{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 2164,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0009242144177449168,
      "grad_norm": 61.5,
      "learning_rate": 1.8348623853211012e-07,
      "loss": 3.1195,
      "step": 1
    },
    {
      "epoch": 0.0018484288354898336,
      "grad_norm": 61.5,
      "learning_rate": 3.6697247706422023e-07,
      "loss": 3.2083,
      "step": 2
    },
    {
      "epoch": 0.0027726432532347504,
      "grad_norm": 62.75,
      "learning_rate": 5.504587155963304e-07,
      "loss": 3.1632,
      "step": 3
    },
    {
      "epoch": 0.0036968576709796672,
      "grad_norm": 61.25,
      "learning_rate": 7.339449541284405e-07,
      "loss": 3.0885,
      "step": 4
    },
    {
      "epoch": 0.0046210720887245845,
      "grad_norm": 63.5,
      "learning_rate": 9.174311926605506e-07,
      "loss": 3.1428,
      "step": 5
    },
    {
      "epoch": 0.005545286506469501,
      "grad_norm": 57.75,
      "learning_rate": 1.1009174311926608e-06,
      "loss": 3.0217,
      "step": 6
    },
    {
      "epoch": 0.006469500924214418,
      "grad_norm": 56.75,
      "learning_rate": 1.2844036697247707e-06,
      "loss": 3.0014,
      "step": 7
    },
    {
      "epoch": 0.0073937153419593345,
      "grad_norm": 57.75,
      "learning_rate": 1.467889908256881e-06,
      "loss": 2.9919,
      "step": 8
    },
    {
      "epoch": 0.00831792975970425,
      "grad_norm": 56.5,
      "learning_rate": 1.6513761467889911e-06,
      "loss": 2.9574,
      "step": 9
    },
    {
      "epoch": 0.009242144177449169,
      "grad_norm": 56.75,
      "learning_rate": 1.8348623853211011e-06,
      "loss": 2.971,
      "step": 10
    },
    {
      "epoch": 0.010166358595194085,
      "grad_norm": 62.0,
      "learning_rate": 2.0183486238532113e-06,
      "loss": 3.1683,
      "step": 11
    },
    {
      "epoch": 0.011090573012939002,
      "grad_norm": 55.75,
      "learning_rate": 2.2018348623853215e-06,
      "loss": 2.9952,
      "step": 12
    },
    {
      "epoch": 0.012014787430683918,
      "grad_norm": 54.75,
      "learning_rate": 2.3853211009174317e-06,
      "loss": 2.9961,
      "step": 13
    },
    {
      "epoch": 0.012939001848428836,
      "grad_norm": 54.0,
      "learning_rate": 2.5688073394495415e-06,
      "loss": 2.9625,
      "step": 14
    },
    {
      "epoch": 0.013863216266173753,
      "grad_norm": 55.5,
      "learning_rate": 2.7522935779816517e-06,
      "loss": 3.0311,
      "step": 15
    },
    {
      "epoch": 0.014787430683918669,
      "grad_norm": 45.5,
      "learning_rate": 2.935779816513762e-06,
      "loss": 2.8209,
      "step": 16
    },
    {
      "epoch": 0.015711645101663587,
      "grad_norm": 46.75,
      "learning_rate": 3.119266055045872e-06,
      "loss": 2.8857,
      "step": 17
    },
    {
      "epoch": 0.0166358595194085,
      "grad_norm": 40.5,
      "learning_rate": 3.3027522935779823e-06,
      "loss": 2.7243,
      "step": 18
    },
    {
      "epoch": 0.01756007393715342,
      "grad_norm": 37.0,
      "learning_rate": 3.486238532110092e-06,
      "loss": 2.605,
      "step": 19
    },
    {
      "epoch": 0.018484288354898338,
      "grad_norm": 36.0,
      "learning_rate": 3.6697247706422022e-06,
      "loss": 2.6036,
      "step": 20
    },
    {
      "epoch": 0.019408502772643253,
      "grad_norm": 39.0,
      "learning_rate": 3.853211009174313e-06,
      "loss": 2.6174,
      "step": 21
    },
    {
      "epoch": 0.02033271719038817,
      "grad_norm": 26.125,
      "learning_rate": 4.036697247706423e-06,
      "loss": 2.4368,
      "step": 22
    },
    {
      "epoch": 0.021256931608133085,
      "grad_norm": 26.75,
      "learning_rate": 4.220183486238532e-06,
      "loss": 2.4722,
      "step": 23
    },
    {
      "epoch": 0.022181146025878003,
      "grad_norm": 22.75,
      "learning_rate": 4.403669724770643e-06,
      "loss": 2.4402,
      "step": 24
    },
    {
      "epoch": 0.02310536044362292,
      "grad_norm": 18.875,
      "learning_rate": 4.587155963302753e-06,
      "loss": 2.458,
      "step": 25
    },
    {
      "epoch": 0.024029574861367836,
      "grad_norm": 18.625,
      "learning_rate": 4.770642201834863e-06,
      "loss": 2.332,
      "step": 26
    },
    {
      "epoch": 0.024953789279112754,
      "grad_norm": 13.1875,
      "learning_rate": 4.954128440366973e-06,
      "loss": 2.2852,
      "step": 27
    },
    {
      "epoch": 0.025878003696857672,
      "grad_norm": 11.9375,
      "learning_rate": 5.137614678899083e-06,
      "loss": 2.3154,
      "step": 28
    },
    {
      "epoch": 0.026802218114602587,
      "grad_norm": 10.5625,
      "learning_rate": 5.3211009174311936e-06,
      "loss": 2.3205,
      "step": 29
    },
    {
      "epoch": 0.027726432532347505,
      "grad_norm": 8.5,
      "learning_rate": 5.504587155963303e-06,
      "loss": 2.3319,
      "step": 30
    },
    {
      "epoch": 0.02865064695009242,
      "grad_norm": 8.875,
      "learning_rate": 5.688073394495414e-06,
      "loss": 2.2993,
      "step": 31
    },
    {
      "epoch": 0.029574861367837338,
      "grad_norm": 7.6875,
      "learning_rate": 5.871559633027524e-06,
      "loss": 2.3397,
      "step": 32
    },
    {
      "epoch": 0.030499075785582256,
      "grad_norm": 7.03125,
      "learning_rate": 6.0550458715596335e-06,
      "loss": 2.1804,
      "step": 33
    },
    {
      "epoch": 0.031423290203327174,
      "grad_norm": 11.625,
      "learning_rate": 6.238532110091744e-06,
      "loss": 2.1016,
      "step": 34
    },
    {
      "epoch": 0.03234750462107209,
      "grad_norm": 18.875,
      "learning_rate": 6.422018348623854e-06,
      "loss": 2.1097,
      "step": 35
    },
    {
      "epoch": 0.033271719038817,
      "grad_norm": 5.6875,
      "learning_rate": 6.6055045871559645e-06,
      "loss": 2.0425,
      "step": 36
    },
    {
      "epoch": 0.034195933456561925,
      "grad_norm": 4.9375,
      "learning_rate": 6.788990825688074e-06,
      "loss": 2.1982,
      "step": 37
    },
    {
      "epoch": 0.03512014787430684,
      "grad_norm": 5.09375,
      "learning_rate": 6.972477064220184e-06,
      "loss": 2.1257,
      "step": 38
    },
    {
      "epoch": 0.036044362292051754,
      "grad_norm": 4.5625,
      "learning_rate": 7.155963302752295e-06,
      "loss": 2.1028,
      "step": 39
    },
    {
      "epoch": 0.036968576709796676,
      "grad_norm": 4.40625,
      "learning_rate": 7.3394495412844045e-06,
      "loss": 2.0811,
      "step": 40
    },
    {
      "epoch": 0.03789279112754159,
      "grad_norm": 6.09375,
      "learning_rate": 7.522935779816515e-06,
      "loss": 2.1204,
      "step": 41
    },
    {
      "epoch": 0.038817005545286505,
      "grad_norm": 4.09375,
      "learning_rate": 7.706422018348626e-06,
      "loss": 2.0407,
      "step": 42
    },
    {
      "epoch": 0.03974121996303143,
      "grad_norm": 3.875,
      "learning_rate": 7.889908256880735e-06,
      "loss": 2.0729,
      "step": 43
    },
    {
      "epoch": 0.04066543438077634,
      "grad_norm": 4.1875,
      "learning_rate": 8.073394495412845e-06,
      "loss": 2.0598,
      "step": 44
    },
    {
      "epoch": 0.041589648798521256,
      "grad_norm": 3.609375,
      "learning_rate": 8.256880733944956e-06,
      "loss": 2.0645,
      "step": 45
    },
    {
      "epoch": 0.04251386321626617,
      "grad_norm": 3.390625,
      "learning_rate": 8.440366972477065e-06,
      "loss": 2.0481,
      "step": 46
    },
    {
      "epoch": 0.04343807763401109,
      "grad_norm": 3.5,
      "learning_rate": 8.623853211009175e-06,
      "loss": 2.0675,
      "step": 47
    },
    {
      "epoch": 0.04436229205175601,
      "grad_norm": 3.234375,
      "learning_rate": 8.807339449541286e-06,
      "loss": 2.0347,
      "step": 48
    },
    {
      "epoch": 0.04528650646950092,
      "grad_norm": 3.640625,
      "learning_rate": 8.990825688073395e-06,
      "loss": 2.024,
      "step": 49
    },
    {
      "epoch": 0.04621072088724584,
      "grad_norm": 3.359375,
      "learning_rate": 9.174311926605506e-06,
      "loss": 1.9933,
      "step": 50
    },
    {
      "epoch": 0.04713493530499076,
      "grad_norm": 3.171875,
      "learning_rate": 9.357798165137616e-06,
      "loss": 1.9816,
      "step": 51
    },
    {
      "epoch": 0.04805914972273567,
      "grad_norm": 4.65625,
      "learning_rate": 9.541284403669727e-06,
      "loss": 2.0251,
      "step": 52
    },
    {
      "epoch": 0.048983364140480594,
      "grad_norm": 3.171875,
      "learning_rate": 9.724770642201836e-06,
      "loss": 2.1156,
      "step": 53
    },
    {
      "epoch": 0.04990757855822551,
      "grad_norm": 3.109375,
      "learning_rate": 9.908256880733946e-06,
      "loss": 2.0297,
      "step": 54
    },
    {
      "epoch": 0.05083179297597042,
      "grad_norm": 3.078125,
      "learning_rate": 1.0091743119266055e-05,
      "loss": 1.9833,
      "step": 55
    },
    {
      "epoch": 0.051756007393715345,
      "grad_norm": 3.078125,
      "learning_rate": 1.0275229357798166e-05,
      "loss": 2.0087,
      "step": 56
    },
    {
      "epoch": 0.05268022181146026,
      "grad_norm": 2.96875,
      "learning_rate": 1.0458715596330275e-05,
      "loss": 2.0128,
      "step": 57
    },
    {
      "epoch": 0.053604436229205174,
      "grad_norm": 3.0625,
      "learning_rate": 1.0642201834862387e-05,
      "loss": 1.9904,
      "step": 58
    },
    {
      "epoch": 0.054528650646950096,
      "grad_norm": 3.078125,
      "learning_rate": 1.0825688073394496e-05,
      "loss": 2.0955,
      "step": 59
    },
    {
      "epoch": 0.05545286506469501,
      "grad_norm": 3.046875,
      "learning_rate": 1.1009174311926607e-05,
      "loss": 2.023,
      "step": 60
    },
    {
      "epoch": 0.056377079482439925,
      "grad_norm": 3.046875,
      "learning_rate": 1.1192660550458716e-05,
      "loss": 2.0217,
      "step": 61
    },
    {
      "epoch": 0.05730129390018484,
      "grad_norm": 2.96875,
      "learning_rate": 1.1376146788990828e-05,
      "loss": 1.9281,
      "step": 62
    },
    {
      "epoch": 0.05822550831792976,
      "grad_norm": 2.9375,
      "learning_rate": 1.1559633027522937e-05,
      "loss": 2.0056,
      "step": 63
    },
    {
      "epoch": 0.059149722735674676,
      "grad_norm": 3.046875,
      "learning_rate": 1.1743119266055047e-05,
      "loss": 2.021,
      "step": 64
    },
    {
      "epoch": 0.06007393715341959,
      "grad_norm": 3.0,
      "learning_rate": 1.1926605504587156e-05,
      "loss": 1.9854,
      "step": 65
    },
    {
      "epoch": 0.06099815157116451,
      "grad_norm": 3.109375,
      "learning_rate": 1.2110091743119267e-05,
      "loss": 2.0083,
      "step": 66
    },
    {
      "epoch": 0.06192236598890943,
      "grad_norm": 2.953125,
      "learning_rate": 1.2293577981651376e-05,
      "loss": 1.9579,
      "step": 67
    },
    {
      "epoch": 0.06284658040665435,
      "grad_norm": 2.984375,
      "learning_rate": 1.2477064220183488e-05,
      "loss": 1.9632,
      "step": 68
    },
    {
      "epoch": 0.06377079482439926,
      "grad_norm": 3.125,
      "learning_rate": 1.2660550458715597e-05,
      "loss": 1.9839,
      "step": 69
    },
    {
      "epoch": 0.06469500924214418,
      "grad_norm": 3.03125,
      "learning_rate": 1.2844036697247708e-05,
      "loss": 2.0081,
      "step": 70
    },
    {
      "epoch": 0.06561922365988909,
      "grad_norm": 3.0,
      "learning_rate": 1.3027522935779817e-05,
      "loss": 1.8978,
      "step": 71
    },
    {
      "epoch": 0.066543438077634,
      "grad_norm": 2.984375,
      "learning_rate": 1.3211009174311929e-05,
      "loss": 1.9977,
      "step": 72
    },
    {
      "epoch": 0.06746765249537892,
      "grad_norm": 2.953125,
      "learning_rate": 1.3394495412844038e-05,
      "loss": 1.9309,
      "step": 73
    },
    {
      "epoch": 0.06839186691312385,
      "grad_norm": 3.09375,
      "learning_rate": 1.3577981651376149e-05,
      "loss": 1.9731,
      "step": 74
    },
    {
      "epoch": 0.06931608133086876,
      "grad_norm": 2.953125,
      "learning_rate": 1.3761467889908258e-05,
      "loss": 1.8775,
      "step": 75
    },
    {
      "epoch": 0.07024029574861368,
      "grad_norm": 3.125,
      "learning_rate": 1.3944954128440368e-05,
      "loss": 1.884,
      "step": 76
    },
    {
      "epoch": 0.0711645101663586,
      "grad_norm": 3.09375,
      "learning_rate": 1.4128440366972477e-05,
      "loss": 1.9861,
      "step": 77
    },
    {
      "epoch": 0.07208872458410351,
      "grad_norm": 3.1875,
      "learning_rate": 1.431192660550459e-05,
      "loss": 1.9244,
      "step": 78
    },
    {
      "epoch": 0.07301293900184842,
      "grad_norm": 2.953125,
      "learning_rate": 1.4495412844036698e-05,
      "loss": 1.9323,
      "step": 79
    },
    {
      "epoch": 0.07393715341959335,
      "grad_norm": 3.015625,
      "learning_rate": 1.4678899082568809e-05,
      "loss": 1.887,
      "step": 80
    },
    {
      "epoch": 0.07486136783733827,
      "grad_norm": 3.09375,
      "learning_rate": 1.4862385321100918e-05,
      "loss": 1.928,
      "step": 81
    },
    {
      "epoch": 0.07578558225508318,
      "grad_norm": 3.203125,
      "learning_rate": 1.504587155963303e-05,
      "loss": 1.8454,
      "step": 82
    },
    {
      "epoch": 0.0767097966728281,
      "grad_norm": 3.03125,
      "learning_rate": 1.5229357798165139e-05,
      "loss": 1.9323,
      "step": 83
    },
    {
      "epoch": 0.07763401109057301,
      "grad_norm": 2.953125,
      "learning_rate": 1.541284403669725e-05,
      "loss": 1.8914,
      "step": 84
    },
    {
      "epoch": 0.07855822550831792,
      "grad_norm": 2.90625,
      "learning_rate": 1.559633027522936e-05,
      "loss": 1.7869,
      "step": 85
    },
    {
      "epoch": 0.07948243992606285,
      "grad_norm": 2.90625,
      "learning_rate": 1.577981651376147e-05,
      "loss": 1.9624,
      "step": 86
    },
    {
      "epoch": 0.08040665434380777,
      "grad_norm": 2.78125,
      "learning_rate": 1.5963302752293578e-05,
      "loss": 1.9335,
      "step": 87
    },
    {
      "epoch": 0.08133086876155268,
      "grad_norm": 2.71875,
      "learning_rate": 1.614678899082569e-05,
      "loss": 1.8357,
      "step": 88
    },
    {
      "epoch": 0.0822550831792976,
      "grad_norm": 2.921875,
      "learning_rate": 1.63302752293578e-05,
      "loss": 1.8874,
      "step": 89
    },
    {
      "epoch": 0.08317929759704251,
      "grad_norm": 2.765625,
      "learning_rate": 1.6513761467889912e-05,
      "loss": 1.9741,
      "step": 90
    },
    {
      "epoch": 0.08410351201478743,
      "grad_norm": 2.65625,
      "learning_rate": 1.669724770642202e-05,
      "loss": 1.8833,
      "step": 91
    },
    {
      "epoch": 0.08502772643253234,
      "grad_norm": 2.96875,
      "learning_rate": 1.688073394495413e-05,
      "loss": 1.8588,
      "step": 92
    },
    {
      "epoch": 0.08595194085027727,
      "grad_norm": 2.703125,
      "learning_rate": 1.706422018348624e-05,
      "loss": 1.9266,
      "step": 93
    },
    {
      "epoch": 0.08687615526802218,
      "grad_norm": 2.578125,
      "learning_rate": 1.724770642201835e-05,
      "loss": 1.8432,
      "step": 94
    },
    {
      "epoch": 0.0878003696857671,
      "grad_norm": 2.9375,
      "learning_rate": 1.743119266055046e-05,
      "loss": 1.9043,
      "step": 95
    },
    {
      "epoch": 0.08872458410351201,
      "grad_norm": 2.546875,
      "learning_rate": 1.7614678899082572e-05,
      "loss": 1.8099,
      "step": 96
    },
    {
      "epoch": 0.08964879852125693,
      "grad_norm": 2.53125,
      "learning_rate": 1.779816513761468e-05,
      "loss": 1.8868,
      "step": 97
    },
    {
      "epoch": 0.09057301293900184,
      "grad_norm": 2.828125,
      "learning_rate": 1.798165137614679e-05,
      "loss": 1.8583,
      "step": 98
    },
    {
      "epoch": 0.09149722735674677,
      "grad_norm": 2.734375,
      "learning_rate": 1.81651376146789e-05,
      "loss": 1.881,
      "step": 99
    },
    {
      "epoch": 0.09242144177449169,
      "grad_norm": 2.671875,
      "learning_rate": 1.834862385321101e-05,
      "loss": 1.8457,
      "step": 100
    },
    {
      "epoch": 0.0933456561922366,
      "grad_norm": 2.65625,
      "learning_rate": 1.853211009174312e-05,
      "loss": 1.8356,
      "step": 101
    },
    {
      "epoch": 0.09426987060998152,
      "grad_norm": 2.546875,
      "learning_rate": 1.8715596330275232e-05,
      "loss": 1.8369,
      "step": 102
    },
    {
      "epoch": 0.09519408502772643,
      "grad_norm": 2.703125,
      "learning_rate": 1.889908256880734e-05,
      "loss": 1.9394,
      "step": 103
    },
    {
      "epoch": 0.09611829944547134,
      "grad_norm": 2.453125,
      "learning_rate": 1.9082568807339454e-05,
      "loss": 1.8417,
      "step": 104
    },
    {
      "epoch": 0.09704251386321626,
      "grad_norm": 2.65625,
      "learning_rate": 1.9266055045871563e-05,
      "loss": 1.8609,
      "step": 105
    },
    {
      "epoch": 0.09796672828096119,
      "grad_norm": 2.53125,
      "learning_rate": 1.944954128440367e-05,
      "loss": 1.8663,
      "step": 106
    },
    {
      "epoch": 0.0988909426987061,
      "grad_norm": 2.65625,
      "learning_rate": 1.963302752293578e-05,
      "loss": 1.8382,
      "step": 107
    },
    {
      "epoch": 0.09981515711645102,
      "grad_norm": 2.625,
      "learning_rate": 1.9816513761467893e-05,
      "loss": 1.854,
      "step": 108
    },
    {
      "epoch": 0.10073937153419593,
      "grad_norm": 2.515625,
      "learning_rate": 2e-05,
      "loss": 1.7959,
      "step": 109
    },
    {
      "epoch": 0.10166358595194085,
      "grad_norm": 2.640625,
      "learning_rate": 1.999998831453463e-05,
      "loss": 1.8384,
      "step": 110
    },
    {
      "epoch": 0.10258780036968576,
      "grad_norm": 2.625,
      "learning_rate": 1.9999953258165833e-05,
      "loss": 1.8335,
      "step": 111
    },
    {
      "epoch": 0.10351201478743069,
      "grad_norm": 2.578125,
      "learning_rate": 1.9999894830975532e-05,
      "loss": 1.8598,
      "step": 112
    },
    {
      "epoch": 0.1044362292051756,
      "grad_norm": 2.546875,
      "learning_rate": 1.999981303310028e-05,
      "loss": 1.7008,
      "step": 113
    },
    {
      "epoch": 0.10536044362292052,
      "grad_norm": 2.53125,
      "learning_rate": 1.9999707864731247e-05,
      "loss": 1.8535,
      "step": 114
    },
    {
      "epoch": 0.10628465804066543,
      "grad_norm": 2.515625,
      "learning_rate": 1.9999579326114224e-05,
      "loss": 1.8864,
      "step": 115
    },
    {
      "epoch": 0.10720887245841035,
      "grad_norm": 2.625,
      "learning_rate": 1.9999427417549612e-05,
      "loss": 1.8675,
      "step": 116
    },
    {
      "epoch": 0.10813308687615526,
      "grad_norm": 2.609375,
      "learning_rate": 1.999925213939244e-05,
      "loss": 1.8583,
      "step": 117
    },
    {
      "epoch": 0.10905730129390019,
      "grad_norm": 2.46875,
      "learning_rate": 1.9999053492052347e-05,
      "loss": 1.8889,
      "step": 118
    },
    {
      "epoch": 0.1099815157116451,
      "grad_norm": 2.765625,
      "learning_rate": 1.999883147599359e-05,
      "loss": 1.8804,
      "step": 119
    },
    {
      "epoch": 0.11090573012939002,
      "grad_norm": 2.609375,
      "learning_rate": 1.9998586091735044e-05,
      "loss": 1.8546,
      "step": 120
    },
    {
      "epoch": 0.11182994454713494,
      "grad_norm": 2.609375,
      "learning_rate": 1.9998317339850192e-05,
      "loss": 1.8745,
      "step": 121
    },
    {
      "epoch": 0.11275415896487985,
      "grad_norm": 2.65625,
      "learning_rate": 1.9998025220967134e-05,
      "loss": 1.7908,
      "step": 122
    },
    {
      "epoch": 0.11367837338262476,
      "grad_norm": 2.546875,
      "learning_rate": 1.9997709735768578e-05,
      "loss": 1.8669,
      "step": 123
    },
    {
      "epoch": 0.11460258780036968,
      "grad_norm": 2.5,
      "learning_rate": 1.9997370884991842e-05,
      "loss": 1.8663,
      "step": 124
    },
    {
      "epoch": 0.11552680221811461,
      "grad_norm": 2.53125,
      "learning_rate": 1.999700866942885e-05,
      "loss": 1.8389,
      "step": 125
    },
    {
      "epoch": 0.11645101663585952,
      "grad_norm": 2.578125,
      "learning_rate": 1.9996623089926138e-05,
      "loss": 1.864,
      "step": 126
    },
    {
      "epoch": 0.11737523105360444,
      "grad_norm": 2.390625,
      "learning_rate": 1.999621414738484e-05,
      "loss": 1.8493,
      "step": 127
    },
    {
      "epoch": 0.11829944547134935,
      "grad_norm": 2.5,
      "learning_rate": 1.999578184276069e-05,
      "loss": 1.8744,
      "step": 128
    },
    {
      "epoch": 0.11922365988909427,
      "grad_norm": 2.484375,
      "learning_rate": 1.999532617706403e-05,
      "loss": 1.8266,
      "step": 129
    },
    {
      "epoch": 0.12014787430683918,
      "grad_norm": 2.53125,
      "learning_rate": 1.9994847151359786e-05,
      "loss": 1.8983,
      "step": 130
    },
    {
      "epoch": 0.12107208872458411,
      "grad_norm": 2.375,
      "learning_rate": 1.9994344766767485e-05,
      "loss": 1.8332,
      "step": 131
    },
    {
      "epoch": 0.12199630314232902,
      "grad_norm": 2.40625,
      "learning_rate": 1.9993819024461255e-05,
      "loss": 1.7669,
      "step": 132
    },
    {
      "epoch": 0.12292051756007394,
      "grad_norm": 2.578125,
      "learning_rate": 1.9993269925669798e-05,
      "loss": 1.8584,
      "step": 133
    },
    {
      "epoch": 0.12384473197781885,
      "grad_norm": 2.640625,
      "learning_rate": 1.9992697471676413e-05,
      "loss": 1.8922,
      "step": 134
    },
    {
      "epoch": 0.12476894639556377,
      "grad_norm": 2.546875,
      "learning_rate": 1.9992101663818976e-05,
      "loss": 1.9421,
      "step": 135
    },
    {
      "epoch": 0.1256931608133087,
      "grad_norm": 2.828125,
      "learning_rate": 1.999148250348994e-05,
      "loss": 1.921,
      "step": 136
    },
    {
