796e-09,
      "loss": 0.4157,
      "step": 1564
    },
    {
      "epoch": 2.980952380952381,
      "grad_norm": 4.0826884736518325,
      "learning_rate": 1.0581478676369161e-09,
      "loss": 0.4237,
      "step": 1565
    },
    {
      "epoch": 2.982857142857143,
      "grad_norm": 2.775846320928122,
      "learning_rate": 8.571055169870334e-10,
      "loss": 0.4322,
      "step": 1566
    },
    {
      "epoch": 2.9847619047619047,
      "grad_norm": 4.287841206088843,
      "learning_rate": 6.772232348140062e-10,
      "loss": 0.4391,
      "step": 1567
    },
    {
      "epoch": 2.986666666666667,
      "grad_norm": 3.237023079895034,
      "learning_rate": 5.185017825126704e-10,
      "loss": 0.4706,
      "step": 1568
    },
    {
      "epoch": 2.9885714285714284,
      "grad_norm": 3.029608142517041,
      "learning_rate": 3.809418319095093e-10,
      "loss": 0.4228,
      "step": 1569
    },
    {
      "epoch": 2.9904761904761905,
      "grad_norm": 4.206662515966221,
      "learning_rate": 2.64543965259878e-10,
      "loss": 0.628,
      "step": 1570
    },
    {
      "epoch": 2.9923809523809526,
      "grad_norm": 2.9028948979191402,
      "learning_rate": 1.6930867524633798e-10,
      "loss": 0.4524,
      "step": 1571
    },
    {
      "epoch": 2.994285714285714,
      "grad_norm": 2.7800671548159284,
      "learning_rate": 9.523636497366095e-11,
      "loss": 0.4379,
      "step": 1572
    },
    {
      "epoch": 2.9961904761904763,
      "grad_norm": 4.067541717640019,
      "learning_rate": 4.232734797215976e-11,
      "loss": 0.4566,
      "step": 1573
    },
    {
      "epoch": 2.998095238095238,
      "grad_norm": 2.445989675628037,
      "learning_rate": 1.0581848190471811e-11,
      "loss": 0.3629,
      "step": 1574
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.71914974410319,
      "learning_rate": 0.0,
      "loss": 0.4207,
      "step": 1575
    },
    {
      "epoch": 3.0,
      "step": 1575,
      "total_flos": 247329718272000.0,
      "train_loss": 0.7211369104044778,
      "train_runtime": 32331.6664,
      "train_samples_per_second": 2.337,
      "train_steps_per_second": 0.049
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1575,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100000,
  "total_flos": 247329718272000.0,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}