      "epoch": 0.1266173752310536,
      "grad_norm": 2.640625,
      "learning_rate": 1.9990839992136354e-05,
      "loss": 1.801,
      "step": 137
    },
    {
      "epoch": 0.12754158964879853,
      "grad_norm": 2.59375,
      "learning_rate": 1.9990174131259813e-05,
      "loss": 1.7953,
      "step": 138
    },
    {
      "epoch": 0.12846580406654343,
      "grad_norm": 2.546875,
      "learning_rate": 1.9989484922416503e-05,
      "loss": 1.8359,
      "step": 139
    },
    {
      "epoch": 0.12939001848428835,
      "grad_norm": 2.5,
      "learning_rate": 1.998877236721717e-05,
      "loss": 1.8894,
      "step": 140
    },
    {
      "epoch": 0.13031423290203328,
      "grad_norm": 2.890625,
      "learning_rate": 1.9988036467327116e-05,
      "loss": 1.9296,
      "step": 141
    },
    {
      "epoch": 0.13123844731977818,
      "grad_norm": 2.4375,
      "learning_rate": 1.9987277224466216e-05,
      "loss": 1.7654,
      "step": 142
    },
    {
      "epoch": 0.1321626617375231,
      "grad_norm": 2.453125,
      "learning_rate": 1.998649464040888e-05,
      "loss": 1.8479,
      "step": 143
    },
    {
      "epoch": 0.133086876155268,
      "grad_norm": 2.484375,
      "learning_rate": 1.998568871698409e-05,
      "loss": 1.9108,
      "step": 144
    },
    {
      "epoch": 0.13401109057301294,
      "grad_norm": 2.53125,
      "learning_rate": 1.998485945607536e-05,
      "loss": 1.8495,
      "step": 145
    },
    {
      "epoch": 0.13493530499075784,
      "grad_norm": 2.375,
      "learning_rate": 1.998400685962075e-05,
      "loss": 1.7533,
      "step": 146
    },
    {
      "epoch": 0.13585951940850277,
      "grad_norm": 2.421875,
      "learning_rate": 1.998313092961286e-05,
      "loss": 1.8244,
      "step": 147
    },
    {
      "epoch": 0.1367837338262477,
      "grad_norm": 2.4375,
      "learning_rate": 1.9982231668098814e-05,
      "loss": 1.835,
      "step": 148
    },
    {
      "epoch": 0.1377079482439926,
      "grad_norm": 2.40625,
      "learning_rate": 1.9981309077180273e-05,
      "loss": 1.7918,
      "step": 149
    },
    {
      "epoch": 0.13863216266173753,
      "grad_norm": 2.421875,
      "learning_rate": 1.9980363159013422e-05,
      "loss": 1.8796,
      "step": 150
    },
    {
      "epoch": 0.13955637707948243,
      "grad_norm": 2.421875,
      "learning_rate": 1.997939391580895e-05,
      "loss": 1.8271,
      "step": 151
    },
    {
      "epoch": 0.14048059149722736,
      "grad_norm": 2.5,
      "learning_rate": 1.997840134983208e-05,
      "loss": 1.8558,
      "step": 152
    },
    {
      "epoch": 0.1414048059149723,
      "grad_norm": 2.421875,
      "learning_rate": 1.9977385463402522e-05,
      "loss": 1.7849,
      "step": 153
    },
    {
      "epoch": 0.1423290203327172,
      "grad_norm": 2.421875,
      "learning_rate": 1.9976346258894502e-05,
      "loss": 1.792,
      "step": 154
    },
    {
      "epoch": 0.14325323475046212,
      "grad_norm": 2.71875,
      "learning_rate": 1.9975283738736734e-05,
      "loss": 1.8889,
      "step": 155
    },
    {
      "epoch": 0.14417744916820702,
      "grad_norm": 2.734375,
      "learning_rate": 1.997419790541243e-05,
      "loss": 1.9106,
      "step": 156
    },
    {
      "epoch": 0.14510166358595195,
      "grad_norm": 2.46875,
      "learning_rate": 1.9973088761459286e-05,
      "loss": 1.8254,
      "step": 157
    },
    {
      "epoch": 0.14602587800369685,
      "grad_norm": 2.578125,
      "learning_rate": 1.9971956309469468e-05,
      "loss": 1.8388,
      "step": 158
    },
    {
      "epoch": 0.14695009242144177,
      "grad_norm": 2.484375,
      "learning_rate": 1.9970800552089623e-05,
      "loss": 1.8595,
      "step": 159
    },
    {
      "epoch": 0.1478743068391867,
      "grad_norm": 2.421875,
      "learning_rate": 1.996962149202087e-05,
      "loss": 1.8562,
      "step": 160
    },
    {
      "epoch": 0.1487985212569316,
      "grad_norm": 2.46875,
      "learning_rate": 1.9968419132018774e-05,
      "loss": 1.896,
      "step": 161
    },
    {
      "epoch": 0.14972273567467653,
      "grad_norm": 2.515625,
      "learning_rate": 1.9967193474893367e-05,
      "loss": 1.8293,
      "step": 162
    },
    {
      "epoch": 0.15064695009242143,
      "grad_norm": 2.59375,
      "learning_rate": 1.996594452350912e-05,
      "loss": 1.8038,
      "step": 163
    },
    {
      "epoch": 0.15157116451016636,
      "grad_norm": 2.453125,
      "learning_rate": 1.9964672280784953e-05,
      "loss": 1.9295,
      "step": 164
    },
    {
      "epoch": 0.15249537892791126,
      "grad_norm": 2.609375,
      "learning_rate": 1.9963376749694215e-05,
      "loss": 1.8757,
      "step": 165
    },
    {
      "epoch": 0.1534195933456562,
      "grad_norm": 2.375,
      "learning_rate": 1.996205793326468e-05,
      "loss": 1.7343,
      "step": 166
    },
    {
      "epoch": 0.15434380776340112,
      "grad_norm": 2.359375,
      "learning_rate": 1.996071583457855e-05,
      "loss": 1.8195,
      "step": 167
    },
    {
      "epoch": 0.15526802218114602,
      "grad_norm": 2.46875,
      "learning_rate": 1.9959350456772424e-05,
      "loss": 1.796,
      "step": 168
    },
    {
      "epoch": 0.15619223659889095,
      "grad_norm": 2.578125,
      "learning_rate": 1.9957961803037325e-05,
      "loss": 1.9167,
      "step": 169
    },
    {
      "epoch": 0.15711645101663585,
      "grad_norm": 2.390625,
      "learning_rate": 1.9956549876618668e-05,
      "loss": 1.8011,
      "step": 170
    },
    {
      "epoch": 0.15804066543438078,
      "grad_norm": 2.40625,
      "learning_rate": 1.9955114680816253e-05,
      "loss": 1.8689,
      "step": 171
    },
    {
      "epoch": 0.1589648798521257,
      "grad_norm": 2.625,
      "learning_rate": 1.9953656218984263e-05,
      "loss": 1.8865,
      "step": 172
    },
    {
      "epoch": 0.1598890942698706,
      "grad_norm": 2.515625,
      "learning_rate": 1.9952174494531266e-05,
      "loss": 1.8288,
      "step": 173
    },
    {
      "epoch": 0.16081330868761554,
      "grad_norm": 2.40625,
      "learning_rate": 1.9950669510920184e-05,
      "loss": 1.9267,
      "step": 174
    },
    {
      "epoch": 0.16173752310536044,
      "grad_norm": 2.375,
      "learning_rate": 1.9949141271668305e-05,
      "loss": 1.785,
      "step": 175
    },
    {
      "epoch": 0.16266173752310537,
      "grad_norm": 2.40625,
      "learning_rate": 1.994758978034727e-05,
      "loss": 1.8777,
      "step": 176
    },
    {
      "epoch": 0.16358595194085027,
      "grad_norm": 2.5,
      "learning_rate": 1.994601504058306e-05,
      "loss": 1.7895,
      "step": 177
    },
    {
      "epoch": 0.1645101663585952,
      "grad_norm": 2.46875,
      "learning_rate": 1.9944417056055978e-05,
      "loss": 1.934,
      "step": 178
    },
    {
      "epoch": 0.16543438077634012,
      "grad_norm": 2.390625,
      "learning_rate": 1.994279583050067e-05,
      "loss": 1.7927,
      "step": 179
    },
    {
      "epoch": 0.16635859519408502,
      "grad_norm": 2.4375,
      "learning_rate": 1.9941151367706097e-05,
      "loss": 1.8511,
      "step": 180
    },
    {
      "epoch": 0.16728280961182995,
      "grad_norm": 2.40625,
      "learning_rate": 1.9939483671515508e-05,
      "loss": 1.8345,
      "step": 181
    },
    {
      "epoch": 0.16820702402957485,
      "grad_norm": 2.3125,
      "learning_rate": 1.993779274582647e-05,
      "loss": 1.8126,
      "step": 182
    },
    {
      "epoch": 0.16913123844731978,
      "grad_norm": 2.484375,
      "learning_rate": 1.993607859459084e-05,
      "loss": 1.7867,
      "step": 183
    },
    {
      "epoch": 0.17005545286506468,
      "grad_norm": 2.4375,
      "learning_rate": 1.993434122181474e-05,
      "loss": 1.8409,
      "step": 184
    },
    {
      "epoch": 0.1709796672828096,
      "grad_norm": 2.3125,
      "learning_rate": 1.9932580631558575e-05,
      "loss": 1.8303,
      "step": 185
    },
    {
      "epoch": 0.17190388170055454,
      "grad_norm": 2.546875,
      "learning_rate": 1.993079682793701e-05,
      "loss": 1.7322,
      "step": 186
    },
    {
      "epoch": 0.17282809611829944,
      "grad_norm": 2.390625,
      "learning_rate": 1.9928989815118963e-05,
      "loss": 1.7711,
      "step": 187
    },
    {
      "epoch": 0.17375231053604437,
      "grad_norm": 2.46875,
      "learning_rate": 1.9927159597327583e-05,
      "loss": 1.8086,
      "step": 188
    },
    {
      "epoch": 0.17467652495378927,
      "grad_norm": 2.4375,
      "learning_rate": 1.9925306178840263e-05,
      "loss": 1.8853,
      "step": 189
    },
    {
      "epoch": 0.1756007393715342,
      "grad_norm": 2.515625,
      "learning_rate": 1.9923429563988613e-05,
      "loss": 1.8893,
      "step": 190
    },
    {
      "epoch": 0.17652495378927913,
      "grad_norm": 2.53125,
      "learning_rate": 1.992152975715846e-05,
      "loss": 1.8385,
      "step": 191
    },
    {
      "epoch": 0.17744916820702403,
      "grad_norm": 7.625,
      "learning_rate": 1.991960676278983e-05,
      "loss": 1.8494,
      "step": 192
    },
    {
      "epoch": 0.17837338262476896,
      "grad_norm": 2.578125,
      "learning_rate": 1.9917660585376936e-05,
      "loss": 1.7777,
      "step": 193
    },
    {
      "epoch": 0.17929759704251386,
      "grad_norm": 5.59375,
      "learning_rate": 1.991569122946818e-05,
      "loss": 1.8623,
      "step": 194
    },
    {
      "epoch": 0.18022181146025879,
      "grad_norm": 2.5625,
      "learning_rate": 1.9913698699666128e-05,
      "loss": 1.8877,
      "step": 195
    },
    {
      "epoch": 0.18114602587800369,
      "grad_norm": 2.375,
      "learning_rate": 1.9911683000627504e-05,
      "loss": 1.8179,
      "step": 196
    },
    {
      "epoch": 0.18207024029574861,
      "grad_norm": 2.546875,
      "learning_rate": 1.9909644137063186e-05,
      "loss": 1.8584,
      "step": 197
    },
    {
      "epoch": 0.18299445471349354,
      "grad_norm": 2.59375,
      "learning_rate": 1.9907582113738195e-05,
      "loss": 1.7721,
      "step": 198
    },
    {
      "epoch": 0.18391866913123844,
      "grad_norm": 2.40625,
      "learning_rate": 1.990549693547166e-05,
      "loss": 1.8218,
      "step": 199
    },
    {
      "epoch": 0.18484288354898337,
      "grad_norm": 2.34375,
      "learning_rate": 1.9903388607136846e-05,
      "loss": 1.8184,
      "step": 200
    },
    {
      "epoch": 0.18576709796672827,
      "grad_norm": 2.515625,
      "learning_rate": 1.9901257133661106e-05,
      "loss": 1.8708,
      "step": 201
    },
    {
      "epoch": 0.1866913123844732,
      "grad_norm": 2.578125,
      "learning_rate": 1.9899102520025896e-05,
      "loss": 1.828,
      "step": 202
    },
    {
      "epoch": 0.1876155268022181,
      "grad_norm": 2.34375,
      "learning_rate": 1.989692477126675e-05,
      "loss": 1.7782,
      "step": 203
    },
    {
      "epoch": 0.18853974121996303,
      "grad_norm": 2.453125,
      "learning_rate": 1.9894723892473264e-05,
      "loss": 1.8454,
      "step": 204
    },
    {
      "epoch": 0.18946395563770796,
      "grad_norm": 2.375,
      "learning_rate": 1.9892499888789097e-05,
      "loss": 1.8232,
      "step": 205
    },
    {
      "epoch": 0.19038817005545286,
      "grad_norm": 2.515625,
      "learning_rate": 1.9890252765411962e-05,
      "loss": 1.8581,
      "step": 206
    },
    {
      "epoch": 0.1913123844731978,
      "grad_norm": 2.65625,
      "learning_rate": 1.9887982527593584e-05,
      "loss": 1.859,
      "step": 207
    },
    {
      "epoch": 0.1922365988909427,
      "grad_norm": 2.375,
      "learning_rate": 1.9885689180639725e-05,
      "loss": 1.8113,
      "step": 208
    },
    {
      "epoch": 0.19316081330868762,
      "grad_norm": 2.71875,
      "learning_rate": 1.9883372729910154e-05,
      "loss": 1.7921,
      "step": 209
    },
    {
      "epoch": 0.19408502772643252,
      "grad_norm": 2.546875,
      "learning_rate": 1.9881033180818622e-05,
      "loss": 1.8889,
      "step": 210
    },
    {
      "epoch": 0.19500924214417745,
      "grad_norm": 2.375,
      "learning_rate": 1.9878670538832882e-05,
      "loss": 1.7741,
      "step": 211
    },
    {
      "epoch": 0.19593345656192238,
      "grad_norm": 2.640625,
      "learning_rate": 1.9876284809474647e-05,
      "loss": 1.8682,
      "step": 212
    },
    {
      "epoch": 0.19685767097966728,
      "grad_norm": 2.421875,
      "learning_rate": 1.9873875998319585e-05,
      "loss": 1.8699,
      "step": 213
    },
    {
      "epoch": 0.1977818853974122,
      "grad_norm": 2.578125,
      "learning_rate": 1.987144411099731e-05,
      "loss": 1.8366,
      "step": 214
    },
    {
      "epoch": 0.1987060998151571,
      "grad_norm": 2.4375,
      "learning_rate": 1.986898915319138e-05,
      "loss": 1.8025,
      "step": 215
    },
    {
      "epoch": 0.19963031423290203,
      "grad_norm": 2.5,
      "learning_rate": 1.9866511130639247e-05,
      "loss": 1.882,
      "step": 216
    },
    {
      "epoch": 0.20055452865064696,
      "grad_norm": 2.484375,
      "learning_rate": 1.9864010049132287e-05,
      "loss": 1.7892,
      "step": 217
    },
    {
      "epoch": 0.20147874306839186,
      "grad_norm": 2.53125,
      "learning_rate": 1.986148591451576e-05,
      "loss": 1.8574,
      "step": 218
    },
    {
      "epoch": 0.2024029574861368,
      "grad_norm": 2.40625,
      "learning_rate": 1.9858938732688803e-05,
      "loss": 1.739,
      "step": 219
    },
    {
      "epoch": 0.2033271719038817,
      "grad_norm": 2.578125,
      "learning_rate": 1.9856368509604414e-05,
      "loss": 1.8676,
      "step": 220
    },
    {
      "epoch": 0.20425138632162662,
      "grad_norm": 2.421875,
      "learning_rate": 1.985377525126945e-05,
      "loss": 1.7483,
      "step": 221
    },
    {
      "epoch": 0.20517560073937152,
      "grad_norm": 2.484375,
      "learning_rate": 1.985115896374459e-05,
      "loss": 1.8347,
      "step": 222
    },
    {
      "epoch": 0.20609981515711645,
      "grad_norm": 2.515625,
      "learning_rate": 1.9848519653144348e-05,
      "loss": 1.7917,
      "step": 223
    },
    {
      "epoch": 0.20702402957486138,
      "grad_norm": 2.453125,
      "learning_rate": 1.984585732563703e-05,
      "loss": 1.7943,
      "step": 224
    },
    {
      "epoch": 0.20794824399260628,
      "grad_norm": 2.375,
      "learning_rate": 1.9843171987444753e-05,
      "loss": 1.8078,
      "step": 225
    },
    {
      "epoch": 0.2088724584103512,
      "grad_norm": 2.65625,
      "learning_rate": 1.9840463644843397e-05,
      "loss": 1.7609,
      "step": 226
    },
    {
      "epoch": 0.2097966728280961,
      "grad_norm": 2.453125,
      "learning_rate": 1.9837732304162612e-05,
      "loss": 1.7535,
      "step": 227
    },
    {
      "epoch": 0.21072088724584104,
      "grad_norm": 2.875,
      "learning_rate": 1.9834977971785792e-05,
      "loss": 1.8484,
      "step": 228
    },
    {
      "epoch": 0.21164510166358594,
      "grad_norm": 2.328125,
      "learning_rate": 1.9832200654150077e-05,
      "loss": 1.8863,
      "step": 229
    },
    {
      "epoch": 0.21256931608133087,
      "grad_norm": 2.625,
      "learning_rate": 1.9829400357746307e-05,
      "loss": 1.8598,
      "step": 230
    },
    {
      "epoch": 0.2134935304990758,
      "grad_norm": 2.484375,
      "learning_rate": 1.9826577089119038e-05,
      "loss": 1.8361,
      "step": 231
    },
    {
      "epoch": 0.2144177449168207,
      "grad_norm": 2.390625,
      "learning_rate": 1.982373085486651e-05,
      "loss": 1.7892,
      "step": 232
    },
    {
      "epoch": 0.21534195933456562,
      "grad_norm": 2.375,
      "learning_rate": 1.9820861661640646e-05,
      "loss": 1.8467,
      "step": 233
    },
    {
      "epoch": 0.21626617375231053,
      "grad_norm": 2.5,
      "learning_rate": 1.981796951614701e-05,
      "loss": 1.8468,
      "step": 234
    },
    {
      "epoch": 0.21719038817005545,
      "grad_norm": 2.5,
      "learning_rate": 1.9815054425144817e-05,
      "loss": 1.8043,
      "step": 235
    },
    {
      "epoch": 0.21811460258780038,
      "grad_norm": 5.71875,
      "learning_rate": 1.9812116395446906e-05,
      "loss": 1.8287,
      "step": 236
    },
    {
      "epoch": 0.21903881700554528,
      "grad_norm": 2.328125,
      "learning_rate": 1.9809155433919726e-05,
      "loss": 1.7701,
      "step": 237
    },
    {
      "epoch": 0.2199630314232902,
      "grad_norm": 2.484375,
      "learning_rate": 1.9806171547483317e-05,
      "loss": 1.8887,
      "step": 238
    },
    {
      "epoch": 0.2208872458410351,
      "grad_norm": 2.765625,
      "learning_rate": 1.9803164743111304e-05,
      "loss": 1.7298,
      "step": 239
    },
    {
      "epoch": 0.22181146025878004,
      "grad_norm": 2.390625,
      "learning_rate": 1.9800135027830866e-05,
      "loss": 1.8693,
      "step": 240
    },
    {
      "epoch": 0.22273567467652494,
      "grad_norm": 2.65625,
      "learning_rate": 1.979708240872273e-05,
      "loss": 1.9084,
      "step": 241
    },
    {
      "epoch": 0.22365988909426987,
      "grad_norm": 3.3125,
      "learning_rate": 1.979400689292115e-05,
      "loss": 1.8188,
      "step": 242
    },
    {
      "epoch": 0.2245841035120148,
      "grad_norm": 2.71875,
      "learning_rate": 1.9790908487613895e-05,
      "loss": 1.7873,
      "step": 243
    },
    {
      "epoch": 0.2255083179297597,
      "grad_norm": 15.5,
      "learning_rate": 1.9787787200042224e-05,
      "loss": 1.8445,
      "step": 244
    },
    {
      "epoch": 0.22643253234750463,
      "grad_norm": 3.375,
      "learning_rate": 1.9784643037500877e-05,
      "loss": 1.8383,
      "step": 245
    },
    {
      "epoch": 0.22735674676524953,
      "grad_norm": 6.84375,
      "learning_rate": 1.9781476007338058e-05,
      "loss": 1.834,
      "step": 246
    },
    {
      "epoch": 0.22828096118299446,
      "grad_norm": 3.15625,
      "learning_rate": 1.9778286116955407e-05,
      "loss": 1.7983,
      "step": 247
    },
    {
      "epoch": 0.22920517560073936,
      "grad_norm": 3.125,
      "learning_rate": 1.9775073373807995e-05,
      "loss": 1.8401,
      "step": 248
    },
    {
      "epoch": 0.2301293900184843,
      "grad_norm": 2.890625,
      "learning_rate": 1.9771837785404308e-05,
      "loss": 1.8153,
      "step": 249
    },
    {
      "epoch": 0.23105360443622922,
      "grad_norm": 29.75,
      "learning_rate": 1.9768579359306204e-05,
      "loss": 1.8784,
      "step": 250
    },
    {
      "epoch": 0.23197781885397412,
      "grad_norm": 2.484375,
      "learning_rate": 1.9765298103128945e-05,
      "loss": 1.8219,
      "step": 251
    },
    {
      "epoch": 0.23290203327171904,
      "grad_norm": 2.484375,
      "learning_rate": 1.9761994024541118e-05,
      "loss": 1.7413,
      "step": 252
    },
    {
      "epoch": 0.23382624768946395,
      "grad_norm": 2.40625,
      "learning_rate": 1.975866713126467e-05,
      "loss": 1.8455,
      "step": 253
    },
    {
      "epoch": 0.23475046210720887,
      "grad_norm": 2.515625,
      "learning_rate": 1.9755317431074857e-05,
      "loss": 1.7784,
      "step": 254
    },
    {
      "epoch": 0.2356746765249538,
      "grad_norm": 2.515625,
      "learning_rate": 1.975194493180024e-05,
      "loss": 1.836,
      "step": 255
    },
    {
      "epoch": 0.2365988909426987,
      "grad_norm": 2.40625,
      "learning_rate": 1.9748549641322668e-05,
      "loss": 1.7377,
      "step": 256
    },
    {
      "epoch": 0.23752310536044363,
      "grad_norm": 2.546875,
      "learning_rate": 1.974513156757725e-05,
      "loss": 1.853,
      "step": 257
    },
    {
      "epoch": 0.23844731977818853,
      "grad_norm": 2.515625,
      "learning_rate": 1.974169071855233e-05,
      "loss": 1.7921,
      "step": 258
    },
    {
      "epoch": 0.23937153419593346,
      "grad_norm": 2.375,
      "learning_rate": 1.973822710228951e-05,
      "loss": 1.8444,
      "step": 259
    },
    {
      "epoch": 0.24029574861367836,
      "grad_norm": 2.4375,
      "learning_rate": 1.9734740726883572e-05,
      "loss": 1.8632,
      "step": 260
    },
    {
      "epoch": 0.2412199630314233,
      "grad_norm": 2.5,
      "learning_rate": 1.9731231600482507e-05,
      "loss": 1.7819,
      "step": 261
    },
    {
      "epoch": 0.24214417744916822,
      "grad_norm": 2.484375,
      "learning_rate": 1.9727699731287465e-05,
      "loss": 1.7956,
      "step": 262
    },
    {
      "epoch": 0.24306839186691312,
      "grad_norm": 2.53125,
      "learning_rate": 1.9724145127552756e-05,
      "loss": 1.8289,
      "step": 263
    },
    {
      "epoch": 0.24399260628465805,
      "grad_norm": 2.40625,
      "learning_rate": 1.9720567797585815e-05,
      "loss": 1.7983,
      "step": 264
    },
    {
      "epoch": 0.24491682070240295,
      "grad_norm": 2.296875,
      "learning_rate": 1.9716967749747206e-05,
      "loss": 1.8215,
      "step": 265
    },
    {
      "epoch": 0.24584103512014788,
      "grad_norm": 2.421875,
      "learning_rate": 1.9713344992450564e-05,
      "loss": 1.871,
      "step": 266
    },
    {
      "epoch": 0.24676524953789278,
      "grad_norm": 2.4375,
      "learning_rate": 1.9709699534162615e-05,
      "loss": 1.8487,
      "step": 267
    },
    {
      "epoch": 0.2476894639556377,
      "grad_norm": 2.421875,
      "learning_rate": 1.9706031383403133e-05,
      "loss": 1.8694,
      "step": 268
    },
    {
      "epoch": 0.24861367837338263,
      "grad_norm": 2.40625,
      "learning_rate": 1.970234054874493e-05,
      "loss": 1.8236,
      "step": 269
    },
    {
      "epoch": 0.24953789279112754,
      "grad_norm": 2.6875,
      "learning_rate": 1.969862703881383e-05,
      "loss": 1.7475,
      "step": 270
    },
    {
      "epoch": 0.25046210720887246,
      "grad_norm": 2.375,
      "learning_rate": 1.9694890862288647e-05,
      "loss": 1.7394,
      "step": 271
    },
    {
      "epoch": 0.2513863216266174,
      "grad_norm": 2.5,
      "learning_rate": 1.969113202790118e-05,
      "loss": 1.8577,
      "step": 272
    },
    {
      "epoch": 0.25231053604436227,
      "grad_norm": 2.703125,
      "learning_rate": 1.9687350544436167e-05,
      "loss": 1.8408,
      "step": 273
    },
    {
      "epoch": 0.2532347504621072,
      "grad_norm": 2.5625,
      "learning_rate": 1.9683546420731292e-05,
      "loss": 1.813,
      "step": 274
    },
    {
      "epoch": 0.2541589648798521,
      "grad_norm": 2.515625,
      "learning_rate": 1.9679719665677148e-05,
      "loss": 1.7782,
      "step": 275
    },
    {
      "epoch": 0.25508317929759705,
      "grad_norm": 2.75,
      "learning_rate": 1.967587028821721e-05,
      "loss": 1.8207,
      "step": 276
    },
    {
      "epoch": 0.256007393715342,
      "grad_norm": 2.46875,
      "learning_rate": 1.967199829734784e-05,
      "loss": 1.8807,
      "step": 277
    },
    {
      "epoch": 0.25693160813308685,
      "grad_norm": 2.421875,
      "learning_rate": 1.9668103702118238e-05,
      "loss": 1.8008,
      "step": 278
    },
    {
      "epoch": 0.2578558225508318,
      "grad_norm": 2.359375,
      "learning_rate": 1.9664186511630433e-05,
      "loss": 1.8277,
      "step": 279
    },
    {
      "epoch": 0.2587800369685767,
      "grad_norm": 2.765625,
      "learning_rate": 1.966024673503927e-05,
      "loss": 1.8582,
      "step": 280
    },
    {
      "epoch": 0.25970425138632164,
      "grad_norm": 2.359375,
      "learning_rate": 1.9656284381552364e-05,
      "loss": 1.729,
      "step": 281
    },
    {
      "epoch": 0.26062846580406657,
      "grad_norm": 2.53125,
      "learning_rate": 1.965229946043011e-05,
      "loss": 1.8339,
      "step": 282
    },
    {
      "epoch": 0.26155268022181144,
      "grad_norm": 2.453125,
      "learning_rate": 1.964829198098564e-05,
      "loss": 1.8304,
      "step": 283
    },
    {
      "epoch": 0.26247689463955637,
      "grad_norm": 2.59375,
      "learning_rate": 1.9644261952584804e-05,
      "loss": 1.8421,
      "step": 284
    },
    {
      "epoch": 0.2634011090573013,
      "grad_norm": 2.828125,
      "learning_rate": 1.9640209384646155e-05,
      "loss": 1.909,
      "step": 285
    },
    {
      "epoch": 0.2643253234750462,
      "grad_norm": 2.34375,
      "learning_rate": 1.963613428664092e-05,
      "loss": 1.8871,
      "step": 286
    },
    {
      "epoch": 0.26524953789279115,
      "grad_norm": 2.5625,
      "learning_rate": 1.9632036668092988e-05,
      "loss": 1.8236,
      "step": 287
    },
    {
      "epoch": 0.266173752310536,
      "grad_norm": 2.984375,
      "learning_rate": 1.9627916538578865e-05,
      "loss": 1.8041,
      "step": 288
    },
    {
      "epoch": 0.26709796672828096,
      "grad_norm": 2.375,
      "learning_rate": 1.9623773907727682e-05,
      "loss": 1.7939,
      "step": 289
    },
    {
      "epoch": 0.2680221811460259,
      "grad_norm": 2.4375,
      "learning_rate": 1.9619608785221155e-05,
      "loss": 1.8597,
      "step": 290
    },
    {
      "epoch": 0.2689463955637708,
      "grad_norm": 2.453125,
      "learning_rate": 1.9615421180793564e-05,
      "loss": 1.8199,
      "step": 291
    },
    {
      "epoch": 0.2698706099815157,
      "grad_norm": 2.5625,
      "learning_rate": 1.9611211104231724e-05,
      "loss": 1.8084,
      "step": 292
    },
    {
      "epoch": 0.2707948243992606,
      "grad_norm": 2.375,
      "learning_rate": 1.960697856537498e-05,
      "loss": 1.8826,
      "step": 293
    },
    {
      "epoch": 0.27171903881700554,
      "grad_norm": 2.375,
      "learning_rate": 1.960272357411517e-05,
      "loss": 1.805,
      "step": 294
    },
    {
      "epoch": 0.27264325323475047,
      "grad_norm": 2.546875,
      "learning_rate": 1.9598446140396607e-05,
      "loss": 1.7651,
      "step": 295
    },
    {
      "epoch": 0.2735674676524954,
      "grad_norm": 2.671875,
      "learning_rate": 1.959414627421604e-05,
      "loss": 1.7758,
      "step": 296
    },
    {
      "epoch": 0.2744916820702403,
      "grad_norm": 2.53125,
      "learning_rate": 1.958982398562267e-05,
      "loss": 1.8417,
      "step": 297
    },
    {
      "epoch": 0.2754158964879852,
      "grad_norm": 2.390625,
      "learning_rate": 1.9585479284718077e-05,
      "loss": 1.7954,
      "step": 298
    },
    {
      "epoch": 0.27634011090573013,
      "grad_norm": 2.34375,
      "learning_rate": 1.9581112181656242e-05,
      "loss": 1.8072,
      "step": 299
    },
    {
      "epoch": 0.27726432532347506,
      "grad_norm": 2.296875,
      "learning_rate": 1.9576722686643483e-05,
      "loss": 1.7432,
      "step": 300
    },
    {
      "epoch": 0.27818853974122,
      "grad_norm": 2.5,
      "learning_rate": 1.957231080993846e-05,
      "loss": 1.8771,
      "step": 301
    },
    {
      "epoch": 0.27911275415896486,
      "grad_norm": 2.359375,
      "learning_rate": 1.9567876561852147e-05,
      "loss": 1.8472,
      "step": 302
    },
    {
      "epoch": 0.2800369685767098,
      "grad_norm": 2.3125,
      "learning_rate": 1.956341995274778e-05,
      "loss": 1.8367,
      "step": 303
    },
    {
      "epoch": 0.2809611829944547,
      "grad_norm": 2.375,
      "learning_rate": 1.9558940993040885e-05,
      "loss": 1.7979,
      "step": 304
    },
    {
      "epoch": 0.28188539741219965,
      "grad_norm": 2.375,
      "learning_rate": 1.9554439693199194e-05,
      "loss": 1.8235,
      "step": 305
    },
    {
      "epoch": 0.2828096118299446,
      "grad_norm": 2.4375,
      "learning_rate": 1.9549916063742673e-05,
      "loss": 1.8021,
      "step": 306
    },
    {
      "epoch": 0.28373382624768945,
      "grad_norm": 2.359375,
      "learning_rate": 1.9545370115243463e-05,
      "loss": 1.8305,
      "step": 307
    },
    {
      "epoch": 0.2846580406654344,
      "grad_norm": 2.734375,
      "learning_rate": 1.9540801858325865e-05,
      "loss": 1.9333,
      "step": 308
    },
    {
      "epoch": 0.2855822550831793,
      "grad_norm": 3.875,
      "learning_rate": 1.9536211303666324e-05,
      "loss": 1.8398,
      "step": 309
    },
    {
      "epoch": 0.28650646950092423,
      "grad_norm": 2.875,
      "learning_rate": 1.953159846199339e-05,
      "loss": 1.8228,
      "step": 310
    },
    {
      "epoch": 0.2874306839186691,
      "grad_norm": 2.484375,
      "learning_rate": 1.9526963344087714e-05,
      "loss": 1.8473,
      "step": 311
    },
    {
      "epoch": 0.28835489833641403,
      "grad_norm": 2.515625,
      "learning_rate": 1.9522305960781986e-05,
      "loss": 1.835,
      "step": 312
    },
    {
      "epoch": 0.28927911275415896,
      "grad_norm": 2.359375,
      "learning_rate": 1.9517626322960946e-05,
      "loss": 1.7645,
      "step": 313
    },
    {
      "epoch": 0.2902033271719039,
      "grad_norm": 2.328125,
      "learning_rate": 1.9512924441561348e-05,
      "loss": 1.7436,
      "step": 314
    },
    {
      "epoch": 0.2911275415896488,
      "grad_norm": 2.40625,
      "learning_rate": 1.950820032757192e-05,
      "loss": 1.7826,
      "step": 315
    },
    {
      "epoch": 0.2920517560073937,
      "grad_norm": 2.40625,
      "learning_rate": 1.950345399203337e-05,
      "loss": 1.7733,
      "step": 316
    },
    {
      "epoch": 0.2929759704251386,
      "grad_norm": 2.453125,
      "learning_rate": 1.9498685446038306e-05,
      "loss": 1.7706,
      "step": 317
    },
    {
      "epoch": 0.29390018484288355,
      "grad_norm": 2.40625,
      "learning_rate": 1.949389470073128e-05,
      "loss": 1.8443,
      "step": 318
    },
    {
      "epoch": 0.2948243992606285,
      "grad_norm": 2.46875,
      "learning_rate": 1.9489081767308696e-05,
      "loss": 1.8424,
      "step": 319
    },
    {
      "epoch": 0.2957486136783734,
      "grad_norm": 2.296875,
      "learning_rate": 1.948424665701884e-05,
      "loss": 1.7751,
      "step": 320
    },
    {
      "epoch": 0.2966728280961183,
      "grad_norm": 2.390625,
      "learning_rate": 1.947938938116181e-05,
      "loss": 1.8094,
      "step": 321
    },
    {
      "epoch": 0.2975970425138632,
      "grad_norm": 2.375,
      "learning_rate": 1.9474509951089508e-05,
      "loss": 1.8415,
      "step": 322
    },
    {
      "epoch": 0.29852125693160814,
      "grad_norm": 2.453125,
      "learning_rate": 1.9469608378205624e-05,
      "loss": 1.7953,
      "step": 323
    },
    {
      "epoch": 0.29944547134935307,
      "grad_norm": 2.390625,
      "learning_rate": 1.9464684673965583e-05,
      "loss": 1.8729,
      "step": 324
    },
    {
      "epoch": 0.300369685767098,
      "grad_norm": 2.3125,
      "learning_rate": 1.9459738849876545e-05,
      "loss": 1.7748,
      "step": 325
    },
    {
      "epoch": 0.30129390018484287,
      "grad_norm": 2.484375,
      "learning_rate": 1.9454770917497357e-05,
      "loss": 1.828,
      "step": 326
    },
    {
      "epoch": 0.3022181146025878,
      "grad_norm": 2.453125,
      "learning_rate": 1.944978088843854e-05,
      "loss": 1.8357,
      "step": 327
    },
    {
      "epoch": 0.3031423290203327,
      "grad_norm": 2.390625,
      "learning_rate": 1.944476877436226e-05,
      "loss": 1.7777,
      "step": 328
    },
    {
      "epoch": 0.30406654343807765,
      "grad_norm": 2.3125,
      "learning_rate": 1.9439734586982292e-05,
      "loss": 1.8503,
      "step": 329
    },
    {
      "epoch": 0.3049907578558225,
      "grad_norm": 2.359375,
      "learning_rate": 1.9434678338063997e-05,
      "loss": 1.8526,
      "step": 330
    },
    {
      "epoch": 0.30591497227356745,
      "grad_norm": 2.359375,
      "learning_rate": 1.9429600039424306e-05,
      "loss": 1.804,
      "step": 331
    },
    {
      "epoch": 0.3068391866913124,
      "grad_norm": 2.46875,
      "learning_rate": 1.942449970293167e-05,
      "loss": 1.7715,
      "step": 332
    },
    {
      "epoch": 0.3077634011090573,
      "grad_norm": 2.484375,
      "learning_rate": 1.9419377340506055e-05,
      "loss": 1.7458,
      "step": 333
    },
    {
      "epoch": 0.30868761552680224,
      "grad_norm": 2.359375,
      "learning_rate": 1.9414232964118893e-05,
      "loss": 1.8572,
      "step": 334
    },
    {
      "epoch": 0.3096118299445471,
      "grad_norm": 2.40625,
      "learning_rate": 1.9409066585793072e-05,
      "loss": 1.8761,
      "step": 335
    },
    {
      "epoch": 0.31053604436229204,
      "grad_norm": 2.390625,
      "learning_rate": 1.9403878217602905e-05,
      "loss": 1.7616,
      "step": 336
    },
    {
      "epoch": 0.31146025878003697,
      "grad_norm": 2.328125,
      "learning_rate": 1.9398667871674082e-05,
      "loss": 1.8547,
      "step": 337
    },
    {
      "epoch": 0.3123844731977819,
      "grad_norm": 2.34375,
      "learning_rate": 1.9393435560183676e-05,
      "loss": 1.8155,
      "step": 338
    },
    {
      "epoch": 0.3133086876155268,
      "grad_norm": 2.53125,
      "learning_rate": 1.9388181295360077e-05,
      "loss": 1.8441,
      "step": 339
    },
    {
      "epoch": 0.3142329020332717,
      "grad_norm": 2.421875,
      "learning_rate": 1.9382905089482994e-05,
      "loss": 1.7487,
      "step": 340
    },
    {
      "epoch": 0.31515711645101663,
      "grad_norm": 2.453125,
      "learning_rate": 1.937760695488342e-05,
      "loss": 1.8294,
      "step": 341
    },
    {
      "epoch": 0.31608133086876156,
      "grad_norm": 2.484375,
      "learning_rate": 1.9372286903943572e-05,
      "loss": 1.8724,
      "step": 342
    },
    {
      "epoch": 0.3170055452865065,
      "grad_norm": 2.640625,
      "learning_rate": 1.9366944949096917e-05,
      "loss": 1.8414,
      "step": 343
    },
    {
      "epoch": 0.3179297597042514,
      "grad_norm": 2.40625,
      "learning_rate": 1.9361581102828095e-05,
      "loss": 1.8197,
      "step": 344
    },
    {
      "epoch": 0.3188539741219963,
      "grad_norm": 2.375,
      "learning_rate": 1.935619537767292e-05,
      "loss": 1.7762,
      "step": 345
    },
    {
      "epoch": 0.3197781885397412,
      "grad_norm": 2.40625,
      "learning_rate": 1.9350787786218325e-05,
      "loss": 1.8243,
      "step": 346
    },
    {
      "epoch": 0.32070240295748614,
      "grad_norm": 2.375,
      "learning_rate": 1.9345358341102358e-05,
      "loss": 1.8216,
      "step": 347
    },
    {
      "epoch": 0.32162661737523107,
      "grad_norm": 2.328125,
      "learning_rate": 1.933990705501414e-05,
      "loss": 1.8647,
      "step": 348
    },
    {
      "epoch": 0.32255083179297594,
      "grad_norm": 2.4375,
      "learning_rate": 1.933443394069383e-05,
      "loss": 1.7994,
      "step": 349
    },
    {
      "epoch": 0.3234750462107209,
      "grad_norm": 2.5625,
      "learning_rate": 1.9328939010932607e-05,
      "loss": 1.7937,
      "step": 350
    },
    {
      "epoch": 0.3243992606284658,
      "grad_norm": 2.421875,
      "learning_rate": 1.9323422278572635e-05,
      "loss": 1.8238,
      "step": 351
    },
    {
      "epoch": 0.32532347504621073,
      "grad_norm": 2.34375,
      "learning_rate": 1.9317883756507026e-05,
      "loss": 1.7627,
      "step": 352
    },
    {
      "epoch": 0.32624768946395566,
      "grad_norm": 2.609375,
      "learning_rate": 1.9312323457679827e-05,
      "loss": 1.8865,
      "step": 353
    },
    {
      "epoch": 0.32717190388170053,
      "grad_norm": 2.546875,
      "learning_rate": 1.9306741395085977e-05,
      "loss": 1.7641,
      "step": 354
    },
    {
      "epoch": 0.32809611829944546,
      "grad_norm": 2.34375,
      "learning_rate": 1.9301137581771266e-05,
      "loss": 1.763,
      "step": 355
    },
    {
      "epoch": 0.3290203327171904,
      "grad_norm": 2.4375,
      "learning_rate": 1.9295512030832336e-05,
      "loss": 1.7827,
      "step": 356
    },
    {
      "epoch": 0.3299445471349353,
      "grad_norm": 2.6875,
      "learning_rate": 1.9289864755416615e-05,
      "loss": 1.6917,
      "step": 357
    },
    {
      "epoch": 0.33086876155268025,
      "grad_norm": 2.578125,
      "learning_rate": 1.928419576872232e-05,
      "loss": 1.7584,
      "step": 358
    },
    {
      "epoch": 0.3317929759704251,
      "grad_norm": 2.625,
      "learning_rate": 1.9278505083998393e-05,
      "loss": 1.8805,
      "step": 359
    },
    {
      "epoch": 0.33271719038817005,
      "grad_norm": 3.1875,
      "learning_rate": 1.9272792714544493e-05,
      "loss": 1.8331,
      "step": 360
    },
    {
      "epoch": 0.333641404805915,
      "grad_norm": 2.765625,
      "learning_rate": 1.9267058673710967e-05,
      "loss": 1.8396,
      "step": 361
    },
    {
      "epoch": 0.3345656192236599,
      "grad_norm": 2.859375,
      "learning_rate": 1.9261302974898798e-05,
      "loss": 1.8088,
      "step": 362
    },
    {
      "epoch": 0.33548983364140483,
      "grad_norm": 2.296875,
      "learning_rate": 1.925552563155959e-05,
      "loss": 1.843,
      "step": 363
    },
    {
      "epoch": 0.3364140480591497,
      "grad_norm": 2.625,
      "learning_rate": 1.9249726657195534e-05,
      "loss": 1.8551,
      "step": 364
    },
    {
      "epoch": 0.33733826247689463,
      "grad_norm": 2.953125,
      "learning_rate": 1.924390606535937e-05,
      "loss": 1.846,
      "step": 365
    },
    {
      "epoch": 0.33826247689463956,
      "grad_norm": 3.0625,
      "learning_rate": 1.923806386965436e-05,
      "loss": 1.8198,
      "step": 366
    },
    {
      "epoch": 0.3391866913123845,
      "grad_norm": 2.34375,
      "learning_rate": 1.9232200083734267e-05,
      "loss": 1.7839,
      "step": 367
    },
    {
      "epoch": 0.34011090573012936,
      "grad_norm": 2.65625,
      "learning_rate": 1.92263147213033e-05,
      "loss": 1.9284,
      "step": 368
    },
    {
      "epoch": 0.3410351201478743,
      "grad_norm": 2.578125,
      "learning_rate": 1.9220407796116098e-05,
      "loss": 1.8056,
      "step": 369
    },
    {
      "epoch": 0.3419593345656192,
      "grad_norm": 2.78125,
      "learning_rate": 1.9214479321977697e-05,
      "loss": 1.8224,
      "step": 370
    },
    {
      "epoch": 0.34288354898336415,
      "grad_norm": 2.4375,
      "learning_rate": 1.920852931274349e-05,
      "loss": 1.8813,
      "step": 371
    },
    {
      "epoch": 0.3438077634011091,
      "grad_norm": 2.609375,
      "learning_rate": 1.9202557782319207e-05,
      "loss": 1.8575,
      "step": 372
    },
    {
      "epoch": 0.34473197781885395,
      "grad_norm": 2.640625,
      "learning_rate": 1.9196564744660867e-05,
      "loss": 1.7455,
      "step": 373
    },
    {
      "epoch": 0.3456561922365989,
      "grad_norm": 2.421875,
      "learning_rate": 1.9190550213774757e-05,
      "loss": 1.832,
      "step": 374
    },
    {
      "epoch": 0.3465804066543438,
      "grad_norm": 2.3125,
      "learning_rate": 1.9184514203717394e-05,
      "loss": 1.8495,
      "step": 375
    },
    {
      "epoch": 0.34750462107208874,
      "grad_norm": 2.34375,
      "learning_rate": 1.9178456728595497e-05,
      "loss": 1.7797,
      "step": 376
    },
    {
      "epoch": 0.34842883548983367,
      "grad_norm": 2.375,
      "learning_rate": 1.9172377802565957e-05,
      "loss": 1.8169,
      "step": 377
    },
    {
      "epoch": 0.34935304990757854,
      "grad_norm": 2.375,
      "learning_rate": 1.9166277439835773e-05,
      "loss": 1.795,
      "step": 378
    },
    {
      "epoch": 0.35027726432532347,
      "grad_norm": 2.390625,
      "learning_rate": 1.9160155654662075e-05,
      "loss": 1.7354,
      "step": 379
    },
    {
      "epoch": 0.3512014787430684,
      "grad_norm": 2.296875,
      "learning_rate": 1.9154012461352035e-05,
      "loss": 1.6805,
      "step": 380
    },
    {
      "epoch": 0.3521256931608133,
      "grad_norm": 2.609375,
      "learning_rate": 1.914784787426288e-05,
      "loss": 1.7878,
      "step": 381
    },
    {
      "epoch": 0.35304990757855825,
      "grad_norm": 2.640625,
      "learning_rate": 1.9141661907801808e-05,
      "loss": 1.8455,
      "step": 382
    },
    {
      "epoch": 0.3539741219963031,
      "grad_norm": 2.375,
      "learning_rate": 1.913545457642601e-05,
      "loss": 1.7985,
      "step": 383
    },
    {
      "epoch": 0.35489833641404805,
      "grad_norm": 2.296875,
      "learning_rate": 1.9129225894642594e-05,
      "loss": 1.8432,
      "step": 384
    },
    {
      "epoch": 0.355822550831793,
      "grad_norm": 2.71875,
      "learning_rate": 1.9122975877008567e-05,
      "loss": 1.7886,
      "step": 385
    },
    {
      "epoch": 0.3567467652495379,
      "grad_norm": 3.015625,
      "learning_rate": 1.9116704538130803e-05,
      "loss": 1.8201,
      "step": 386
    },
    {
      "epoch": 0.3576709796672828,
      "grad_norm": 2.28125,
      "learning_rate": 1.9110411892666006e-05,
      "loss": 1.8003,
      "step": 387
    },
    {
      "epoch": 0.3585951940850277,
      "grad_norm": 2.609375,
      "learning_rate": 1.9104097955320672e-05,
      "loss": 1.9047,
      "step": 388
    },
    {
      "epoch": 0.35951940850277264,
      "grad_norm": 2.609375,
      "learning_rate": 1.9097762740851063e-05,
      "loss": 1.8118,
      "step": 389
    },
    {
      "epoch": 0.36044362292051757,
      "grad_norm": 2.671875,
      "learning_rate": 1.9091406264063163e-05,
      "loss": 1.8673,
      "step": 390
    },
    {
      "epoch": 0.3613678373382625,
      "grad_norm": 2.390625,
      "learning_rate": 1.9085028539812648e-05,
      "loss": 1.8354,
      "step": 391
    },
    {
      "epoch": 0.36229205175600737,
      "grad_norm": 2.484375,
      "learning_rate": 1.907862958300486e-05,
      "loss": 1.7845,
      "step": 392
    },
    {
      "epoch": 0.3632162661737523,
      "grad_norm": 2.5625,
      "learning_rate": 1.907220940859475e-05,
      "loss": 1.85,
      "step": 393
    },
    {
      "epoch": 0.36414048059149723,
      "grad_norm": 2.4375,
      "learning_rate": 1.9065768031586864e-05,
      "loss": 1.8279,
      "step": 394
    },
    {
      "epoch": 0.36506469500924216,
      "grad_norm": 2.28125,
      "learning_rate": 1.90593054670353e-05,
      "loss": 1.7901,
      "step": 395
    },
    {
      "epoch": 0.3659889094269871,
      "grad_norm": 2.515625,
      "learning_rate": 1.9052821730043676e-05,
      "loss": 1.7901,
      "step": 396
    },
    {
      "epoch": 0.36691312384473196,
      "grad_norm": 2.4375,
      "learning_rate": 1.9046316835765083e-05,
      "loss": 1.8505,
      "step": 397
    },
    {
      "epoch": 0.3678373382624769,
      "grad_norm": 2.546875,
      "learning_rate": 1.9039790799402072e-05,
      "loss": 1.8252,
      "step": 398
    },
    {
      "epoch": 0.3687615526802218,
      "grad_norm": 2.328125,
      "learning_rate": 1.903324363620659e-05,
      "loss": 1.7952,
      "step": 399
    },
    {
      "epoch": 0.36968576709796674,
      "grad_norm": 2.671875,
      "learning_rate": 1.902667536147997e-05,
      "loss": 1.8418,
      "step": 400
    },
    {
      "epoch": 0.3706099815157117,
      "grad_norm": 2.46875,
      "learning_rate": 1.9020085990572882e-05,
      "loss": 1.7914,
      "step": 401
    },
    {
      "epoch": 0.37153419593345655,
      "grad_norm": 2.40625,
      "learning_rate": 1.9013475538885297e-05,
      "loss": 1.7756,
      "step": 402
    },
    {
      "epoch": 0.3724584103512015,
      "grad_norm": 2.3125,
      "learning_rate": 1.900684402186646e-05,
      "loss": 1.7733,
      "step": 403
    },
    {
      "epoch": 0.3733826247689464,
      "grad_norm": 2.5,
      "learning_rate": 1.900019145501484e-05,
      "loss": 1.7646,
      "step": 404
    },
    {
      "epoch": 0.37430683918669133,
      "grad_norm": 2.296875,
      "learning_rate": 1.8993517853878105e-05,
      "loss": 1.7381,
      "step": 405
    },
    {
      "epoch": 0.3752310536044362,
      "grad_norm": 2.328125,
      "learning_rate": 1.898682323405308e-05,
      "loss": 1.7489,
      "step": 406
    },
    {
      "epoch": 0.37615526802218113,
      "grad_norm": 2.390625,
      "learning_rate": 1.8980107611185724e-05,
      "loss": 1.8204,
      "step": 407
    },
    {
      "epoch": 0.37707948243992606,
      "grad_norm": 2.375,
      "learning_rate": 1.897337100097106e-05,
      "loss": 1.781,
      "step": 408
    },
    {
      "epoch": 0.378003696857671,
      "grad_norm": 2.46875,
      "learning_rate": 1.896661341915318e-05,
      "loss": 1.8209,
      "step": 409
    },
    {
      "epoch": 0.3789279112754159,
      "grad_norm": 2.375,
      "learning_rate": 1.8959834881525183e-05,
      "loss": 1.8092,
      "step": 410
    },
    {
      "epoch": 0.3798521256931608,
      "grad_norm": 2.515625,
      "learning_rate": 1.895303540392914e-05,
      "loss": 1.8058,
      "step": 411
    },
    {
      "epoch": 0.3807763401109057,
      "grad_norm": 2.484375,
      "learning_rate": 1.894621500225606e-05,
      "loss": 1.8722,
      "step": 412
    },
    {
      "epoch": 0.38170055452865065,
      "grad_norm": 2.28125,
      "learning_rate": 1.8939373692445863e-05,
      "loss": 1.7865,
      "step": 413
    },
    {
      "epoch": 0.3826247689463956,
      "grad_norm": 2.703125,
      "learning_rate": 1.893251149048732e-05,
      "loss": 1.8205,
      "step": 414
    },
    {
      "epoch": 0.3835489833641405,
      "grad_norm": 2.46875,
      "learning_rate": 1.8925628412418042e-05,
      "loss": 1.8676,
      "step": 415
    },
    {
      "epoch": 0.3844731977818854,
      "grad_norm": 2.359375,
      "learning_rate": 1.8918724474324418e-05,
      "loss": 1.748,
      "step": 416
    },
    {
      "epoch": 0.3853974121996303,
      "grad_norm": 2.3125,
      "learning_rate": 1.8911799692341598e-05,
      "loss": 1.7704,
      "step": 417
    },
    {
      "epoch": 0.38632162661737524,
      "grad_norm": 2.5,
      "learning_rate": 1.8904854082653438e-05,
      "loss": 1.8642,
      "step": 418
    },
    {
      "epoch": 0.38724584103512016,
      "grad_norm": 2.640625,
      "learning_rate": 1.8897887661492476e-05,
      "loss": 1.7762,
      "step": 419
    },
    {
      "epoch": 0.38817005545286504,
      "grad_norm": 2.5625,
      "learning_rate": 1.889090044513989e-05,
      "loss": 1.8046,
      "step": 420
    },
    {
      "epoch": 0.38909426987060997,
      "grad_norm": 2.546875,
      "learning_rate": 1.888389244992545e-05,
      "loss": 1.7127,
      "step": 421
    },
    {
      "epoch": 0.3900184842883549,
      "grad_norm": 2.546875,
      "learning_rate": 1.8876863692227492e-05,
      "loss": 1.8352,
      "step": 422
    },
    {
      "epoch": 0.3909426987060998,
      "grad_norm": 2.6875,
      "learning_rate": 1.8869814188472883e-05,
      "loss": 1.8322,
      "step": 423
    },
    {
      "epoch": 0.39186691312384475,
      "grad_norm": 2.28125,
      "learning_rate": 1.8862743955136966e-05,
      "loss": 1.7639,
      "step": 424
    },
    {
      "epoch": 0.3927911275415896,
      "grad_norm": 2.34375,
      "learning_rate": 1.8855653008743533e-05,
      "loss": 1.759,
      "step": 425
    },
    {
      "epoch": 0.39371534195933455,
      "grad_norm": 2.453125,
      "learning_rate": 1.884854136586479e-05,
      "loss": 1.7579,
      "step": 426
    },
    {
      "epoch": 0.3946395563770795,
      "grad_norm": 2.359375,
      "learning_rate": 1.8841409043121306e-05,
      "loss": 1.8499,
      "step": 427
    },
    {
      "epoch": 0.3955637707948244,
      "grad_norm": 2.390625,
      "learning_rate": 1.8834256057181978e-05,
      "loss": 1.7694,
      "step": 428
    },
    {
      "epoch": 0.39648798521256934,
      "grad_norm": 2.40625,
      "learning_rate": 1.882708242476401e-05,
      "loss": 1.8417,
      "step": 429
    },
    {
      "epoch": 0.3974121996303142,
      "grad_norm": 2.375,
      "learning_rate": 1.881988816263284e-05,
      "loss": 1.7308,
      "step": 430
    },
    {
      "epoch": 0.39833641404805914,
      "grad_norm": 2.46875,
      "learning_rate": 1.8812673287602136e-05,
      "loss": 1.7876,
      "step": 431
    },
    {
      "epoch": 0.39926062846580407,
      "grad_norm": 2.40625,
      "learning_rate": 1.8805437816533723e-05,
      "loss": 1.8453,
      "step": 432
    },
    {
      "epoch": 0.400184842883549,
      "grad_norm": 2.40625,
      "learning_rate": 1.879818176633758e-05,
      "loss": 1.8424,
      "step": 433
    },
    {
      "epoch": 0.4011090573012939,
      "grad_norm": 2.359375,
      "learning_rate": 1.879090515397176e-05,
      "loss": 1.7315,
      "step": 434
    },
    {
      "epoch": 0.4020332717190388,
      "grad_norm": 2.40625,
      "learning_rate": 1.8783607996442393e-05,
      "loss": 1.8596,
      "step": 435
    },
    {
      "epoch": 0.4029574861367837,
      "grad_norm": 2.34375,
      "learning_rate": 1.8776290310803616e-05,
      "loss": 1.8376,
      "step": 436
    },
    {
      "epoch": 0.40388170055452866,
      "grad_norm": 2.5,
      "learning_rate": 1.8768952114157535e-05,
      "loss": 1.8336,
      "step": 437
    },
    {
      "epoch": 0.4048059149722736,
      "grad_norm": 2.453125,
      "learning_rate": 1.8761593423654202e-05,
      "loss": 1.8435,
      "step": 438
    },
    {
      "epoch": 0.40573012939001846,
      "grad_norm": 2.28125,
      "learning_rate": 1.8754214256491564e-05,
      "loss": 1.729,
      "step": 439
    },
    {
      "epoch": 0.4066543438077634,
      "grad_norm": 2.3125,
      "learning_rate": 1.8746814629915414e-05,
      "loss": 1.7737,
      "step": 440
    },
    {
      "epoch": 0.4075785582255083,
      "grad_norm": 2.421875,
      "learning_rate": 1.8739394561219378e-05,
      "loss": 1.751,
      "step": 441
    },
    {
      "epoch": 0.40850277264325324,
      "grad_norm": 2.375,
      "learning_rate": 1.8731954067744835e-05,
      "loss": 1.8782,
      "step": 442
    },
    {
      "epoch": 0.40942698706099817,
      "grad_norm": 2.4375,
      "learning_rate": 1.8724493166880924e-05,
      "loss": 1.7271,
      "step": 443
    },
    {
      "epoch": 0.41035120147874304,
      "grad_norm": 2.421875,
      "learning_rate": 1.8717011876064453e-05,
      "loss": 1.8491,
      "step": 444
    },
    {
      "epoch": 0.411275415896488,
      "grad_norm": 2.40625,
      "learning_rate": 1.8709510212779906e-05,
      "loss": 1.7633,
      "step": 445
    },
    {
      "epoch": 0.4121996303142329,
      "grad_norm": 2.40625,
      "learning_rate": 1.870198819455936e-05,
      "loss": 1.7974,
      "step": 446
    },
    {
      "epoch": 0.41312384473197783,
      "grad_norm": 2.421875,
      "learning_rate": 1.8694445838982473e-05,
      "loss": 1.7966,
      "step": 447
    },
    {
      "epoch": 0.41404805914972276,
      "grad_norm": 2.328125,
      "learning_rate": 1.8686883163676434e-05,
      "loss": 1.752,
      "step": 448
    },
    {
      "epoch": 0.41497227356746763,
      "grad_norm": 2.515625,
      "learning_rate": 1.8679300186315918e-05,
      "loss": 1.8489,
      "step": 449
    },
    {
      "epoch": 0.41589648798521256,
      "grad_norm": 2.4375,
      "learning_rate": 1.867169692462305e-05,
      "loss": 1.8105,
      "step": 450
    },
    {
      "epoch": 0.4168207024029575,
      "grad_norm": 2.4375,
      "learning_rate": 1.8664073396367363e-05,
      "loss": 1.7778,
      "step": 451
    },
    {
      "epoch": 0.4177449168207024,
      "grad_norm": 2.3125,
      "learning_rate": 1.8656429619365744e-05,
      "loss": 1.6695,
      "step": 452
    },
    {
      "epoch": 0.41866913123844735,
      "grad_norm": 2.421875,
      "learning_rate": 1.8648765611482417e-05,
      "loss": 1.828,
      "step": 453
    },
    {
      "epoch": 0.4195933456561922,
      "grad_norm": 2.328125,
      "learning_rate": 1.864108139062888e-05,
      "loss": 1.8154,
      "step": 454
    },
    {
      "epoch": 0.42051756007393715,
      "grad_norm": 2.40625,
      "learning_rate": 1.8633376974763874e-05,
      "loss": 1.8107,
      "step": 455
    },
    {
      "epoch": 0.4214417744916821,
      "grad_norm": 2.3125,
      "learning_rate": 1.8625652381893335e-05,
      "loss": 1.8369,
      "step": 456
    },
    {
      "epoch": 0.422365988909427,
      "grad_norm": 2.4375,
      "learning_rate": 1.8617907630070353e-05,
      "loss": 1.8016,
      "step": 457
    },
    {
      "epoch": 0.4232902033271719,
      "grad_norm": 2.328125,
      "learning_rate": 1.861014273739514e-05,
      "loss": 1.7897,
      "step": 458
    },
    {
      "epoch": 0.4242144177449168,
      "grad_norm": 2.4375,
      "learning_rate": 1.8602357722014966e-05,
      "loss": 1.7401,
      "step": 459
    },
    {
      "epoch": 0.42513863216266173,
      "grad_norm": 2.46875,
      "learning_rate": 1.859455260212414e-05,
      "loss": 1.7512,
      "step": 460
    },
    {
      "epoch": 0.42606284658040666,
      "grad_norm": 2.4375,
      "learning_rate": 1.8586727395963952e-05,
      "loss": 1.8365,
      "step": 461
    },
    {
      "epoch": 0.4269870609981516,
      "grad_norm": 2.390625,
      "learning_rate": 1.8578882121822643e-05,
      "loss": 1.817,
      "step": 462
    },
    {
      "epoch": 0.42791127541589646,
      "grad_norm": 2.515625,
      "learning_rate": 1.8571016798035343e-05,
      "loss": 1.714,
      "step": 463
    },
    {
      "epoch": 0.4288354898336414,
      "grad_norm": 2.5,
      "learning_rate": 1.8563131442984045e-05,
      "loss": 1.7522,
      "step": 464
    },
    {
      "epoch": 0.4297597042513863,
      "grad_norm": 2.328125,
      "learning_rate": 1.8555226075097558e-05,
      "loss": 1.7814,
      "step": 465
    },
    {
      "epoch": 0.43068391866913125,
      "grad_norm": 2.359375,
      "learning_rate": 1.8547300712851467e-05,
      "loss": 1.8426,
      "step": 466
    },
    {
      "epoch": 0.4316081330868762,
      "grad_norm": 2.484375,
      "learning_rate": 1.853935537476808e-05,
      "loss": 1.7951,
      "step": 467
    },
    {
      "epoch": 0.43253234750462105,
      "grad_norm": 2.4375,
      "learning_rate": 1.8531390079416386e-05,
      "loss": 1.7889,
      "step": 468
    },
    {
      "epoch": 0.433456561922366,
      "grad_norm": 2.390625,
      "learning_rate": 1.8523404845412028e-05,
      "loss": 1.807,
      "step": 469
    },
    {
      "epoch": 0.4343807763401109,
      "grad_norm": 2.375,
      "learning_rate": 1.8515399691417236e-05,
      "loss": 1.7887,
      "step": 470
    },
    {
      "epoch": 0.43530499075785584,
      "grad_norm": 2.328125,
      "learning_rate": 1.8507374636140806e-05,
      "loss": 1.7585,
      "step": 471
    },
    {
      "epoch": 0.43622920517560076,
      "grad_norm": 2.3125,
      "learning_rate": 1.8499329698338035e-05,
      "loss": 1.8336,
      "step": 472
    },
    {
      "epoch": 0.43715341959334564,
      "grad_norm": 2.5,
      "learning_rate": 1.8491264896810694e-05,
      "loss": 1.8155,
      "step": 473
    },
    {
      "epoch": 0.43807763401109057,
      "grad_norm": 2.4375,
      "learning_rate": 1.8483180250406973e-05,
      "loss": 1.8227,
      "step": 474
    },
    {
      "epoch": 0.4390018484288355,
      "grad_norm": 2.59375,
      "learning_rate": 1.847507577802144e-05,
      "loss": 1.8449,
      "step": 475
    },
    {
      "epoch": 0.4399260628465804,
      "grad_norm": 2.59375,
      "learning_rate": 1.8466951498595005e-05,
      "loss": 1.8046,
      "step": 476
    },
    {
      "epoch": 0.4408502772643253,
      "grad_norm": 2.546875,
      "learning_rate": 1.8458807431114863e-05,
      "loss": 1.8108,
      "step": 477
    },
    {
      "epoch": 0.4417744916820702,
      "grad_norm": 2.328125,
      "learning_rate": 1.8450643594614463e-05,
      "loss": 1.7701,
      "step": 478
    },
    {
      "epoch": 0.44269870609981515,
      "grad_norm": 2.8125,
      "learning_rate": 1.8442460008173447e-05,
      "loss": 1.8811,
      "step": 479
    },
    {
      "epoch": 0.4436229205175601,
      "grad_norm": 2.265625,
      "learning_rate": 1.8434256690917616e-05,
      "loss": 1.7186,
      "step": 480
    },
    {
      "epoch": 0.444547134935305,
      "grad_norm": 2.4375,
      "learning_rate": 1.842603366201889e-05,
      "loss": 1.7922,
      "step": 481
    },
    {
      "epoch": 0.4454713493530499,
      "grad_norm": 2.65625,
      "learning_rate": 1.841779094069525e-05,
      "loss": 1.7742,
      "step": 482
    },
    {
      "epoch": 0.4463955637707948,
      "grad_norm": 2.5,
      "learning_rate": 1.84095285462107e-05,
      "loss": 1.7208,
      "step": 483
    },
    {
      "epoch": 0.44731977818853974,
      "grad_norm": 2.359375,
      "learning_rate": 1.8401246497875238e-05,
      "loss": 1.8605,
      "step": 484
    },
    {
      "epoch": 0.44824399260628467,
      "grad_norm": 2.578125,
      "learning_rate": 1.8392944815044767e-05,
      "loss": 1.9009,
      "step": 485
    },
    {
      "epoch": 0.4491682070240296,
      "grad_norm": 2.671875,
      "learning_rate": 1.8384623517121095e-05,
      "loss": 1.8106,
      "step": 486
    },
    {
      "epoch": 0.45009242144177447,
      "grad_norm": 2.34375,
      "learning_rate": 1.837628262355188e-05,
      "loss": 1.7558,
      "step": 487
    },
    {
      "epoch": 0.4510166358595194,
      "grad_norm": 2.390625,
      "learning_rate": 1.8367922153830556e-05,
      "loss": 1.8173,
      "step": 488
    },
    {
      "epoch": 0.4519408502772643,
      "grad_norm": 2.453125,
      "learning_rate": 1.835954212749632e-05,
      "loss": 1.8143,
      "step": 489
    },
    {
      "epoch": 0.45286506469500926,
      "grad_norm": 2.546875,
      "learning_rate": 1.835114256413408e-05,
      "loss": 1.8251,
      "step": 490
    },
    {
      "epoch": 0.4537892791127542,
      "grad_norm": 2.453125,
      "learning_rate": 1.834272348337439e-05,
      "loss": 1.8316,
      "step": 491
    },
    {
      "epoch": 0.45471349353049906,
      "grad_norm": 2.296875,
      "learning_rate": 1.833428490489343e-05,
      "loss": 1.7902,
      "step": 492
    },
    {
      "epoch": 0.455637707948244,
      "grad_norm": 2.4375,
      "learning_rate": 1.8325826848412945e-05,
      "loss": 1.7923,
      "step": 493
    },
    {
      "epoch": 0.4565619223659889,
      "grad_norm": 2.265625,
      "learning_rate": 1.8317349333700192e-05,
      "loss": 1.6527,
      "step": 494
    },
    {
      "epoch": 0.45748613678373384,
      "grad_norm": 2.359375,
      "learning_rate": 1.830885238056792e-05,
      "loss": 1.8055,
      "step": 495
    },
    {
      "epoch": 0.4584103512014787,
      "grad_norm": 2.40625,
      "learning_rate": 1.8300336008874297e-05,
      "loss": 1.7207,
      "step": 496
    },
    {
      "epoch": 0.45933456561922364,
      "grad_norm": 2.359375,
      "learning_rate": 1.8291800238522872e-05,
      "loss": 1.8763,
      "step": 497
    },
    {
      "epoch": 0.4602587800369686,
      "grad_norm": 2.515625,
      "learning_rate": 1.828324508946254e-05,
      "loss": 1.7244,
      "step": 498
    },
    {
      "epoch": 0.4611829944547135,
      "grad_norm": 2.375,
      "learning_rate": 1.827467058168748e-05,
      "loss": 1.8059,
      "step": 499
    },
    {
      "epoch": 0.46210720887245843,
      "grad_norm": 2.25,
      "learning_rate": 1.8266076735237113e-05,
      "loss": 1.7651,
      "step": 500
    },
    {
      "epoch": 0.4630314232902033,
      "grad_norm": 2.375,
      "learning_rate": 1.825746357019606e-05,
      "loss": 1.8413,
      "step": 501
    },
    {
      "epoch": 0.46395563770794823,
      "grad_norm": 2.296875,
      "learning_rate": 1.8248831106694088e-05,
      "loss": 1.7642,
      "step": 502
    },
    {
      "epoch": 0.46487985212569316,
      "grad_norm": 2.328125,
      "learning_rate": 1.8240179364906064e-05,
      "loss": 1.7855,
      "step": 503
    },
    {
      "epoch": 0.4658040665434381,
      "grad_norm": 2.453125,
      "learning_rate": 1.8231508365051924e-05,
      "loss": 1.7445,
      "step": 504
    },
    {
      "epoch": 0.466728280961183,
      "grad_norm": 2.4375,
      "learning_rate": 1.822281812739659e-05,
      "loss": 1.8306,
      "step": 505
    },
    {
      "epoch": 0.4676524953789279,
      "grad_norm": 2.421875,
      "learning_rate": 1.8214108672249963e-05,
      "loss": 1.7766,
      "step": 506
    },
    {
      "epoch": 0.4685767097966728,
      "grad_norm": 2.5625,
      "learning_rate": 1.820538001996685e-05,
      "loss": 1.833,
      "step": 507
    },
    {
      "epoch": 0.46950092421441775,
      "grad_norm": 2.375,
      "learning_rate": 1.819663219094692e-05,
      "loss": 1.7733,
      "step": 508
    },
    {
      "epoch": 0.4704251386321627,
      "grad_norm": 2.609375,
      "learning_rate": 1.818786520563467e-05,
      "loss": 1.7451,
      "step": 509
    },
    {
      "epoch": 0.4713493530499076,
      "grad_norm": 2.515625,
      "learning_rate": 1.8179079084519356e-05,
      "loss": 1.915,
      "step": 510
    },
    {
      "epoch": 0.4722735674676525,
      "grad_norm": 2.34375,
      "learning_rate": 1.8170273848134958e-05,
      "loss": 1.7984,
      "step": 511
    },
    {
      "epoch": 0.4731977818853974,
      "grad_norm": 2.40625,
      "learning_rate": 1.8161449517060142e-05,
      "loss": 1.7826,
      "step": 512
    },
    {
      "epoch": 0.47412199630314233,
      "grad_norm": 2.515625,
      "learning_rate": 1.815260611191818e-05,
      "loss": 1.8685,
      "step": 513
    },
    {
      "epoch": 0.47504621072088726,
      "grad_norm": 2.40625,
      "learning_rate": 1.8143743653376944e-05,
      "loss": 1.7808,
      "step": 514
    },
    {
      "epoch": 0.47597042513863214,
      "grad_norm": 2.34375,
      "learning_rate": 1.8134862162148816e-05,
      "loss": 1.7452,
      "step": 515
    },
    {
      "epoch": 0.47689463955637706,
      "grad_norm": 2.296875,
      "learning_rate": 1.8125961658990673e-05,
      "loss": 1.7051,
      "step": 516
    },
    {
      "epoch": 0.477818853974122,
      "grad_norm": 2.46875,
      "learning_rate": 1.8117042164703816e-05,
      "loss": 1.8248,
      "step": 517
    },
    {
      "epoch": 0.4787430683918669,
      "grad_norm": 2.53125,
      "learning_rate": 1.8108103700133936e-05,
      "loss": 1.8173,
      "step": 518
    },
    {
      "epoch": 0.47966728280961185,
      "grad_norm": 2.609375,
      "learning_rate": 1.809914628617105e-05,
      "loss": 1.7394,
      "step": 519
    },
    {
      "epoch": 0.4805914972273567,
      "grad_norm": 2.328125,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 1.8523,
      "step": 520
    },
    {
      "epoch": 0.48151571164510165,
      "grad_norm": 2.328125,
      "learning_rate": 1.8081174693847758e-05,
      "loss": 1.8535,
      "step": 521
    },
    {
      "epoch": 0.4824399260628466,
      "grad_norm": 2.390625,
      "learning_rate": 1.8072160557488632e-05,
      "loss": 1.8176,
      "step": 522
    },
    {
      "epoch": 0.4833641404805915,
      "grad_norm": 2.484375,
      "learning_rate": 1.806312755573898e-05,
      "loss": 1.8457,
      "step": 523
    },
    {
      "epoch": 0.48428835489833644,
      "grad_norm": 2.546875,
      "learning_rate": 1.805407570970976e-05,
      "loss": 1.8794,
      "step": 524
    },
    {
      "epoch": 0.4852125693160813,
      "grad_norm": 2.328125,
      "learning_rate": 1.804500504055598e-05,
      "loss": 1.7496,
      "step": 525
    },
    {
      "epoch": 0.48613678373382624,
      "grad_norm": 2.390625,
      "learning_rate": 1.8035915569476638e-05,
      "loss": 1.8289,
      "step": 526
    },
    {
      "epoch": 0.48706099815157117,
      "grad_norm": 2.46875,
      "learning_rate": 1.802680731771468e-05,
      "loss": 1.7759,
      "step": 527
    },
    {
      "epoch": 0.4879852125693161,
      "grad_norm": 2.328125,
      "learning_rate": 1.8017680306556928e-05,
      "loss": 1.7268,
      "step": 528
    },
    {
      "epoch": 0.488909426987061,
      "grad_norm": 2.515625,
      "learning_rate": 1.8008534557334064e-05,
      "loss": 1.8667,
      "step": 529
    },
    {
      "epoch": 0.4898336414048059,
      "grad_norm": 2.484375,
      "learning_rate": 1.7999370091420554e-05,
      "loss": 1.8681,
      "step": 530
    },
    {
      "epoch": 0.4907578558225508,
      "grad_norm": 2.3125,
      "learning_rate": 1.7990186930234606e-05,
      "loss": 1.7786,
      "step": 531
    },
    {
      "epoch": 0.49168207024029575,
      "grad_norm": 2.421875,
      "learning_rate": 1.7980985095238123e-05,
      "loss": 1.7993,
      "step": 532
    },
    {
      "epoch": 0.4926062846580407,
      "grad_norm": 2.484375,
      "learning_rate": 1.797176460793665e-05,
      "loss": 1.8106,
      "step": 533
    },
    {
      "epoch": 0.49353049907578556,
      "grad_norm": 2.40625,
      "learning_rate": 1.7962525489879324e-05,
      "loss": 1.7916,
      "step": 534
    },
    {
      "epoch": 0.4944547134935305,
      "grad_norm": 2.4375,
      "learning_rate": 1.7953267762658826e-05,
      "loss": 1.7595,
      "step": 535
    },
    {
      "epoch": 0.4953789279112754,
      "grad_norm": 2.375,
      "learning_rate": 1.7943991447911322e-05,
      "loss": 1.7203,
      "step": 536
    },
    {
      "epoch": 0.49630314232902034,
      "grad_norm": 2.46875,
      "learning_rate": 1.7934696567316426e-05,
      "loss": 1.7498,
      "step": 537
    },
    {
      "epoch": 0.49722735674676527,
      "grad_norm": 2.546875,
      "learning_rate": 1.7925383142597136e-05,
      "loss": 1.826,
      "step": 538
    },
    {
      "epoch": 0.49815157116451014,
      "grad_norm": 2.546875,
      "learning_rate": 1.7916051195519797e-05,
      "loss": 1.8746,
      "step": 539
    },
    {
      "epoch": 0.49907578558225507,
      "grad_norm": 2.328125,
      "learning_rate": 1.7906700747894035e-05,
      "loss": 1.7825,
      "step": 540
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.53125,
      "learning_rate": 1.7897331821572716e-05,
      "loss": 1.7949,
      "step": 541
    },
    {
      "epoch": 0.5009242144177449,
      "grad_norm": 2.421875,
      "learning_rate": 1.7887944438451893e-05,
      "loss": 1.705,
      "step": 542
    },
    {
      "epoch": 0.5018484288354899,
      "grad_norm": 2.453125,
      "learning_rate": 1.7878538620470755e-05,
      "loss": 1.8534,
      "step": 543
    },
    {
      "epoch": 0.5027726432532348,
      "grad_norm": 2.484375,
      "learning_rate": 1.7869114389611574e-05,
      "loss": 1.7709,
      "step": 544
    },
    {
      "epoch": 0.5036968576709797,
      "grad_norm": 2.46875,
      "learning_rate": 1.7859671767899655e-05,
      "loss": 1.798,
      "step": 545
    },
    {
      "epoch": 0.5046210720887245,
      "grad_norm": 2.296875,
      "learning_rate": 1.7850210777403283e-05,
      "loss": 1.8015,
      "step": 546
    },
    {
      "epoch": 0.5055452865064695,
      "grad_norm": 2.390625,
      "learning_rate": 1.7840731440233673e-05,
      "loss": 1.7518,
      "step": 547
    },
    {
      "epoch": 0.5064695009242144,
      "grad_norm": 2.546875,
      "learning_rate": 1.783123377854492e-05,
      "loss": 1.8501,
      "step": 548
    },
    {
      "epoch": 0.5073937153419593,
      "grad_norm": 2.390625,
      "learning_rate": 1.7821717814533942e-05,
      "loss": 1.8816,
      "step": 549
    },
    {
      "epoch": 0.5083179297597042,
      "grad_norm": 2.34375,
      "learning_rate": 1.781218357044043e-05,
      "loss": 1.7582,
      "step": 550
    },
    {
      "epoch": 0.5092421441774492,
      "grad_norm": 2.4375,
      "learning_rate": 1.7802631068546805e-05,
      "loss": 1.7972,
      "step": 551
    },
    {
      "epoch": 0.5101663585951941,
      "grad_norm": 2.40625,
      "learning_rate": 1.779306033117815e-05,
      "loss": 1.8232,
      "step": 552
    },
    {
      "epoch": 0.511090573012939,
      "grad_norm": 2.296875,
      "learning_rate": 1.7783471380702167e-05,
      "loss": 1.7349,
      "step": 553
    },
    {
      "epoch": 0.512014787430684,
      "grad_norm": 2.328125,
      "learning_rate": 1.777386423952913e-05,
      "loss": 1.786,
      "step": 554
    },
    {
      "epoch": 0.5129390018484289,
      "grad_norm": 2.46875,
      "learning_rate": 1.7764238930111825e-05,
      "loss": 1.8225,
      "step": 555
    },
    {
      "epoch": 0.5138632162661737,
      "grad_norm": 2.375,
      "learning_rate": 1.7754595474945486e-05,
      "loss": 1.8403,
      "step": 556
    },
    {
      "epoch": 0.5147874306839186,
      "grad_norm": 2.25,
      "learning_rate": 1.7744933896567773e-05,
      "loss": 1.8103,
      "step": 557
    },
    {
      "epoch": 0.5157116451016636,
      "grad_norm": 2.25,
      "learning_rate": 1.773525421755869e-05,
      "loss": 1.7873,
      "step": 558
    },
    {
      "epoch": 0.5166358595194085,
      "grad_norm": 2.3125,
      "learning_rate": 1.7725556460540553e-05,
      "loss": 1.7976,
      "step": 559
    },
    {
      "epoch": 0.5175600739371534,
      "grad_norm": 2.265625,
      "learning_rate": 1.7715840648177916e-05,
      "loss": 1.7229,
      "step": 560
    },
    {
      "epoch": 0.5184842883548983,
      "grad_norm": 2.8125,
      "learning_rate": 1.7706106803177538e-05,
      "loss": 1.903,
      "step": 561
    },
    {
      "epoch": 0.5194085027726433,
      "grad_norm": 2.359375,
      "learning_rate": 1.769635494828833e-05,
      "loss": 1.8074,
      "step": 562
    },
    {
      "epoch": 0.5203327171903882,
      "grad_norm": 2.515625,
      "learning_rate": 1.768658510630127e-05,
      "loss": 1.781,
      "step": 563
    },
    {
      "epoch": 0.5212569316081331,
      "grad_norm": 2.4375,
      "learning_rate": 1.7676797300049395e-05,
      "loss": 1.8054,
      "step": 564
    },
    {
      "epoch": 0.522181146025878,
      "grad_norm": 2.515625,
      "learning_rate": 1.7666991552407723e-05,
      "loss": 1.7849,
      "step": 565
    },
    {
      "epoch": 0.5231053604436229,
      "grad_norm": 2.265625,
      "learning_rate": 1.7657167886293196e-05,
      "loss": 1.7184,
      "step": 566
    },
    {
      "epoch": 0.5240295748613678,
      "grad_norm": 2.5625,
      "learning_rate": 1.7647326324664633e-05,
      "loss": 1.7871,
      "step": 567
    },
    {
      "epoch": 0.5249537892791127,
      "grad_norm": 2.703125,
      "learning_rate": 1.763746689052268e-05,
      "loss": 1.7195,
      "step": 568
    },
    {
      "epoch": 0.5258780036968577,
      "grad_norm": 2.46875,
      "learning_rate": 1.7627589606909756e-05,
      "loss": 1.7034,
      "step": 569
    },
    {
      "epoch": 0.5268022181146026,
      "grad_norm": 2.359375,
      "learning_rate": 1.7617694496909994e-05,
      "loss": 1.7465,
      "step": 570
    },
    {
      "epoch": 0.5277264325323475,
      "grad_norm": 2.5,
      "learning_rate": 1.760778158364918e-05,
      "loss": 1.789,
      "step": 571
    },
    {
      "epoch": 0.5286506469500925,
      "grad_norm": 3.015625,
      "learning_rate": 1.7597850890294718e-05,
      "loss": 1.7155,
      "step": 572
    },
    {
      "epoch": 0.5295748613678374,
      "grad_norm": 2.546875,
      "learning_rate": 1.758790244005556e-05,
      "loss": 1.7861,
      "step": 573
    },
    {
      "epoch": 0.5304990757855823,
      "grad_norm": 2.25,
      "learning_rate": 1.757793625618217e-05,
      "loss": 1.7542,
      "step": 574
    },
    {
      "epoch": 0.5314232902033271,
      "grad_norm": 2.4375,
      "learning_rate": 1.7567952361966435e-05,
      "loss": 1.7936,
      "step": 575
    },
    {
      "epoch": 0.532347504621072,
      "grad_norm": 2.5625,
      "learning_rate": 1.7557950780741652e-05,
      "loss": 1.7876,
      "step": 576
    },
    {
      "epoch": 0.533271719038817,
      "grad_norm": 2.609375,
      "learning_rate": 1.7547931535882445e-05,
      "loss": 1.8651,
      "step": 577
    },
    {
      "epoch": 0.5341959334565619,
      "grad_norm": 2.359375,
      "learning_rate": 1.7537894650804725e-05,
      "loss": 1.834,
      "step": 578
    },
    {
      "epoch": 0.5351201478743068,
      "grad_norm": 2.3125,
      "learning_rate": 1.752784014896562e-05,
      "loss": 1.7637,
      "step": 579
    },
    {
      "epoch": 0.5360443622920518,
      "grad_norm": 2.375,
      "learning_rate": 1.751776805386344e-05,
      "loss": 1.8324,
      "step": 580
    },
    {
      "epoch": 0.5369685767097967,
      "grad_norm": 2.203125,
      "learning_rate": 1.7507678389037613e-05,
      "loss": 1.7368,
      "step": 581
    },
    {
      "epoch": 0.5378927911275416,
      "grad_norm": 2.53125,
      "learning_rate": 1.749757117806862e-05,
      "loss": 1.7988,
      "step": 582
    },
    {
      "epoch": 0.5388170055452866,
      "grad_norm": 2.46875,
      "learning_rate": 1.7487446444577955e-05,
      "loss": 1.7895,
      "step": 583
    },
    {
      "epoch": 0.5397412199630314,
      "grad_norm": 2.53125,
      "learning_rate": 1.747730421222806e-05,
      "loss": 1.8359,
      "step": 584
    },
    {
      "epoch": 0.5406654343807763,
      "grad_norm": 2.59375,
      "learning_rate": 1.746714450472228e-05,
      "loss": 1.7945,
      "step": 585
    },
    {
      "epoch": 0.5415896487985212,
      "grad_norm": 2.359375,
      "learning_rate": 1.7456967345804794e-05,
      "loss": 1.8113,
      "step": 586
    },
    {
      "epoch": 0.5425138632162662,
      "grad_norm": 2.34375,
      "learning_rate": 1.744677275926057e-05,
      "loss": 1.7825,
      "step": 587
    },
    {
      "epoch": 0.5434380776340111,
      "grad_norm": 2.515625,
      "learning_rate": 1.7436560768915312e-05,
      "loss": 1.7549,
      "step": 588
    },
    {
      "epoch": 0.544362292051756,
      "grad_norm": 2.3125,
      "learning_rate": 1.742633139863538e-05,
      "loss": 1.7137,
      "step": 589
    },
    {
      "epoch": 0.5452865064695009,
      "grad_norm": 2.28125,
      "learning_rate": 1.7416084672327774e-05,
      "loss": 1.7222,
      "step": 590
    },
    {
      "epoch": 0.5462107208872459,
      "grad_norm": 2.53125,
      "learning_rate": 1.7405820613940044e-05,
      "loss": 1.8542,
      "step": 591
    },
    {
      "epoch": 0.5471349353049908,
      "grad_norm": 2.25,
      "learning_rate": 1.7395539247460252e-05,
      "loss": 1.7546,
      "step": 592
    },
    {
      "epoch": 0.5480591497227357,
      "grad_norm": 2.359375,
      "learning_rate": 1.7385240596916905e-05,
      "loss": 1.6853,
      "step": 593
    },
    {
      "epoch": 0.5489833641404805,
      "grad_norm": 2.328125,
      "learning_rate": 1.7374924686378906e-05,
      "loss": 1.8847,
      "step": 594
    },
    {
      "epoch": 0.5499075785582255,
      "grad_norm": 2.375,
      "learning_rate": 1.7364591539955503e-05,
      "loss": 1.84,
      "step": 595
    },
    {
      "epoch": 0.5508317929759704,
      "grad_norm": 2.265625,
      "learning_rate": 1.7354241181796218e-05,
      "loss": 1.715,
      "step": 596
    },
    {
      "epoch": 0.5517560073937153,
      "grad_norm": 2.53125,
      "learning_rate": 1.73438736360908e-05,
      "loss": 1.7637,
      "step": 597
    },
    {
      "epoch": 0.5526802218114603,
      "grad_norm": 2.40625,
      "learning_rate": 1.733348892706918e-05,
      "loss": 1.7145,
      "step": 598
    },
    {
      "epoch": 0.5536044362292052,
      "grad_norm": 2.453125,
      "learning_rate": 1.7323087079001372e-05,
      "loss": 1.7938,
      "step": 599
    },
    {
      "epoch": 0.5545286506469501,
      "grad_norm": 2.296875,
      "learning_rate": 1.731266811619747e-05,
      "loss": 1.7954,
      "step": 600
    },
    {
      "epoch": 0.555452865064695,
      "grad_norm": 2.5,
      "learning_rate": 1.730223206300757e-05,
      "loss": 1.6671,
      "step": 601
    },
    {
      "epoch": 0.55637707948244,
      "grad_norm": 2.578125,
      "learning_rate": 1.729177894382168e-05,
      "loss": 1.8419,
      "step": 602
    },
    {
      "epoch": 0.5573012939001848,
      "grad_norm": 2.296875,
      "learning_rate": 1.7281308783069727e-05,
      "loss": 1.7749,
      "step": 603
    },
    {
      "epoch": 0.5582255083179297,
      "grad_norm": 2.40625,
      "learning_rate": 1.7270821605221448e-05,
      "loss": 1.8182,
      "step": 604
    },
    {
      "epoch": 0.5591497227356746,
      "grad_norm": 2.515625,
      "learning_rate": 1.7260317434786357e-05,
      "loss": 1.8423,
      "step": 605
    },
    {
      "epoch": 0.5600739371534196,
      "grad_norm": 2.484375,
      "learning_rate": 1.7249796296313668e-05,
      "loss": 1.7371,
      "step": 606
    },
    {
      "epoch": 0.5609981515711645,
      "grad_norm": 2.328125,
      "learning_rate": 1.723925821439227e-05,
      "loss": 1.8257,
      "step": 607
    },
    {
      "epoch": 0.5619223659889094,
      "grad_norm": 2.296875,
      "learning_rate": 1.7228703213650635e-05,
      "loss": 1.7477,
      "step": 608
    },
    {
      "epoch": 0.5628465804066544,
      "grad_norm": 2.28125,
      "learning_rate": 1.721813131875679e-05,
      "loss": 1.77,
      "step": 609
    },
    {
      "epoch": 0.5637707948243993,
      "grad_norm": 2.359375,
      "learning_rate": 1.720754255441823e-05,
      "loss": 1.7389,
      "step": 610
    },
    {
      "epoch": 0.5646950092421442,
      "grad_norm": 2.421875,
      "learning_rate": 1.7196936945381885e-05,
      "loss": 1.8286,
      "step": 611
    },
    {
      "epoch": 0.5656192236598891,
      "grad_norm": 2.390625,
      "learning_rate": 1.718631451643405e-05,
      "loss": 1.7894,
      "step": 612
    },
    {
      "epoch": 0.566543438077634,
      "grad_norm": 2.390625,
      "learning_rate": 1.717567529240033e-05,
      "loss": 1.8781,
      "step": 613
    },
    {
      "epoch": 0.5674676524953789,
      "grad_norm": 2.484375,
      "learning_rate": 1.7165019298145587e-05,
      "loss": 1.799,
      "step": 614
    },
    {
      "epoch": 0.5683918669131238,
      "grad_norm": 2.3125,
      "learning_rate": 1.7154346558573863e-05,
      "loss": 1.8503,
      "step": 615
    },
    {
      "epoch": 0.5693160813308688,
      "grad_norm": 2.28125,
      "learning_rate": 1.7143657098628353e-05,
      "loss": 1.7614,
      "step": 616
    },
    {
      "epoch": 0.5702402957486137,
      "grad_norm": 2.265625,
      "learning_rate": 1.713295094329131e-05,
      "loss": 1.7831,
      "step": 617
    },
    {
      "epoch": 0.5711645101663586,
      "grad_norm": 2.328125,
      "learning_rate": 1.712222811758402e-05,
      "loss": 1.8115,
      "step": 618
    },
    {
      "epoch": 0.5720887245841035,
      "grad_norm": 2.40625,
      "learning_rate": 1.7111488646566728e-05,
      "loss": 1.7405,
      "step": 619
    },
    {
      "epoch": 0.5730129390018485,
      "grad_norm": 2.328125,
      "learning_rate": 1.7100732555338573e-05,
      "loss": 1.9411,
      "step": 620
    },
    {
      "epoch": 0.5739371534195934,
      "grad_norm": 2.28125,
      "learning_rate": 1.7089959869037544e-05,
      "loss": 1.7312,
      "step": 621
    },
    {
      "epoch": 0.5748613678373382,
      "grad_norm": 2.328125,
      "learning_rate": 1.7079170612840404e-05,
      "loss": 1.7759,
      "step": 622
    },
    {
      "epoch": 0.5757855822550831,
      "grad_norm": 2.40625,
      "learning_rate": 1.7068364811962662e-05,
      "loss": 1.7094,
      "step": 623
    },
    {
      "epoch": 0.5767097966728281,
      "grad_norm": 2.390625,
      "learning_rate": 1.7057542491658467e-05,
      "loss": 1.7928,
      "step": 624
    },
    {
      "epoch": 0.577634011090573,
      "grad_norm": 2.265625,
      "learning_rate": 1.70467036772206e-05,
      "loss": 1.7901,
      "step": 625
    },
    {
      "epoch": 0.5785582255083179,
      "grad_norm": 2.328125,
      "learning_rate": 1.7035848393980372e-05,
      "loss": 1.8203,
      "step": 626
    },
    {
      "epoch": 0.5794824399260629,
      "grad_norm": 2.265625,
      "learning_rate": 1.7024976667307594e-05,
      "loss": 1.7715,
      "step": 627
    },
    {
      "epoch": 0.5804066543438078,
      "grad_norm": 2.328125,
      "learning_rate": 1.70140885226105e-05,
      "loss": 1.9386,
      "step": 628
    },
    {
      "epoch": 0.5813308687615527,
      "grad_norm": 2.546875,
      "learning_rate": 1.70031839853357e-05,
      "loss": 1.8386,
      "step": 629
    },
    {
      "epoch": 0.5822550831792976,
      "grad_norm": 2.40625,
      "learning_rate": 1.6992263080968112e-05,
      "loss": 1.7751,
      "step": 630
    },
    {
      "epoch": 0.5831792975970426,
      "grad_norm": 2.359375,
      "learning_rate": 1.6981325835030905e-05,
      "loss": 1.7702,
      "step": 631
    },
    {
      "epoch": 0.5841035120147874,
      "grad_norm": 2.390625,
      "learning_rate": 1.697037227308544e-05,
      "loss": 1.8312,
      "step": 632
    },
    {
      "epoch": 0.5850277264325323,
      "grad_norm": 2.375,
      "learning_rate": 1.695940242073122e-05,
      "loss": 1.809,
      "step": 633
    },
    {
      "epoch": 0.5859519408502772,
      "grad_norm": 2.390625,
      "learning_rate": 1.6948416303605796e-05,
      "loss": 1.7757,
      "step": 634
    },
    {
      "epoch": 0.5868761552680222,
      "grad_norm": 2.484375,
      "learning_rate": 1.6937413947384756e-05,
      "loss": 1.7591,
      "step": 635
    },
    {
      "epoch": 0.5878003696857671,
      "grad_norm": 2.28125,
      "learning_rate": 1.692639537778163e-05,
      "loss": 1.8168,
      "step": 636
    },
    {
      "epoch": 0.588724584103512,
      "grad_norm": 2.21875,
      "learning_rate": 1.691536062054783e-05,
      "loss": 1.8131,
      "step": 637
    },
    {
      "epoch": 0.589648798521257,
      "grad_norm": 2.484375,
      "learning_rate": 1.6904309701472627e-05,
      "loss": 1.7659,
      "step": 638
    },
    {
      "epoch": 0.5905730129390019,
      "grad_norm": 2.328125,
      "learning_rate": 1.689324264638304e-05,
      "loss": 1.7739,
      "step": 639
    },
    {
      "epoch": 0.5914972273567468,
      "grad_norm": 2.328125,
      "learning_rate": 1.68821594811438e-05,
      "loss": 1.792,
      "step": 640
    },
    {
      "epoch": 0.5924214417744916,
      "grad_norm": 2.25,
      "learning_rate": 1.6871060231657306e-05,
      "loss": 1.7787,
      "step": 641
    },
    {
      "epoch": 0.5933456561922366,
      "grad_norm": 2.296875,
      "learning_rate": 1.6859944923863536e-05,
      "loss": 1.8021,
      "step": 642
    },
    {
      "epoch": 0.5942698706099815,
      "grad_norm": 2.40625,
      "learning_rate": 1.6848813583739995e-05,
      "loss": 1.698,
      "step": 643
    },
    {
      "epoch": 0.5951940850277264,
      "grad_norm": 2.265625,
      "learning_rate": 1.683766623730166e-05,
      "loss": 1.7306,
      "step": 644
    },
    {
      "epoch": 0.5961182994454713,
      "grad_norm": 2.3125,
      "learning_rate": 1.6826502910600924e-05,
      "loss": 1.8539,
      "step": 645
    },
    {
      "epoch": 0.5970425138632163,
      "grad_norm": 2.40625,
      "learning_rate": 1.6815323629727513e-05,
      "loss": 1.8593,
      "step": 646
    },
    {
      "epoch": 0.5979667282809612,
      "grad_norm": 2.390625,
      "learning_rate": 1.680412842080845e-05,
      "loss": 1.8015,
      "step": 647
    },
    {
      "epoch": 0.5988909426987061,
      "grad_norm": 2.484375,
      "learning_rate": 1.679291731000798e-05,
      "loss": 1.7695,
      "step": 648
    },
    {
      "epoch": 0.5998151571164511,
      "grad_norm": 2.359375,
      "learning_rate": 1.6781690323527512e-05,
      "loss": 1.882,
      "step": 649
    },
    {
      "epoch": 0.600739371534196,
      "grad_norm": 2.21875,
      "learning_rate": 1.6770447487605558e-05,
      "loss": 1.7352,
      "step": 650
    },
    {
      "epoch": 0.6016635859519408,
      "grad_norm": 2.390625,
      "learning_rate": 1.6759188828517672e-05,
      "loss": 1.6697,
      "step": 651
    },
    {
      "epoch": 0.6025878003696857,
      "grad_norm": 2.25,
      "learning_rate": 1.6747914372576394e-05,
      "loss": 1.7724,
      "step": 652
    },
    {
      "epoch": 0.6035120147874307,
      "grad_norm": 2.296875,
      "learning_rate": 1.6736624146131165e-05,
      "loss": 1.7867,
      "step": 653
    },
    {
      "epoch": 0.6044362292051756,
      "grad_norm": 2.3125,
      "learning_rate": 1.6725318175568307e-05,
      "loss": 1.7952,
      "step": 654
    },
    {
      "epoch": 0.6053604436229205,
      "grad_norm": 2.25,
      "learning_rate": 1.6713996487310918e-05,
      "loss": 1.7732,
      "step": 655
    },
    {
      "epoch": 0.6062846580406654,
      "grad_norm": 2.265625,
      "learning_rate": 1.670265910781884e-05,
      "loss": 1.7278,
      "step": 656
    },
    {
      "epoch": 0.6072088724584104,
      "grad_norm": 2.203125,
      "learning_rate": 1.6691306063588583e-05,
      "loss": 1.7756,
      "step": 657
    },
    {
      "epoch": 0.6081330868761553,
      "grad_norm": 2.296875,
      "learning_rate": 1.667993738115327e-05,
      "loss": 1.7691,
      "step": 658
    },
    {
      "epoch": 0.6090573012939002,
      "grad_norm": 2.390625,
      "learning_rate": 1.666855308708257e-05,
      "loss": 1.7943,
      "step": 659
    },
    {
      "epoch": 0.609981515711645,
      "grad_norm": 2.6875,
      "learning_rate": 1.6657153207982633e-05,
      "loss": 1.872,
      "step": 660
    },
    {
      "epoch": 0.61090573012939,
      "grad_norm": 2.359375,
      "learning_rate": 1.6645737770496043e-05,
      "loss": 1.757,
      "step": 661
    },
    {
      "epoch": 0.6118299445471349,
      "grad_norm": 2.390625,
      "learning_rate": 1.6634306801301737e-05,
      "loss": 1.8742,
      "step": 662
    },
    {
      "epoch": 0.6127541589648798,
      "grad_norm": 2.328125,
      "learning_rate": 1.6622860327114953e-05,
      "loss": 1.8074,
      "step": 663
    },
    {
      "epoch": 0.6136783733826248,
      "grad_norm": 2.328125,
      "learning_rate": 1.6611398374687172e-05,
      "loss": 1.753,
      "step": 664
    },
    {
      "epoch": 0.6146025878003697,
      "grad_norm": 2.375,
      "learning_rate": 1.6599920970806035e-05,
      "loss": 1.7479,
      "step": 665
    },
    {
      "epoch": 0.6155268022181146,
      "grad_norm": 2.390625,
      "learning_rate": 1.6588428142295313e-05,
      "loss": 1.8057,
      "step": 666
    },
    {
      "epoch": 0.6164510166358595,
      "grad_norm": 2.296875,
      "learning_rate": 1.6576919916014808e-05,
      "loss": 1.7698,
      "step": 667
    },
    {
      "epoch": 0.6173752310536045,
      "grad_norm": 2.21875,
      "learning_rate": 1.656539631886032e-05,
      "loss": 1.6938,
      "step": 668
    },
    {
      "epoch": 0.6182994454713494,
      "grad_norm": 2.203125,
      "learning_rate": 1.6553857377763565e-05,
      "loss": 1.6874,
      "step": 669
    },
    {
      "epoch": 0.6192236598890942,
      "grad_norm": 2.40625,
      "learning_rate": 1.654230311969213e-05,
      "loss": 1.8034,
      "step": 670
    },
    {
      "epoch": 0.6201478743068392,
      "grad_norm": 2.34375,
      "learning_rate": 1.653073357164938e-05,
      "loss": 1.7782,
      "step": 671
    },
    {
      "epoch": 0.6210720887245841,
      "grad_norm": 2.28125,
      "learning_rate": 1.6519148760674435e-05,
      "loss": 1.7162,
      "step": 672
    },
    {
      "epoch": 0.621996303142329,
      "grad_norm": 2.375,
      "learning_rate": 1.650754871384207e-05,
      "loss": 1.8538,
      "step": 673
    },
    {
      "epoch": 0.6229205175600739,
      "grad_norm": 2.28125,
      "learning_rate": 1.649593345826268e-05,
      "loss": 1.7061,
      "step": 674
    },
    {
      "epoch": 0.6238447319778189,
      "grad_norm": 2.3125,
      "learning_rate": 1.648430302108219e-05,
      "loss": 1.7933,
      "step": 675
    },
    {
      "epoch": 0.6247689463955638,
      "grad_norm": 2.296875,
      "learning_rate": 1.6472657429482026e-05,
      "loss": 1.774,
      "step": 676
    },
    {
      "epoch": 0.6256931608133087,
      "grad_norm": 2.296875,
      "learning_rate": 1.6460996710679013e-05,
      "loss": 1.8082,
      "step": 677
    },
    {
      "epoch": 0.6266173752310537,
      "grad_norm": 2.34375,
      "learning_rate": 1.6449320891925334e-05,
      "loss": 1.767,
      "step": 678
    },
    {
      "epoch": 0.6275415896487985,
      "grad_norm": 2.328125,
      "learning_rate": 1.6437630000508466e-05,
      "loss": 1.7577,
      "step": 679
    },
    {
      "epoch": 0.6284658040665434,
      "grad_norm": 2.203125,
      "learning_rate": 1.6425924063751112e-05,
      "loss": 1.7203,
      "step": 680
    },
    {
      "epoch": 0.6293900184842883,
      "grad_norm": 2.3125,
      "learning_rate": 1.6414203109011132e-05,
      "loss": 1.7882,
      "step": 681
    },
    {
      "epoch": 0.6303142329020333,
      "grad_norm": 2.265625,
      "learning_rate": 1.640246716368149e-05,
      "loss": 1.7407,
      "step": 682
    },
    {
      "epoch": 0.6312384473197782,
      "grad_norm": 2.3125,
      "learning_rate": 1.6390716255190186e-05,
      "loss": 1.8388,
      "step": 683
    },
    {
      "epoch": 0.6321626617375231,
      "grad_norm": 2.390625,
      "learning_rate": 1.6378950411000183e-05,
      "loss": 1.7253,
      "step": 684
    },
    {
      "epoch": 0.633086876155268,
      "grad_norm": 2.28125,
      "learning_rate": 1.6367169658609353e-05,
      "loss": 1.7185,
      "step": 685
    },
    {
      "epoch": 0.634011090573013,
      "grad_norm": 2.28125,
      "learning_rate": 1.635537402555042e-05,
      "loss": 1.8591,
      "step": 686
    },
    {
      "epoch": 0.6349353049907579,
      "grad_norm": 2.234375,
      "learning_rate": 1.634356353939086e-05,
      "loss": 1.764,
      "step": 687
    },
    {
      "epoch": 0.6358595194085028,
      "grad_norm": 2.515625,
      "learning_rate": 1.633173822773289e-05,
      "loss": 1.8386,
      "step": 688
    },
    {
      "epoch": 0.6367837338262476,
      "grad_norm": 2.390625,
      "learning_rate": 1.6319898118213365e-05,
      "loss": 1.7715,
      "step": 689
    },
    {
      "epoch": 0.6377079482439926,
      "grad_norm": 2.484375,
      "learning_rate": 1.6308043238503717e-05,
      "loss": 1.8002,
      "step": 690
    },
    {
      "epoch": 0.6386321626617375,
      "grad_norm": 2.3125,
      "learning_rate": 1.6296173616309908e-05,
      "loss": 1.8008,
      "step": 691
    },
    {
      "epoch": 0.6395563770794824,
      "grad_norm": 2.390625,
      "learning_rate": 1.6284289279372343e-05,
      "loss": 1.7519,
      "step": 692
    },
    {
      "epoch": 0.6404805914972274,
      "grad_norm": 2.34375,
      "learning_rate": 1.627239025546583e-05,
      "loss": 1.7481,
      "step": 693
    },
    {
      "epoch": 0.6414048059149723,
      "grad_norm": 2.265625,
      "learning_rate": 1.6260476572399494e-05,
      "loss": 1.8054,
      "step": 694
    },
    {
      "epoch": 0.6423290203327172,
      "grad_norm": 2.234375,
      "learning_rate": 1.6248548258016722e-05,
      "loss": 1.7836,
      "step": 695
    },
    {
      "epoch": 0.6432532347504621,
      "grad_norm": 2.390625,
      "learning_rate": 1.6236605340195093e-05,
      "loss": 1.8227,
      "step": 696
    },
    {
      "epoch": 0.6441774491682071,
      "grad_norm": 2.328125,
      "learning_rate": 1.6224647846846314e-05,
      "loss": 1.8094,
      "step": 697
    },
    {
      "epoch": 0.6451016635859519,
      "grad_norm": 2.28125,
      "learning_rate": 1.621267580591617e-05,
      "loss": 1.8357,
      "step": 698
    },
    {
      "epoch": 0.6460258780036968,
      "grad_norm": 2.296875,
      "learning_rate": 1.6200689245384424e-05,
      "loss": 1.8122,
      "step": 699
    },
    {
      "epoch": 0.6469500924214417,
      "grad_norm": 2.3125,
      "learning_rate": 1.618868819326479e-05,
      "loss": 1.8756,
      "step": 700
    },
    {
      "epoch": 0.6478743068391867,
      "grad_norm": 2.484375,
      "learning_rate": 1.6176672677604844e-05,
      "loss": 1.7895,
      "step": 701
    },
    {
      "epoch": 0.6487985212569316,
      "grad_norm": 2.28125,
      "learning_rate": 1.616464272648596e-05,
      "loss": 1.7842,
      "step": 702
    },
    {
      "epoch": 0.6497227356746765,
      "grad_norm": 2.359375,
      "learning_rate": 1.615259836802326e-05,
      "loss": 1.6746,
      "step": 703
    },
    {
      "epoch": 0.6506469500924215,
      "grad_norm": 2.375,
      "learning_rate": 1.614053963036552e-05,
      "loss": 1.8615,
      "step": 704
    },
    {
      "epoch": 0.6515711645101664,
      "grad_norm": 2.3125,
      "learning_rate": 1.6128466541695144e-05,
      "loss": 1.7695,
      "step": 705
    },
    {
      "epoch": 0.6524953789279113,
      "grad_norm": 2.578125,
      "learning_rate": 1.611637913022806e-05,
      "loss": 1.7845,
      "step": 706
    },
    {
      "epoch": 0.6534195933456562,
      "grad_norm": 2.578125,
      "learning_rate": 1.610427742421367e-05,
      "loss": 1.7942,
      "step": 707
    },
    {
      "epoch": 0.6543438077634011,
      "grad_norm": 2.328125,
      "learning_rate": 1.609216145193479e-05,
      "loss": 1.8649,
      "step": 708
    },
    {
      "epoch": 0.655268022181146,
      "grad_norm": 2.25,
      "learning_rate": 1.608003124170758e-05,
      "loss": 1.7562,
      "step": 709
    },
    {
      "epoch": 0.6561922365988909,
      "grad_norm": 2.390625,
      "learning_rate": 1.606788682188146e-05,
      "loss": 1.7936,
      "step": 710
    },
    {
      "epoch": 0.6571164510166358,
      "grad_norm": 2.3125,
      "learning_rate": 1.6055728220839077e-05,
      "loss": 1.7874,
      "step": 711
    },
    {
      "epoch": 0.6580406654343808,
      "grad_norm": 2.265625,
      "learning_rate": 1.6043555466996206e-05,
      "loss": 1.7525,
      "step": 712
    },
    {
      "epoch": 0.6589648798521257,
      "grad_norm": 2.21875,
      "learning_rate": 1.6031368588801715e-05,
      "loss": 1.7345,
      "step": 713
    },
    {
      "epoch": 0.6598890942698706,
      "grad_norm": 2.328125,
      "learning_rate": 1.601916761473747e-05,
      "loss": 1.7493,
      "step": 714
    },
    {
      "epoch": 0.6608133086876156,
      "grad_norm": 2.4375,
      "learning_rate": 1.600695257331828e-05,
      "loss": 1.7529,
      "step": 715
    },
    {
      "epoch": 0.6617375231053605,
      "grad_norm": 2.328125,
      "learning_rate": 1.5994723493091832e-05,
      "loss": 1.779,
      "step": 716
    },
    {
      "epoch": 0.6626617375231053,
      "grad_norm": 2.3125,
      "learning_rate": 1.598248040263863e-05,
      "loss": 1.8572,
      "step": 717
    },
    {
      "epoch": 0.6635859519408502,
      "grad_norm": 2.296875,
      "learning_rate": 1.5970223330571912e-05,
      "loss": 1.7661,
      "step": 718
    },
    {
      "epoch": 0.6645101663585952,
      "grad_norm": 2.28125,
      "learning_rate": 1.59579523055376e-05,
      "loss": 1.8277,
      "step": 719
    },
    {
      "epoch": 0.6654343807763401,
      "grad_norm": 2.265625,
      "learning_rate": 1.5945667356214216e-05,
      "loss": 1.8224,
      "step": 720
    },
    {
      "epoch": 0.666358595194085,
      "grad_norm": 2.265625,
      "learning_rate": 1.5933368511312836e-05,
      "loss": 1.7635,
      "step": 721
    },
    {
      "epoch": 0.66728280961183,
      "grad_norm": 2.265625,
      "learning_rate": 1.5921055799577004e-05,
      "loss": 1.8137,
      "step": 722
    },
    {
      "epoch": 0.6682070240295749,
      "grad_norm": 2.1875,
      "learning_rate": 1.590872924978267e-05,
      "loss": 1.6704,
      "step": 723
    },
    {
      "epoch": 0.6691312384473198,
      "grad_norm": 2.21875,
      "learning_rate": 1.589638889073813e-05,
      "loss": 1.8304,
      "step": 724
    },
    {
      "epoch": 0.6700554528650647,
      "grad_norm": 2.53125,
      "learning_rate": 1.588403475128395e-05,
      "loss": 1.8284,
      "step": 725
    },
    {
      "epoch": 0.6709796672828097,
      "grad_norm": 2.359375,
      "learning_rate": 1.5871666860292905e-05,
      "loss": 1.8196,
      "step": 726
    },
    {
      "epoch": 0.6719038817005545,
      "grad_norm": 2.265625,
      "learning_rate": 1.5859285246669914e-05,
      "loss": 1.7755,
      "step": 727
    },
    {
      "epoch": 0.6728280961182994,
      "grad_norm": 2.421875,
      "learning_rate": 1.584688993935195e-05,
      "loss": 1.7778,
      "step": 728
    },
    {
      "epoch": 0.6737523105360443,
      "grad_norm": 2.421875,
      "learning_rate": 1.5834480967308005e-05,
      "loss": 1.727,
      "step": 729
    },
    {
      "epoch": 0.6746765249537893,
      "grad_norm": 2.3125,
      "learning_rate": 1.5822058359539002e-05,
      "loss": 1.7436,
      "step": 730
    },
    {
      "epoch": 0.6756007393715342,
      "grad_norm": 2.25,
      "learning_rate": 1.580962214507773e-05,
      "loss": 1.7105,
      "step": 731
    },
    {
      "epoch": 0.6765249537892791,
      "grad_norm": 2.296875,
      "learning_rate": 1.5797172352988775e-05,
      "loss": 1.8137,
      "step": 732
    },
    {
      "epoch": 0.677449168207024,
      "grad_norm": 2.359375,
      "learning_rate": 1.578470901236847e-05,
      "loss": 1.798,
      "step": 733
    },
    {
      "epoch": 0.678373382624769,
      "grad_norm": 2.546875,
      "learning_rate": 1.5772232152344797e-05,
      "loss": 1.8455,
      "step": 734
    },
    {
      "epoch": 0.6792975970425139,
      "grad_norm": 2.25,
      "learning_rate": 1.5759741802077337e-05,
      "loss": 1.739,
      "step": 735
    },
    {
      "epoch": 0.6802218114602587,
      "grad_norm": 2.28125,
      "learning_rate": 1.5747237990757205e-05,
      "loss": 1.803,
      "step": 736
    },
    {
      "epoch": 0.6811460258780037,
      "grad_norm": 2.328125,
      "learning_rate": 1.573472074760697e-05,
      "loss": 1.7791,
      "step": 737
    },
    {
      "epoch": 0.6820702402957486,
      "grad_norm": 2.265625,
      "learning_rate": 1.5722190101880594e-05,
      "loss": 1.7634,
      "step": 738
    },
    {
      "epoch": 0.6829944547134935,
      "grad_norm": 2.3125,
      "learning_rate": 1.570964608286336e-05,
      "loss": 1.7385,
      "step": 739
    },
    {
      "epoch": 0.6839186691312384,
      "grad_norm": 2.28125,
      "learning_rate": 1.5697088719871817e-05,
      "loss": 1.7355,
      "step": 740
    },
    {
      "epoch": 0.6848428835489834,
      "grad_norm": 2.3125,
      "learning_rate": 1.5684518042253677e-05,
      "loss": 1.7753,
      "step": 741
    },
    {
      "epoch": 0.6857670979667283,
      "grad_norm": 2.328125,
      "learning_rate": 1.5671934079387795e-05,
      "loss": 1.7161,
      "step": 742
    },
    {
      "epoch": 0.6866913123844732,
      "grad_norm": 2.765625,
      "learning_rate": 1.5659336860684062e-05,
      "loss": 1.8047,
      "step": 743
    },
    {
      "epoch": 0.6876155268022182,
      "grad_norm": 2.59375,
      "learning_rate": 1.5646726415583344e-05,
      "loss": 1.7531,
      "step": 744
    },
    {
      "epoch": 0.6885397412199631,
      "grad_norm": 2.203125,
      "learning_rate": 1.5634102773557432e-05,
      "loss": 1.6926,
      "step": 745
    },
    {
      "epoch": 0.6894639556377079,
      "grad_norm": 2.484375,
      "learning_rate": 1.5621465964108947e-05,
      "loss": 1.7061,
      "step": 746
    },
    {
      "epoch": 0.6903881700554528,
      "grad_norm": 2.6875,
      "learning_rate": 1.56088160167713e-05,
      "loss": 1.8197,
      "step": 747
    },
    {
      "epoch": 0.6913123844731978,
      "grad_norm": 2.515625,
      "learning_rate": 1.5596152961108577e-05,
      "loss": 1.7569,
      "step": 748
    },
    {
      "epoch": 0.6922365988909427,
      "grad_norm": 2.515625,
      "learning_rate": 1.5583476826715532e-05,
      "loss": 1.7616,
      "step": 749
    },
    {
      "epoch": 0.6931608133086876,
      "grad_norm": 2.328125,
      "learning_rate": 1.5570787643217465e-05,
      "loss": 1.7631,
      "step": 750
    },
    {
      "epoch": 0.6940850277264325,
      "grad_norm": 2.671875,
      "learning_rate": 1.5558085440270177e-05,
      "loss": 1.6993,
      "step": 751
    },
    {
      "epoch": 0.6950092421441775,
      "grad_norm": 2.8125,
      "learning_rate": 1.5545370247559907e-05,
      "loss": 1.8346,
      "step": 752
    },
    {
      "epoch": 0.6959334565619224,
      "grad_norm": 2.484375,
      "learning_rate": 1.553264209480324e-05,
      "loss": 1.7672,
      "step": 753
    },
    {
      "epoch": 0.6968576709796673,
      "grad_norm": 2.453125,
      "learning_rate": 1.5519901011747046e-05,
      "loss": 1.7772,
      "step": 754
    },
    {
      "epoch": 0.6977818853974121,
      "grad_norm": 2.3125,
      "learning_rate": 1.550714702816843e-05,
      "loss": 1.8091,
      "step": 755
    },
    {
      "epoch": 0.6987060998151571,
      "grad_norm": 2.265625,
      "learning_rate": 1.5494380173874638e-05,
      "loss": 1.7847,
      "step": 756
    },
    {
      "epoch": 0.699630314232902,
      "grad_norm": 2.578125,
      "learning_rate": 1.5481600478702995e-05,
      "loss": 1.7667,
      "step": 757
    },
    {
      "epoch": 0.7005545286506469,
      "grad_norm": 2.875,
      "learning_rate": 1.5468807972520837e-05,
      "loss": 1.9047,
      "step": 758
    },
    {
      "epoch": 0.7014787430683919,
      "grad_norm": 2.28125,
      "learning_rate": 1.5456002685225447e-05,
      "loss": 1.7364,
      "step": 759
    },
    {
      "epoch": 0.7024029574861368,
      "grad_norm": 2.515625,
      "learning_rate": 1.5443184646743968e-05,
      "loss": 1.8143,
      "step": 760
    },
    {
      "epoch": 0.7033271719038817,
      "grad_norm": 2.46875,
      "learning_rate": 1.5430353887033352e-05,
      "loss": 1.7119,
      "step": 761
    },
    {
      "epoch": 0.7042513863216266,
      "grad_norm": 2.65625,
      "learning_rate": 1.5417510436080276e-05,
      "loss": 1.7802,
      "step": 762
    },
    {
      "epoch": 0.7051756007393716,
      "grad_norm": 2.3125,
      "learning_rate": 1.5404654323901082e-05,
      "loss": 1.8215,
      "step": 763
    },
    {
      "epoch": 0.7060998151571165,
      "grad_norm": 2.265625,
      "learning_rate": 1.5391785580541697e-05,
      "loss": 1.8022,
      "step": 764
    },
    {
      "epoch": 0.7070240295748613,
      "grad_norm": 2.40625,
      "learning_rate": 1.537890423607758e-05,
      "loss": 1.8711,
      "step": 765
    },
    {
      "epoch": 0.7079482439926063,
      "grad_norm": 2.28125,
      "learning_rate": 1.536601032061362e-05,
      "loss": 1.7202,
      "step": 766
    },
    {
      "epoch": 0.7088724584103512,
      "grad_norm": 2.40625,
      "learning_rate": 1.535310386428411e-05,
      "loss": 1.7944,
      "step": 767
    },
    {
      "epoch": 0.7097966728280961,
      "grad_norm": 2.4375,
      "learning_rate": 1.534018489725263e-05,
      "loss": 1.8318,
      "step": 768
    },
    {
      "epoch": 0.710720887245841,
      "grad_norm": 2.328125,
      "learning_rate": 1.532725344971202e-05,
      "loss": 1.7648,
      "step": 769
    },
    {
      "epoch": 0.711645101663586,
      "grad_norm": 2.28125,
      "learning_rate": 1.5314309551884264e-05,
      "loss": 1.7047,
      "step": 770
    },
    {
      "epoch": 0.7125693160813309,
      "grad_norm": 2.359375,
      "learning_rate": 1.5301353234020462e-05,
      "loss": 1.7886,
      "step": 771
    },
    {
      "epoch": 0.7134935304990758,
      "grad_norm": 2.359375,
      "learning_rate": 1.5288384526400735e-05,
      "loss": 1.7931,
      "step": 772
    },
    {
      "epoch": 0.7144177449168208,
      "grad_norm": 2.296875,
      "learning_rate": 1.5275403459334154e-05,
      "loss": 1.8331,
      "step": 773
    },
    {
      "epoch": 0.7153419593345656,
      "grad_norm": 2.296875,
      "learning_rate": 1.526241006315869e-05,
      "loss": 1.855,
      "step": 774
    },
    {
      "epoch": 0.7162661737523105,
      "grad_norm": 2.375,
      "learning_rate": 1.5249404368241117e-05,
      "loss": 1.8126,
      "step": 775
    },
    {
      "epoch": 0.7171903881700554,
      "grad_norm": 2.234375,
      "learning_rate": 1.5236386404976949e-05,
      "loss": 1.754,
      "step": 776
    },
    {
      "epoch": 0.7181146025878004,
      "grad_norm": 2.359375,
      "learning_rate": 1.5223356203790381e-05,
      "loss": 1.8277,
      "step": 777
    },
    {
      "epoch": 0.7190388170055453,
      "grad_norm": 2.453125,
      "learning_rate": 1.5210313795134203e-05,
      "loss": 1.7469,
      "step": 778
    },
    {
      "epoch": 0.7199630314232902,
      "grad_norm": 2.375,
      "learning_rate": 1.5197259209489748e-05,
      "loss": 1.8264,
      "step": 779
    },
    {
      "epoch": 0.7208872458410351,
      "grad_norm": 2.265625,
      "learning_rate": 1.5184192477366786e-05,
      "loss": 1.7926,
      "step": 780
    },
    {
      "epoch": 0.7218114602587801,
      "grad_norm": 2.34375,
      "learning_rate": 1.517111362930349e-05,
      "loss": 1.7593,
      "step": 781
    },
    {
      "epoch": 0.722735674676525,
      "grad_norm": 2.515625,
      "learning_rate": 1.5158022695866348e-05,
      "loss": 1.793,
      "step": 782
    },
    {
      "epoch": 0.7236598890942699,
      "grad_norm": 2.328125,
      "learning_rate": 1.5144919707650088e-05,
      "loss": 1.7454,
      "step": 783
    },
    {
      "epoch": 0.7245841035120147,
      "grad_norm": 2.3125,
      "learning_rate": 1.5131804695277612e-05,
      "loss": 1.7493,
      "step": 784
    },
    {
      "epoch": 0.7255083179297597,
      "grad_norm": 2.28125,
      "learning_rate": 1.5118677689399923e-05,
      "loss": 1.7089,
      "step": 785
    },
    {
      "epoch": 0.7264325323475046,
      "grad_norm": 2.3125,
      "learning_rate": 1.5105538720696062e-05,
      "loss": 1.7174,
      "step": 786
    },
    {
      "epoch": 0.7273567467652495,
      "grad_norm": 2.359375,
      "learning_rate": 1.5092387819873016e-05,
      "loss": 1.7641,
      "step": 787
    },
    {
      "epoch": 0.7282809611829945,
      "grad_norm": 2.4375,
      "learning_rate": 1.5079225017665668e-05,
      "loss": 1.7438,
      "step": 788
    },
    {
      "epoch": 0.7292051756007394,
      "grad_norm": 2.375,
      "learning_rate": 1.5066050344836708e-05,
      "loss": 1.7347,
      "step": 789
    },
    {
      "epoch": 0.7301293900184843,
      "grad_norm": 2.40625,
      "learning_rate": 1.5052863832176572e-05,
      "loss": 1.8348,
      "step": 790
    },
    {
      "epoch": 0.7310536044362292,
      "grad_norm": 2.40625,
      "learning_rate": 1.5039665510503375e-05,
      "loss": 1.7957,
      "step": 791
    },
    {
      "epoch": 0.7319778188539742,
      "grad_norm": 2.3125,
      "learning_rate": 1.5026455410662814e-05,
      "loss": 1.8147,
      "step": 792
    },
    {
      "epoch": 0.732902033271719,
      "grad_norm": 2.3125,
      "learning_rate": 1.5013233563528129e-05,
      "loss": 1.8141,
      "step": 793
    },
    {
      "epoch": 0.7338262476894639,
      "grad_norm": 2.234375,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.8052,
      "step": 794
    },
    {
      "epoch": 0.7347504621072088,
      "grad_norm": 2.1875,
      "learning_rate": 1.4986754751006505e-05,
      "loss": 1.8449,
      "step": 795
    },
    {
      "epoch": 0.7356746765249538,
      "grad_norm": 2.390625,
      "learning_rate": 1.4973497847503018e-05,
      "loss": 1.7465,
      "step": 796
    },
    {
      "epoch": 0.7365988909426987,
      "grad_norm": 2.4375,
      "learning_rate": 1.4960229320472156e-05,
      "loss": 1.8365,
      "step": 797
    },
    {
      "epoch": 0.7375231053604436,
      "grad_norm": 2.328125,
      "learning_rate": 1.4946949200923708e-05,
      "loss": 1.759,
      "step": 798
    },
    {
      "epoch": 0.7384473197781886,
      "grad_norm": 2.3125,
      "learning_rate": 1.4933657519894542e-05,
      "loss": 1.7429,
      "step": 799
    },
    {
      "epoch": 0.7393715341959335,
      "grad_norm": 2.375,
      "learning_rate": 1.4920354308448557e-05,
      "loss": 1.7776,
      "step": 800
    },
    {
      "epoch": 0.7402957486136784,
      "grad_norm": 2.34375,
      "learning_rate": 1.4907039597676598e-05,
      "loss": 1.7743,
      "step": 801
    },
    {
      "epoch": 0.7412199630314233,
      "grad_norm": 2.40625,
      "learning_rate": 1.4893713418696378e-05,
      "loss": 1.7352,
      "step": 802
    },
    {
      "epoch": 0.7421441774491682,
      "grad_norm": 2.390625,
      "learning_rate": 1.4880375802652423e-05,
      "loss": 1.774,
      "step": 803
    },
    {
      "epoch": 0.7430683918669131,
      "grad_norm": 2.375,
      "learning_rate": 1.4867026780715981e-05,
      "loss": 1.7264,
      "step": 804
    },
    {
      "epoch": 0.743992606284658,
      "grad_norm": 2.328125,
      "learning_rate": 1.4853666384084959e-05,
      "loss": 1.8298,
      "step": 805
    },
    {
      "epoch": 0.744916820702403,
      "grad_norm": 2.328125,
      "learning_rate": 1.4840294643983845e-05,
      "loss": 1.7876,
      "step": 806
    },
    {
      "epoch": 0.7458410351201479,
      "grad_norm": 2.390625,
      "learning_rate": 1.4826911591663644e-05,
      "loss": 1.7342,
      "step": 807
    },
    {
      "epoch": 0.7467652495378928,
      "grad_norm": 2.375,
      "learning_rate": 1.4813517258401793e-05,
      "loss": 1.7839,
      "step": 808
    },
    {
      "epoch": 0.7476894639556377,
      "grad_norm": 2.328125,
      "learning_rate": 1.4800111675502095e-05,
      "loss": 1.7903,
      "step": 809
    },
    {
      "epoch": 0.7486136783733827,
      "grad_norm": 2.34375,
      "learning_rate": 1.4786694874294647e-05,
      "loss": 1.7304,
      "step": 810
    },
    {
      "epoch": 0.7495378927911276,
      "grad_norm": 2.328125,
      "learning_rate": 1.4773266886135762e-05,
      "loss": 1.7142,
      "step": 811
    },
    {
      "epoch": 0.7504621072088724,
      "grad_norm": 2.28125,
      "learning_rate": 1.4759827742407893e-05,
      "loss": 1.6443,
      "step": 812
    },
    {
      "epoch": 0.7513863216266173,
      "grad_norm": 2.265625,
      "learning_rate": 1.4746377474519575e-05,
      "loss": 1.7573,
      "step": 813
    },
    {
      "epoch": 0.7523105360443623,
      "grad_norm": 2.390625,
      "learning_rate": 1.4732916113905336e-05,
      "loss": 1.8038,
      "step": 814
    },
    {
      "epoch": 0.7532347504621072,
      "grad_norm": 2.21875,
      "learning_rate": 1.4719443692025628e-05,
      "loss": 1.7174,
      "step": 815
    },
    {
      "epoch": 0.7541589648798521,
      "grad_norm": 2.390625,
      "learning_rate": 1.4705960240366754e-05,
      "loss": 1.7866,
      "step": 816
    },
    {
      "epoch": 0.755083179297597,
      "grad_norm": 2.375,
      "learning_rate": 1.4692465790440794e-05,
      "loss": 1.835,
      "step": 817
    },
    {
      "epoch": 0.756007393715342,
      "grad_norm": 2.375,
      "learning_rate": 1.4678960373785533e-05,
      "loss": 1.8002,
      "step": 818
    },
    {
      "epoch": 0.7569316081330869,
      "grad_norm": 2.328125,
      "learning_rate": 1.4665444021964392e-05,
      "loss": 1.7899,
      "step": 819
    },
    {
      "epoch": 0.7578558225508318,
      "grad_norm": 2.25,
      "learning_rate": 1.465191676656634e-05,
      "loss": 1.7414,
      "step": 820
    },
    {
      "epoch": 0.7587800369685767,
      "grad_norm": 2.3125,
      "learning_rate": 1.4638378639205833e-05,
      "loss": 1.7971,
      "step": 821
    },
    {
      "epoch": 0.7597042513863216,
      "grad_norm": 2.4375,
      "learning_rate": 1.4624829671522732e-05,
      "loss": 1.6684,
      "step": 822
    },
    {
      "epoch": 0.7606284658040665,
      "grad_norm": 2.265625,
      "learning_rate": 1.4611269895182239e-05,
      "loss": 1.8049,
      "step": 823
    },
    {
      "epoch": 0.7615526802218114,
      "grad_norm": 2.203125,
      "learning_rate": 1.4597699341874806e-05,
      "loss": 1.7747,
      "step": 824
    },
    {
      "epoch": 0.7624768946395564,
      "grad_norm": 2.3125,
      "learning_rate": 1.4584118043316088e-05,
      "loss": 1.8228,
      "step": 825
    },
    {
      "epoch": 0.7634011090573013,
      "grad_norm": 2.25,
      "learning_rate": 1.4570526031246844e-05,
      "loss": 1.7713,
      "step": 826
    },
    {
      "epoch": 0.7643253234750462,
      "grad_norm": 2.34375,
      "learning_rate": 1.4556923337432865e-05,
      "loss": 1.7706,
      "step": 827
    },
    {
      "epoch": 0.7652495378927912,
      "grad_norm": 2.265625,
      "learning_rate": 1.4543309993664916e-05,
      "loss": 1.7494,
      "step": 828
    },
    {
      "epoch": 0.7661737523105361,
      "grad_norm": 2.265625,
      "learning_rate": 1.4529686031758642e-05,
      "loss": 1.8213,
      "step": 829
    },
    {
      "epoch": 0.767097966728281,
      "grad_norm": 2.234375,
      "learning_rate": 1.4516051483554524e-05,
      "loss": 1.6974,
      "step": 830
    },
    {
      "epoch": 0.7680221811460258,
      "grad_norm": 2.3125,
      "learning_rate": 1.4502406380917757e-05,
      "loss": 1.7619,
      "step": 831
    },
    {
      "epoch": 0.7689463955637708,
      "grad_norm": 2.234375,
      "learning_rate": 1.4488750755738224e-05,
      "loss": 1.7384,
      "step": 832
    },
    {
      "epoch": 0.7698706099815157,
      "grad_norm": 2.25,
      "learning_rate": 1.4475084639930386e-05,
      "loss": 1.7818,
      "step": 833
    },
    {
      "epoch": 0.7707948243992606,
      "grad_norm": 2.359375,
      "learning_rate": 1.446140806543323e-05,
      "loss": 1.837,
      "step": 834
    },
    {
      "epoch": 0.7717190388170055,
      "grad_norm": 2.25,
      "learning_rate": 1.4447721064210184e-05,
      "loss": 1.7348,
      "step": 835
    },
    {
      "epoch": 0.7726432532347505,
      "grad_norm": 2.21875,
      "learning_rate": 1.4434023668249046e-05,
      "loss": 1.7972,
      "step": 836
    },
    {
      "epoch": 0.7735674676524954,
      "grad_norm": 2.25,
      "learning_rate": 1.4420315909561901e-05,
      "loss": 1.7634,
      "step": 837
    },
    {
      "epoch": 0.7744916820702403,
      "grad_norm": 2.5625,
      "learning_rate": 1.4406597820185063e-05,
      "loss": 1.7113,
      "step": 838
    },
    {
      "epoch": 0.7754158964879853,
      "grad_norm": 2.375,
      "learning_rate": 1.4392869432178974e-05,
      "loss": 1.6995,
      "step": 839
    },
    {
      "epoch": 0.7763401109057301,
      "grad_norm": 2.28125,
      "learning_rate": 1.4379130777628163e-05,
      "loss": 1.8317,
      "step": 840
    },
    {
      "epoch": 0.777264325323475,
      "grad_norm": 2.28125,
      "learning_rate": 1.436538188864114e-05,
      "loss": 1.7965,
      "step": 841
    },
    {
      "epoch": 0.7781885397412199,
      "grad_norm": 2.296875,
      "learning_rate": 1.4351622797350338e-05,
      "loss": 1.836,
      "step": 842
    },
    {
      "epoch": 0.7791127541589649,
      "grad_norm": 2.3125,
      "learning_rate": 1.4337853535912039e-05,
      "loss": 1.8173,
      "step": 843
    },
    {
      "epoch": 0.7800369685767098,
      "grad_norm": 2.3125,
      "learning_rate": 1.4324074136506283e-05,
      "loss": 1.8036,
      "step": 844
    },
    {
      "epoch": 0.7809611829944547,
      "grad_norm": 2.34375,
      "learning_rate": 1.4310284631336812e-05,
      "loss": 1.7517,
      "step": 845
    },
    {
      "epoch": 0.7818853974121996,
      "grad_norm": 2.28125,
      "learning_rate": 1.429648505263098e-05,
      "loss": 1.7596,
      "step": 846
    },
    {
      "epoch": 0.7828096118299446,
      "grad_norm": 2.25,
      "learning_rate": 1.428267543263969e-05,
      "loss": 1.7631,
      "step": 847
    },
    {
      "epoch": 0.7837338262476895,
      "grad_norm": 2.265625,
      "learning_rate": 1.426885580363731e-05,
      "loss": 1.7933,
      "step": 848
    },
    {
      "epoch": 0.7846580406654344,
      "grad_norm": 2.265625,
      "learning_rate": 1.4255026197921596e-05,
      "loss": 1.7236,
      "step": 849
    },
    {
      "epoch": 0.7855822550831792,
      "grad_norm": 2.28125,
      "learning_rate": 1.4241186647813626e-05,
      "loss": 1.8102,
      "step": 850
    },
    {
      "epoch": 0.7865064695009242,
      "grad_norm": 2.25,
      "learning_rate": 1.4227337185657712e-05,
      "loss": 1.8166,
      "step": 851
    },
    {
      "epoch": 0.7874306839186691,
      "grad_norm": 2.40625,
      "learning_rate": 1.4213477843821345e-05,
      "loss": 1.7751,
      "step": 852
    },
    {
      "epoch": 0.788354898336414,
      "grad_norm": 2.421875,
      "learning_rate": 1.4199608654695088e-05,
      "loss": 1.8383,
      "step": 853
    },
    {
      "epoch": 0.789279112754159,
      "grad_norm": 2.265625,
      "learning_rate": 1.4185729650692533e-05,
      "loss": 1.7932,
      "step": 854
    },
    {
      "epoch": 0.7902033271719039,
      "grad_norm": 2.25,
      "learning_rate": 1.41718408642502e-05,
      "loss": 1.7964,
      "step": 855
    },
    {
      "epoch": 0.7911275415896488,
      "grad_norm": 2.3125,
      "learning_rate": 1.4157942327827477e-05,
      "loss": 1.7234,
      "step": 856
    },
    {
      "epoch": 0.7920517560073937,
      "grad_norm": 2.328125,
      "learning_rate": 1.4144034073906539e-05,
      "loss": 1.8805,
      "step": 857
    },
    {
      "epoch": 0.7929759704251387,
      "grad_norm": 2.21875,
      "learning_rate": 1.4130116134992267e-05,
      "loss": 1.7262,
      "step": 858
    },
    {
      "epoch": 0.7939001848428835,
      "grad_norm": 2.1875,
      "learning_rate": 1.4116188543612182e-05,
      "loss": 1.7126,
      "step": 859
    },
    {
      "epoch": 0.7948243992606284,
      "grad_norm": 2.25,
      "learning_rate": 1.410225133231636e-05,
      "loss": 1.7707,
      "step": 860
    },
    {
      "epoch": 0.7957486136783734,
      "grad_norm": 2.421875,
      "learning_rate": 1.408830453367736e-05,
      "loss": 1.7615,
      "step": 861
    },
    {
      "epoch": 0.7966728280961183,
      "grad_norm": 2.359375,
      "learning_rate": 1.4074348180290151e-05,
      "loss": 1.7847,
      "step": 862
    },
    {
      "epoch": 0.7975970425138632,
      "grad_norm": 2.265625,
      "learning_rate": 1.4060382304772028e-05,
      "loss": 1.7322,
      "step": 863
    },
    {
      "epoch": 0.7985212569316081,
      "grad_norm": 2.28125,
      "learning_rate": 1.4046406939762545e-05,
      "loss": 1.7945,
      "step": 864
    },
    {
      "epoch": 0.7994454713493531,
      "grad_norm": 2.296875,
      "learning_rate": 1.4032422117923427e-05,
      "loss": 1.847,
      "step": 865
    },
    {
      "epoch": 0.800369685767098,
      "grad_norm": 2.328125,
      "learning_rate": 1.4018427871938508e-05,
      "loss": 1.661,
      "step": 866
    },
    {
      "epoch": 0.8012939001848429,
      "grad_norm": 2.3125,
      "learning_rate": 1.4004424234513638e-05,
      "loss": 1.8186,
      "step": 867
    },
    {
      "epoch": 0.8022181146025879,
      "grad_norm": 2.359375,
      "learning_rate": 1.3990411238376622e-05,
      "loss": 1.7442,
      "step": 868
    },
    {
      "epoch": 0.8031423290203327,
      "grad_norm": 2.28125,
      "learning_rate": 1.3976388916277141e-05,
      "loss": 1.7814,
      "step": 869
    },
    {
      "epoch": 0.8040665434380776,
      "grad_norm": 2.265625,
      "learning_rate": 1.3962357300986667e-05,
      "loss": 1.6996,
      "step": 870
    },
    {
      "epoch": 0.8049907578558225,
      "grad_norm": 2.296875,
      "learning_rate": 1.3948316425298386e-05,
      "loss": 1.7575,
      "step": 871
    },
    {
      "epoch": 0.8059149722735675,
      "grad_norm": 2.15625,
      "learning_rate": 1.3934266322027132e-05,
      "loss": 1.6443,
      "step": 872
    },
    {
      "epoch": 0.8068391866913124,
      "grad_norm": 2.453125,
      "learning_rate": 1.3920207024009305e-05,
      "loss": 1.8647,
      "step": 873
    },
    {
      "epoch": 0.8077634011090573,
      "grad_norm": 2.25,
      "learning_rate": 1.3906138564102794e-05,
      "loss": 1.8063,
      "step": 874
    },
    {
      "epoch": 0.8086876155268022,
      "grad_norm": 2.59375,
      "learning_rate": 1.3892060975186899e-05,
      "loss": 1.7265,
      "step": 875
    },
    {
      "epoch": 0.8096118299445472,
      "grad_norm": 2.25,
      "learning_rate": 1.3877974290162256e-05,
      "loss": 1.7838,
      "step": 876
    },
    {
      "epoch": 0.8105360443622921,
      "grad_norm": 2.3125,
      "learning_rate": 1.386387854195076e-05,
      "loss": 1.7849,
      "step": 877
    },
    {
      "epoch": 0.8114602587800369,
      "grad_norm": 2.359375,
      "learning_rate": 1.384977376349548e-05,
      "loss": 1.7519,
      "step": 878
    },
    {
      "epoch": 0.8123844731977818,
      "grad_norm": 2.3125,
      "learning_rate": 1.3835659987760605e-05,
      "loss": 1.8347,
      "step": 879
    },
    {
      "epoch": 0.8133086876155268,
      "grad_norm": 2.515625,
      "learning_rate": 1.3821537247731336e-05,
      "loss": 1.8398,
      "step": 880
    },
    {
      "epoch": 0.8142329020332717,
      "grad_norm": 2.265625,
      "learning_rate": 1.3807405576413833e-05,
      "loss": 1.8579,
      "step": 881
    },
    {
      "epoch": 0.8151571164510166,
      "grad_norm": 2.28125,
      "learning_rate": 1.3793265006835129e-05,
      "loss": 1.7627,
      "step": 882
    },
    {
      "epoch": 0.8160813308687616,
      "grad_norm": 2.3125,
      "learning_rate": 1.377911557204305e-05,
      "loss": 1.7712,
      "step": 883
    },
    {
      "epoch": 0.8170055452865065,
      "grad_norm": 2.21875,
      "learning_rate": 1.3764957305106142e-05,
      "loss": 1.745,
      "step": 884
    },
    {
      "epoch": 0.8179297597042514,
      "grad_norm": 2.34375,
      "learning_rate": 1.3750790239113588e-05,
      "loss": 1.6701,
      "step": 885
    },
    {
      "epoch": 0.8188539741219963,
      "grad_norm": 2.3125,
      "learning_rate": 1.3736614407175146e-05,
      "loss": 1.8181,
      "step": 886
    },
    {
      "epoch": 0.8197781885397413,
      "grad_norm": 2.359375,
      "learning_rate": 1.3722429842421055e-05,
      "loss": 1.8625,
      "step": 887
    },
    {
      "epoch": 0.8207024029574861,
      "grad_norm": 2.28125,
      "learning_rate": 1.3708236578001958e-05,
      "loss": 1.858,
      "step": 888
    },
    {
      "epoch": 0.821626617375231,
      "grad_norm": 2.390625,
      "learning_rate": 1.369403464708884e-05,
      "loss": 1.7923,
      "step": 889
    },
    {
      "epoch": 0.822550831792976,
      "grad_norm": 2.328125,
      "learning_rate": 1.3679824082872932e-05,
      "loss": 1.7148,
      "step": 890
    },
    {
      "epoch": 0.8234750462107209,
      "grad_norm": 2.375,
      "learning_rate": 1.3665604918565646e-05,
      "loss": 1.7588,
      "step": 891
    },
    {
      "epoch": 0.8243992606284658,
      "grad_norm": 2.25,
      "learning_rate": 1.3651377187398494e-05,
      "loss": 1.8226,
      "step": 892
    },
    {
      "epoch": 0.8253234750462107,
      "grad_norm": 2.40625,
      "learning_rate": 1.3637140922623007e-05,
      "loss": 1.8,
      "step": 893
    },
    {
      "epoch": 0.8262476894639557,
      "grad_norm": 2.28125,
      "learning_rate": 1.362289615751066e-05,
      "loss": 1.7712,
      "step": 894
    },
    {
      "epoch": 0.8271719038817006,
      "grad_norm": 2.296875,
      "learning_rate": 1.3608642925352794e-05,
      "loss": 1.7373,
      "step": 895
    },
    {
      "epoch": 0.8280961182994455,
      "grad_norm": 2.265625,
      "learning_rate": 1.359438125946054e-05,
      "loss": 1.8171,
      "step": 896
    },
    {
      "epoch": 0.8290203327171903,
      "grad_norm": 2.3125,
      "learning_rate": 1.358011119316474e-05,
      "loss": 1.7703,
      "step": 897
    },
    {
      "epoch": 0.8299445471349353,
      "grad_norm": 2.328125,
      "learning_rate": 1.3565832759815866e-05,
      "loss": 1.8006,
      "step": 898
    },
    {
      "epoch": 0.8308687615526802,
      "grad_norm": 2.140625,
      "learning_rate": 1.3551545992783948e-05,
      "loss": 1.6905,
      "step": 899
    },
    {
      "epoch": 0.8317929759704251,
      "grad_norm": 2.25,
      "learning_rate": 1.3537250925458488e-05,
      "loss": 1.805,
      "step": 900
    },
    {
      "epoch": 0.83271719038817,
      "grad_norm": 2.25,
      "learning_rate": 1.3522947591248388e-05,
      "loss": 1.7636,
      "step": 901
    },
    {
      "epoch": 0.833641404805915,
      "grad_norm": 2.3125,
      "learning_rate": 1.3508636023581871e-05,
      "loss": 1.7451,
      "step": 902
    },
    {
      "epoch": 0.8345656192236599,
      "grad_norm": 2.171875,
      "learning_rate": 1.3494316255906405e-05,
      "loss": 1.7164,
      "step": 903
    },
    {
      "epoch": 0.8354898336414048,
      "grad_norm": 2.1875,
      "learning_rate": 1.3479988321688619e-05,
      "loss": 1.7113,
      "step": 904
    },
    {
      "epoch": 0.8364140480591498,
      "grad_norm": 2.25,
      "learning_rate": 1.3465652254414229e-05,
      "loss": 1.7954,
      "step": 905
    },
    {
      "epoch": 0.8373382624768947,
      "grad_norm": 2.34375,
      "learning_rate": 1.345130808758796e-05,
      "loss": 1.8064,
      "step": 906
    },
    {
      "epoch": 0.8382624768946395,
      "grad_norm": 2.25,
      "learning_rate": 1.343695585473346e-05,
      "loss": 1.6947,
      "step": 907
    },
    {
      "epoch": 0.8391866913123844,
      "grad_norm": 2.265625,
      "learning_rate": 1.3422595589393237e-05,
      "loss": 1.7491,
      "step": 908
    },
    {
      "epoch": 0.8401109057301294,
      "grad_norm": 2.25,
      "learning_rate": 1.3408227325128569e-05,
      "loss": 1.7537,
      "step": 909
    },
    {
      "epoch": 0.8410351201478743,
      "grad_norm": 2.296875,
      "learning_rate": 1.3393851095519424e-05,
      "loss": 1.8628,
      "step": 910
    },
    {
      "epoch": 0.8419593345656192,
      "grad_norm": 2.328125,
      "learning_rate": 1.3379466934164389e-05,
      "loss": 1.756,
      "step": 911
    },
    {
      "epoch": 0.8428835489833642,
      "grad_norm": 2.359375,
      "learning_rate": 1.3365074874680586e-05,
      "loss": 1.7818,
      "step": 912
    },
    {
      "epoch": 0.8438077634011091,
      "grad_norm": 2.34375,
      "learning_rate": 1.3350674950703601e-05,
      "loss": 1.772,
      "step": 913
    },
    {
      "epoch": 0.844731977818854,
      "grad_norm": 2.234375,
      "learning_rate": 1.3336267195887399e-05,
      "loss": 1.76,
      "step": 914
    },
    {
      "epoch": 0.8456561922365989,
      "grad_norm": 2.34375,
      "learning_rate": 1.3321851643904236e-05,
      "loss": 1.7541,
      "step": 915
    },
    {
      "epoch": 0.8465804066543438,
      "grad_norm": 2.296875,
      "learning_rate": 1.3307428328444605e-05,
      "loss": 1.8326,
      "step": 916
    },
    {
      "epoch": 0.8475046210720887,
      "grad_norm": 2.296875,
      "learning_rate": 1.3292997283217133e-05,
      "loss": 1.8199,
      "step": 917
    },
    {
      "epoch": 0.8484288354898336,
      "grad_norm": 2.21875,
      "learning_rate": 1.327855854194852e-05,
      "loss": 1.7938,
      "step": 918
    },
    {
      "epoch": 0.8493530499075785,
      "grad_norm": 2.25,
      "learning_rate": 1.3264112138383445e-05,
      "loss": 1.805,
      "step": 919
    },
    {
      "epoch": 0.8502772643253235,
      "grad_norm": 2.25,
      "learning_rate": 1.32496581062845e-05,
      "loss": 1.7256,
      "step": 920
    },
    {
      "epoch": 0.8512014787430684,
      "grad_norm": 2.34375,
      "learning_rate": 1.3235196479432099e-05,
      "loss": 1.8061,
      "step": 921
    },
    {
      "epoch": 0.8521256931608133,
      "grad_norm": 2.265625,
      "learning_rate": 1.3220727291624415e-05,
      "loss": 1.7492,
      "step": 922
    },
    {
      "epoch": 0.8530499075785583,
      "grad_norm": 2.21875,
      "learning_rate": 1.3206250576677284e-05,
      "loss": 1.701,
      "step": 923
    },
    {
      "epoch": 0.8539741219963032,
      "grad_norm": 2.28125,
      "learning_rate": 1.3191766368424134e-05,
      "loss": 1.7678,
      "step": 924
    },
    {
      "epoch": 0.8548983364140481,
      "grad_norm": 2.265625,
      "learning_rate": 1.3177274700715914e-05,
      "loss": 1.7748,
      "step": 925
    },
    {
      "epoch": 0.8558225508317929,
      "grad_norm": 2.203125,
      "learning_rate": 1.3162775607420995e-05,
      "loss": 1.7107,
      "step": 926
    },
    {
      "epoch": 0.8567467652495379,
      "grad_norm": 2.28125,
      "learning_rate": 1.314826912242511e-05,
      "loss": 1.7665,
      "step": 927
    },
    {
      "epoch": 0.8576709796672828,
      "grad_norm": 2.421875,
      "learning_rate": 1.3133755279631264e-05,
      "loss": 1.7532,
      "step": 928
    },
    {
      "epoch": 0.8585951940850277,
      "grad_norm": 2.265625,
      "learning_rate": 1.3119234112959656e-05,
      "loss": 1.7546,
      "step": 929
    },
    {
      "epoch": 0.8595194085027726,
      "grad_norm": 2.296875,
      "learning_rate": 1.310470565634761e-05,
      "loss": 1.8021,
      "step": 930
    },
    {
      "epoch": 0.8604436229205176,
      "grad_norm": 2.25,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 1.6847,
      "step": 931
    },
    {
      "epoch": 0.8613678373382625,
      "grad_norm": 2.203125,
      "learning_rate": 1.307562700913657e-05,
      "loss": 1.767,
      "step": 932
    },
    {
      "epoch": 0.8622920517560074,
      "grad_norm": 2.296875,
      "learning_rate": 1.3061076886497082e-05,
      "loss": 1.6855,
      "step": 933
    },
    {
      "epoch": 0.8632162661737524,
      "grad_norm": 2.296875,
      "learning_rate": 1.3046519609836002e-05,
      "loss": 1.7172,
      "step": 934
    },
    {
      "epoch": 0.8641404805914972,
      "grad_norm": 2.25,
      "learning_rate": 1.303195521317504e-05,
      "loss": 1.8457,
      "step": 935
    },
    {
      "epoch": 0.8650646950092421,
      "grad_norm": 2.296875,
      "learning_rate": 1.3017383730552553e-05,
      "loss": 1.669,
      "step": 936
    },
    {
      "epoch": 0.865988909426987,
      "grad_norm": 2.265625,
      "learning_rate": 1.3002805196023447e-05,
      "loss": 1.7659,
      "step": 937
    },
    {
      "epoch": 0.866913123844732,
      "grad_norm": 2.25,
      "learning_rate": 1.2988219643659114e-05,
      "loss": 1.7613,
      "step": 938
    },
    {
      "epoch": 0.8678373382624769,
      "grad_norm": 2.25,
      "learning_rate": 1.2973627107547346e-05,
      "loss": 1.8635,
      "step": 939
    },
    {
      "epoch": 0.8687615526802218,
      "grad_norm": 2.28125,
      "learning_rate": 1.2959027621792266e-05,
      "loss": 1.7871,
      "step": 940
    },
    {
      "epoch": 0.8696857670979667,
      "grad_norm": 2.40625,
      "learning_rate": 1.294442122051422e-05,
      "loss": 1.8043,
      "step": 941
    },
    {
      "epoch": 0.8706099815157117,
      "grad_norm": 2.28125,
      "learning_rate": 1.2929807937849735e-05,
      "loss": 1.7799,
      "step": 942
    },
    {
      "epoch": 0.8715341959334566,
      "grad_norm": 2.25,
      "learning_rate": 1.2915187807951405e-05,
      "loss": 1.791,
      "step": 943
    },
    {
      "epoch": 0.8724584103512015,
      "grad_norm": 2.34375,
      "learning_rate": 1.290056086498785e-05,
      "loss": 1.6993,
      "step": 944
    },
    {
      "epoch": 0.8733826247689463,
      "grad_norm": 2.25,
      "learning_rate": 1.288592714314358e-05,
      "loss": 1.7641,
      "step": 945
    },
    {
      "epoch": 0.8743068391866913,
      "grad_norm": 2.265625,
      "learning_rate": 1.287128667661897e-05,
      "loss": 1.7605,
      "step": 946
    },
    {
      "epoch": 0.8752310536044362,
      "grad_norm": 2.328125,
      "learning_rate": 1.285663949963016e-05,
      "loss": 1.8159,
      "step": 947
    },
    {
      "epoch": 0.8761552680221811,
      "grad_norm": 2.296875,
      "learning_rate": 1.2841985646408954e-05,
      "loss": 1.7955,
      "step": 948
    },
    {
      "epoch": 0.8770794824399261,
      "grad_norm": 2.234375,
      "learning_rate": 1.2827325151202783e-05,
      "loss": 1.717,
      "step": 949
    },
    {
      "epoch": 0.878003696857671,
      "grad_norm": 2.21875,
      "learning_rate": 1.281265804827458e-05,
      "loss": 1.7707,
      "step": 950
    },
    {
      "epoch": 0.8789279112754159,
      "grad_norm": 2.265625,
      "learning_rate": 1.2797984371902732e-05,
      "loss": 1.7301,
      "step": 951
    },
    {
      "epoch": 0.8798521256931608,
      "grad_norm": 2.328125,
      "learning_rate": 1.2783304156380989e-05,
      "loss": 1.7663,
      "step": 952
    },
    {
      "epoch": 0.8807763401109058,
      "grad_norm": 2.359375,
      "learning_rate": 1.276861743601838e-05,
      "loss": 1.8238,
      "step": 953
    },
    {
      "epoch": 0.8817005545286506,
      "grad_norm": 2.390625,
      "learning_rate": 1.2753924245139135e-05,
      "loss": 1.809,
      "step": 954
    },
    {
      "epoch": 0.8826247689463955,
      "grad_norm": 2.265625,
      "learning_rate": 1.2739224618082612e-05,
      "loss": 1.7541,
      "step": 955
    },
    {
      "epoch": 0.8835489833641405,
      "grad_norm": 2.21875,
      "learning_rate": 1.2724518589203204e-05,
      "loss": 1.7941,
      "step": 956
    },
    {
      "epoch": 0.8844731977818854,
      "grad_norm": 2.203125,
      "learning_rate": 1.2709806192870272e-05,
      "loss": 1.682,
      "step": 957
    },
    {
      "epoch": 0.8853974121996303,
      "grad_norm": 2.203125,
      "learning_rate": 1.2695087463468054e-05,
      "loss": 1.7269,
      "step": 958
    },
    {
      "epoch": 0.8863216266173752,
      "grad_norm": 2.265625,
      "learning_rate": 1.2680362435395595e-05,
      "loss": 1.7166,
      "step": 959
    },
    {
      "epoch": 0.8872458410351202,
      "grad_norm": 2.171875,
      "learning_rate": 1.2665631143066647e-05,
      "loss": 1.7366,
      "step": 960
    },
    {
      "epoch": 0.8881700554528651,
      "grad_norm": 2.3125,
      "learning_rate": 1.2650893620909621e-05,
      "loss": 1.765,
      "step": 961
    },
    {
      "epoch": 0.88909426987061,
      "grad_norm": 2.375,
      "learning_rate": 1.2636149903367472e-05,
      "loss": 1.745,
      "step": 962
    },
    {
      "epoch": 0.890018484288355,
      "grad_norm": 2.34375,
      "learning_rate": 1.2621400024897642e-05,
      "loss": 1.7293,
      "step": 963
    },
    {
      "epoch": 0.8909426987060998,
      "grad_norm": 2.390625,
      "learning_rate": 1.2606644019971967e-05,
      "loss": 1.8037,
      "step": 964
    },
    {
      "epoch": 0.8918669131238447,
      "grad_norm": 2.328125,
      "learning_rate": 1.259188192307661e-05,
      "loss": 1.7243,
      "step": 965
    },
    {
      "epoch": 0.8927911275415896,
      "grad_norm": 2.40625,
      "learning_rate": 1.2577113768711959e-05,
      "loss": 1.7914,
      "step": 966
    },
    {
      "epoch": 0.8937153419593346,
      "grad_norm": 2.234375,
      "learning_rate": 1.2562339591392571e-05,
      "loss": 1.7615,
      "step": 967
    },
    {
      "epoch": 0.8946395563770795,
      "grad_norm": 2.328125,
      "learning_rate": 1.2547559425647068e-05,
      "loss": 1.8003,
      "step": 968
    },
    {
      "epoch": 0.8955637707948244,
      "grad_norm": 2.28125,
      "learning_rate": 1.2532773306018075e-05,
      "loss": 1.796,
      "step": 969
    },
    {
      "epoch": 0.8964879852125693,
      "grad_norm": 2.21875,
      "learning_rate": 1.2517981267062133e-05,
      "loss": 1.6651,
      "step": 970
    },
    {
      "epoch": 0.8974121996303143,
      "grad_norm": 2.234375,
      "learning_rate": 1.2503183343349612e-05,
      "loss": 1.8054,
      "step": 971
    },
    {
      "epoch": 0.8983364140480592,
      "grad_norm": 2.359375,
      "learning_rate": 1.2488379569464634e-05,
      "loss": 1.8059,
      "step": 972
    },
    {
      "epoch": 0.899260628465804,
      "grad_norm": 2.28125,
      "learning_rate": 1.2473569980005001e-05,
      "loss": 1.7695,
      "step": 973
    },
    {
      "epoch": 0.9001848428835489,
      "grad_norm": 2.203125,
      "learning_rate": 1.2458754609582098e-05,
      "loss": 1.7513,
      "step": 974
    },
    {
      "epoch": 0.9011090573012939,
      "grad_norm": 2.328125,
      "learning_rate": 1.2443933492820826e-05,
      "loss": 1.6973,
      "step": 975
    },
    {
      "epoch": 0.9020332717190388,
      "grad_norm": 2.265625,
      "learning_rate": 1.2429106664359515e-05,
      "loss": 1.7944,
      "step": 976
    },
    {
      "epoch": 0.9029574861367837,
      "grad_norm": 2.234375,
      "learning_rate": 1.2414274158849845e-05,
      "loss": 1.7366,
      "step": 977
    },
    {
      "epoch": 0.9038817005545287,
      "grad_norm": 2.5,
      "learning_rate": 1.2399436010956753e-05,
      "loss": 1.8152,
      "step": 978
    },
    {
      "epoch": 0.9048059149722736,
      "grad_norm": 2.28125,
      "learning_rate": 1.2384592255358385e-05,
      "loss": 1.7479,
      "step": 979
    },
    {
      "epoch": 0.9057301293900185,
      "grad_norm": 2.3125,
      "learning_rate": 1.2369742926745969e-05,
      "loss": 1.8034,
      "step": 980
    },
    {
      "epoch": 0.9066543438077634,
      "grad_norm": 2.171875,
      "learning_rate": 1.2354888059823773e-05,
      "loss": 1.7525,
      "step": 981
    },
    {
      "epoch": 0.9075785582255084,
      "grad_norm": 2.328125,
      "learning_rate": 1.2340027689309e-05,
      "loss": 1.8198,
      "step": 982
    },
    {
      "epoch": 0.9085027726432532,
      "grad_norm": 2.234375,
      "learning_rate": 1.2325161849931726e-05,
      "loss": 1.7491,
      "step": 983
    },
    {
      "epoch": 0.9094269870609981,
      "grad_norm": 2.34375,
      "learning_rate": 1.2310290576434795e-05,
      "loss": 1.7972,
      "step": 984
    },
    {
      "epoch": 0.910351201478743,
      "grad_norm": 2.1875,
      "learning_rate": 1.2295413903573757e-05,
      "loss": 1.6824,
      "step": 985
    },
    {
      "epoch": 0.911275415896488,
      "grad_norm": 2.375,
      "learning_rate": 1.2280531866116783e-05,
      "loss": 1.7569,
      "step": 986
    },
    {
      "epoch": 0.9121996303142329,
      "grad_norm": 2.296875,
      "learning_rate": 1.226564449884458e-05,
      "loss": 1.7574,
      "step": 987
    },
    {
      "epoch": 0.9131238447319778,
      "grad_norm": 2.296875,
      "learning_rate": 1.2250751836550312e-05,
      "loss": 1.7377,
      "step": 988
    },
    {
      "epoch": 0.9140480591497228,
      "grad_norm": 2.21875,
      "learning_rate": 1.2235853914039515e-05,
      "loss": 1.7504,
      "step": 989
    },
    {
      "epoch": 0.9149722735674677,
      "grad_norm": 2.28125,
      "learning_rate": 1.2220950766130017e-05,
      "loss": 1.7993,
      "step": 990
    },
    {
      "epoch": 0.9158964879852126,
      "grad_norm": 2.15625,
      "learning_rate": 1.2206042427651873e-05,
      "loss": 1.7808,
      "step": 991
    },
    {
      "epoch": 0.9168207024029574,
      "grad_norm": 2.5,
      "learning_rate": 1.2191128933447244e-05,
      "loss": 1.8045,
      "step": 992
    },
    {
      "epoch": 0.9177449168207024,
      "grad_norm": 2.375,
      "learning_rate": 1.2176210318370365e-05,
      "loss": 1.8367,
      "step": 993
    },
    {
      "epoch": 0.9186691312384473,
      "grad_norm": 2.328125,
      "learning_rate": 1.216128661728742e-05,
      "loss": 1.7854,
      "step": 994
    },
    {
      "epoch": 0.9195933456561922,
      "grad_norm": 2.25,
      "learning_rate": 1.214635786507649e-05,
      "loss": 1.8107,
      "step": 995
    },
    {
      "epoch": 0.9205175600739371,
      "grad_norm": 2.359375,
      "learning_rate": 1.2131424096627465e-05,
      "loss": 1.8171,
      "step": 996
    },
    {
      "epoch": 0.9214417744916821,
      "grad_norm": 2.34375,
      "learning_rate": 1.2116485346841939e-05,
      "loss": 1.6712,
      "step": 997
    },
    {
      "epoch": 0.922365988909427,
      "grad_norm": 2.578125,
      "learning_rate": 1.2101541650633173e-05,
      "loss": 1.7513,
      "step": 998
    },
    {
      "epoch": 0.9232902033271719,
      "grad_norm": 2.203125,
      "learning_rate": 1.2086593042925964e-05,
      "loss": 1.697,
      "step": 999
    },
    {
      "epoch": 0.9242144177449169,
      "grad_norm": 2.25,
      "learning_rate": 1.2071639558656614e-05,
      "loss": 1.7592,
      "step": 1000
    },
    {
      "epoch": 0.9251386321626618,
      "grad_norm": 2.28125,
      "learning_rate": 1.2056681232772798e-05,
      "loss": 1.8038,
      "step": 1001
    },
    {
      "epoch": 0.9260628465804066,
      "grad_norm": 2.375,
      "learning_rate": 1.2041718100233513e-05,
      "loss": 1.6955,
      "step": 1002
    },
    {
      "epoch": 0.9269870609981515,
      "grad_norm": 2.140625,
      "learning_rate": 1.2026750196008999e-05,
      "loss": 1.7241,
      "step": 1003
    },
    {
      "epoch": 0.9279112754158965,
      "grad_norm": 2.25,
      "learning_rate": 1.2011777555080639e-05,
      "loss": 1.7432,
      "step": 1004
    },
    {
      "epoch": 0.9288354898336414,
      "grad_norm": 2.265625,
      "learning_rate": 1.1996800212440891e-05,
      "loss": 1.7739,
      "step": 1005
    },
    {
      "epoch": 0.9297597042513863,
      "grad_norm": 2.1875,
      "learning_rate": 1.1981818203093193e-05,
      "loss": 1.7539,
      "step": 1006
    },
    {
      "epoch": 0.9306839186691312,
      "grad_norm": 2.328125,
      "learning_rate": 1.1966831562051899e-05,
      "loss": 1.7583,
      "step": 1007
    },
    {
      "epoch": 0.9316081330868762,
      "grad_norm": 2.234375,
      "learning_rate": 1.1951840324342182e-05,
      "loss": 1.7792,
      "step": 1008
    },
    {
      "epoch": 0.9325323475046211,
      "grad_norm": 2.21875,
      "learning_rate": 1.1936844524999966e-05,
      "loss": 1.8641,
      "step": 1009
    },
    {
      "epoch": 0.933456561922366,
      "grad_norm": 2.4375,
      "learning_rate": 1.1921844199071823e-05,
      "loss": 1.8336,
      "step": 1010
    },
    {
      "epoch": 0.9343807763401109,
      "grad_norm": 2.28125,
      "learning_rate": 1.1906839381614913e-05,
      "loss": 1.6826,
      "step": 1011
    },
    {
      "epoch": 0.9353049907578558,
      "grad_norm": 2.265625,
      "learning_rate": 1.189183010769689e-05,
      "loss": 1.7652,
      "step": 1012
    },
    {
      "epoch": 0.9362292051756007,
      "grad_norm": 2.296875,
      "learning_rate": 1.1876816412395825e-05,
      "loss": 1.818,
      "step": 1013
    },
    {
      "epoch": 0.9371534195933456,
      "grad_norm": 2.140625,
      "learning_rate": 1.1861798330800125e-05,
      "loss": 1.724,
      "step": 1014
    },
    {
      "epoch": 0.9380776340110906,
      "grad_norm": 2.234375,
      "learning_rate": 1.1846775898008438e-05,
      "loss": 1.6802,
      "step": 1015
    },
    {
      "epoch": 0.9390018484288355,
      "grad_norm": 2.328125,
      "learning_rate": 1.183174914912959e-05,
      "loss": 1.7649,
      "step": 1016
    },
    {
      "epoch": 0.9399260628465804,
      "grad_norm": 2.25,
      "learning_rate": 1.1816718119282492e-05,
      "loss": 1.798,
      "step": 1017
    },
    {
      "epoch": 0.9408502772643254,
      "grad_norm": 2.296875,
      "learning_rate": 1.1801682843596064e-05,
      "loss": 1.7898,
      "step": 1018
    },
    {
      "epoch": 0.9417744916820703,
      "grad_norm": 2.265625,
      "learning_rate": 1.1786643357209136e-05,
      "loss": 1.773,
      "step": 1019
    },
    {
      "epoch": 0.9426987060998152,
      "grad_norm": 2.28125,
      "learning_rate": 1.1771599695270393e-05,
      "loss": 1.7778,
      "step": 1020
    },
    {
      "epoch": 0.94362292051756,
      "grad_norm": 2.203125,
      "learning_rate": 1.1756551892938273e-05,
      "loss": 1.7781,
      "step": 1021
    },
    {
      "epoch": 0.944547134935305,
      "grad_norm": 2.3125,
      "learning_rate": 1.1741499985380892e-05,
      "loss": 1.7698,
      "step": 1022
    },
    {
      "epoch": 0.9454713493530499,
      "grad_norm": 2.21875,
      "learning_rate": 1.1726444007775954e-05,
      "loss": 1.7416,
      "step": 1023
    },
    {
      "epoch": 0.9463955637707948,
      "grad_norm": 2.21875,
      "learning_rate": 1.171138399531068e-05,
      "loss": 1.7448,
      "step": 1024
    },
    {
      "epoch": 0.9473197781885397,
      "grad_norm": 2.265625,
      "learning_rate": 1.169631998318173e-05,
      "loss": 1.7282,
      "step": 1025
    },
    {
      "epoch": 0.9482439926062847,
      "grad_norm": 2.234375,
      "learning_rate": 1.1681252006595092e-05,
      "loss": 1.7564,
      "step": 1026
    },
    {
      "epoch": 0.9491682070240296,
      "grad_norm": 2.390625,
      "learning_rate": 1.1666180100766035e-05,
      "loss": 1.7234,
      "step": 1027
    },
    {
      "epoch": 0.9500924214417745,
      "grad_norm": 2.265625,
      "learning_rate": 1.1651104300919005e-05,
      "loss": 1.7624,
      "step": 1028
    },
    {
      "epoch": 0.9510166358595195,
      "grad_norm": 2.28125,
      "learning_rate": 1.1636024642287546e-05,
      "loss": 1.7008,
      "step": 1029
    },
    {
      "epoch": 0.9519408502772643,
      "grad_norm": 2.375,
      "learning_rate": 1.162094116011423e-05,
      "loss": 1.774,
      "step": 1030
    },
    {
      "epoch": 0.9528650646950092,
      "grad_norm": 2.28125,
      "learning_rate": 1.1605853889650553e-05,
      "loss": 1.7911,
      "step": 1031
    },
    {
      "epoch": 0.9537892791127541,
      "grad_norm": 2.25,
      "learning_rate": 1.1590762866156873e-05,
      "loss": 1.7926,
      "step": 1032
    },
    {
      "epoch": 0.9547134935304991,
      "grad_norm": 2.3125,
      "learning_rate": 1.1575668124902317e-05,
      "loss": 1.8248,
      "step": 1033
    },
    {
      "epoch": 0.955637707948244,
      "grad_norm": 2.265625,
      "learning_rate": 1.1560569701164696e-05,
      "loss": 1.6901,
      "step": 1034
    },
    {
      "epoch": 0.9565619223659889,
      "grad_norm": 2.265625,
      "learning_rate": 1.1545467630230442e-05,
      "loss": 1.7256,
      "step": 1035
    },
    {
      "epoch": 0.9574861367837338,
      "grad_norm": 2.34375,
      "learning_rate": 1.1530361947394487e-05,
      "loss": 1.8637,
      "step": 1036
    },
    {
      "epoch": 0.9584103512014788,
      "grad_norm": 2.1875,
      "learning_rate": 1.1515252687960226e-05,
      "loss": 1.7705,
      "step": 1037
    },
    {
      "epoch": 0.9593345656192237,
      "grad_norm": 2.25,
      "learning_rate": 1.1500139887239401e-05,
      "loss": 1.7753,
      "step": 1038
    },
    {
      "epoch": 0.9602587800369686,
      "grad_norm": 2.21875,
      "learning_rate": 1.1485023580552039e-05,
      "loss": 1.6853,
      "step": 1039
    },
    {
      "epoch": 0.9611829944547134,
      "grad_norm": 2.296875,
      "learning_rate": 1.1469903803226351e-05,
      "loss": 1.7484,
      "step": 1040
    },
    {
      "epoch": 0.9621072088724584,
      "grad_norm": 2.25,
      "learning_rate": 1.1454780590598662e-05,
      "loss": 1.7542,
      "step": 1041
    },
    {
      "epoch": 0.9630314232902033,
      "grad_norm": 2.328125,
      "learning_rate": 1.1439653978013334e-05,
      "loss": 1.8801,
      "step": 1042
    },
    {
      "epoch": 0.9639556377079482,
      "grad_norm": 2.328125,
      "learning_rate": 1.1424524000822663e-05,
      "loss": 1.7408,
      "step": 1043
    },
    {
      "epoch": 0.9648798521256932,
      "grad_norm": 2.3125,
      "learning_rate": 1.1409390694386818e-05,
      "loss": 1.7599,
      "step": 1044
    },
    {
      "epoch": 0.9658040665434381,
      "grad_norm": 2.359375,
      "learning_rate": 1.139425409407374e-05,
      "loss": 1.7153,
      "step": 1045
    },
    {
      "epoch": 0.966728280961183,
      "grad_norm": 2.1875,
      "learning_rate": 1.1379114235259077e-05,
      "loss": 1.663,
      "step": 1046
    },
    {
      "epoch": 0.967652495378928,
      "grad_norm": 2.25,
      "learning_rate": 1.1363971153326085e-05,
      "loss": 1.7666,
      "step": 1047
    },
    {
      "epoch": 0.9685767097966729,
      "grad_norm": 2.296875,
      "learning_rate": 1.134882488366556e-05,
      "loss": 1.7926,
      "step": 1048
    },
    {
      "epoch": 0.9695009242144177,
      "grad_norm": 2.28125,
      "learning_rate": 1.133367546167574e-05,
      "loss": 1.7318,
      "step": 1049
    },
    {
      "epoch": 0.9704251386321626,
      "grad_norm": 2.3125,
      "learning_rate": 1.1318522922762233e-05,
      "loss": 1.7762,
      "step": 1050
    },
    {
      "epoch": 0.9713493530499075,
      "grad_norm": 2.359375,
      "learning_rate": 1.1303367302337937e-05,
      "loss": 1.7529,
      "step": 1051
    },
    {
      "epoch": 0.9722735674676525,
      "grad_norm": 2.375,
      "learning_rate": 1.1288208635822947e-05,
      "loss": 1.7774,
      "step": 1052
    },
    {
      "epoch": 0.9731977818853974,
      "grad_norm": 2.203125,
      "learning_rate": 1.1273046958644476e-05,
      "loss": 1.6978,
      "step": 1053
    },
    {
      "epoch": 0.9741219963031423,
      "grad_norm": 2.625,
      "learning_rate": 1.1257882306236776e-05,
      "loss": 1.8488,
      "step": 1054
    },
    {
      "epoch": 0.9750462107208873,
      "grad_norm": 2.296875,
      "learning_rate": 1.1242714714041046e-05,
      "loss": 1.8155,
      "step": 1055
    },
    {
      "epoch": 0.9759704251386322,
      "grad_norm": 2.3125,
      "learning_rate": 1.1227544217505368e-05,
      "loss": 1.741,
      "step": 1056
    },
    {
      "epoch": 0.9768946395563771,
      "grad_norm": 2.140625,
      "learning_rate": 1.1212370852084603e-05,
      "loss": 1.8001,
      "step": 1057
    },
    {
      "epoch": 0.977818853974122,
      "grad_norm": 2.234375,
      "learning_rate": 1.1197194653240314e-05,
      "loss": 1.7306,
      "step": 1058
    },
    {
      "epoch": 0.9787430683918669,
      "grad_norm": 2.203125,
      "learning_rate": 1.1182015656440692e-05,
      "loss": 1.7079,
      "step": 1059
    },
    {
      "epoch": 0.9796672828096118,
      "grad_norm": 2.203125,
      "learning_rate": 1.1166833897160465e-05,
      "loss": 1.8404,
      "step": 1060
    },
    {
      "epoch": 0.9805914972273567,
      "grad_norm": 2.28125,
      "learning_rate": 1.1151649410880823e-05,
      "loss": 1.7201,
      "step": 1061
    },
    {
      "epoch": 0.9815157116451017,
      "grad_norm": 2.265625,
      "learning_rate": 1.1136462233089318e-05,
      "loss": 1.763,
      "step": 1062
    },
    {
      "epoch": 0.9824399260628466,
      "grad_norm": 2.453125,
      "learning_rate": 1.1121272399279795e-05,
      "loss": 1.7126,
      "step": 1063
    },
    {
      "epoch": 0.9833641404805915,
      "grad_norm": 2.25,
      "learning_rate": 1.1106079944952317e-05,
      "loss": 1.7772,
      "step": 1064
    },
    {
      "epoch": 0.9842883548983364,
      "grad_norm": 2.3125,
      "learning_rate": 1.109088490561306e-05,
      "loss": 1.7543,
      "step": 1065
    },
    {
      "epoch": 0.9852125693160814,
      "grad_norm": 2.171875,
      "learning_rate": 1.1075687316774246e-05,
      "loss": 1.6266,
      "step": 1066
    },
    {
      "epoch": 0.9861367837338263,
      "grad_norm": 2.234375,
      "learning_rate": 1.1060487213954053e-05,
      "loss": 1.8199,
      "step": 1067
    },
    {
      "epoch": 0.9870609981515711,
      "grad_norm": 2.25,
      "learning_rate": 1.1045284632676535e-05,
      "loss": 1.6765,
      "step": 1068
    },
    {
      "epoch": 0.987985212569316,
      "grad_norm": 2.3125,
      "learning_rate": 1.1030079608471544e-05,
      "loss": 1.7144,
      "step": 1069
    },
    {
      "epoch": 0.988909426987061,
      "grad_norm": 2.234375,
      "learning_rate": 1.1014872176874635e-05,
      "loss": 1.7569,
      "step": 1070
    },
    {
      "epoch": 0.9898336414048059,
      "grad_norm": 2.375,
      "learning_rate": 1.099966237342699e-05,
      "loss": 1.7365,
      "step": 1071
    },
    {
      "epoch": 0.9907578558225508,
      "grad_norm": 2.1875,
      "learning_rate": 1.0984450233675334e-05,
      "loss": 1.7409,
      "step": 1072
    },
    {
      "epoch": 0.9916820702402958,
      "grad_norm": 2.34375,
      "learning_rate": 1.0969235793171858e-05,
      "loss": 1.7435,
      "step": 1073
    },
    {
      "epoch": 0.9926062846580407,
      "grad_norm": 2.203125,
      "learning_rate": 1.0954019087474123e-05,
      "loss": 1.749,
      "step": 1074
    },
    {
      "epoch": 0.9935304990757856,
      "grad_norm": 2.4375,
      "learning_rate": 1.0938800152144984e-05,
      "loss": 1.8421,
      "step": 1075
    },
    {
      "epoch": 0.9944547134935305,
      "grad_norm": 2.234375,
      "learning_rate": 1.092357902275251e-05,
      "loss": 1.6558,
      "step": 1076
    },
    {
      "epoch": 0.9953789279112755,
      "grad_norm": 2.21875,
      "learning_rate": 1.09083557348699e-05,
      "loss": 1.7726,
      "step": 1077
    },
    {
      "epoch": 0.9963031423290203,
      "grad_norm": 2.203125,
      "learning_rate": 1.0893130324075396e-05,
      "loss": 1.7757,
      "step": 1078
    },
    {
      "epoch": 0.9972273567467652,
      "grad_norm": 2.296875,
      "learning_rate": 1.0877902825952198e-05,
      "loss": 1.7686,
      "step": 1079
    },
    {
      "epoch": 0.9981515711645101,
      "grad_norm": 2.3125,
      "learning_rate": 1.086267327608838e-05,
      "loss": 1.681,
      "step": 1080
    },
    {
      "epoch": 0.9990757855822551,
      "grad_norm": 2.234375,
      "learning_rate": 1.0847441710076826e-05,
      "loss": 1.7529,
      "step": 1081
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.46875,
      "learning_rate": 1.0832208163515123e-05,
      "loss": 1.7201,
      "step": 1082
    },
    {
      "epoch": 1.0009242144177448,
      "grad_norm": 2.375,
      "learning_rate": 1.0816972672005483e-05,
      "loss": 1.7402,
      "step": 1083
    },
    {
      "epoch": 1.0018484288354899,
      "grad_norm": 2.25,
      "learning_rate": 1.080173527115467e-05,
      "loss": 1.6441,
      "step": 1084
    },
    {
      "epoch": 1.0027726432532347,
      "grad_norm": 2.140625,
      "learning_rate": 1.0786495996573907e-05,
      "loss": 1.6349,
      "step": 1085
    },
    {
      "epoch": 1.0036968576709797,
      "grad_norm": 2.1875,
      "learning_rate": 1.0771254883878796e-05,
      "loss": 1.6627,
      "step": 1086
    },
    {
      "epoch": 1.0046210720887245,
      "grad_norm": 2.234375,
      "learning_rate": 1.0756011968689242e-05,
      "loss": 1.6892,
      "step": 1087
    },
    {
      "epoch": 1.0055452865064696,
      "grad_norm": 2.421875,
      "learning_rate": 1.074076728662935e-05,
      "loss": 1.6846,
      "step": 1088
    },
    {
      "epoch": 1.0064695009242144,
      "grad_norm": 2.3125,
      "learning_rate": 1.0725520873327361e-05,
      "loss": 1.7464,
      "step": 1089
    },
    {
      "epoch": 1.0073937153419594,
      "grad_norm": 2.265625,
      "learning_rate": 1.0710272764415566e-05,
      "loss": 1.6771,
      "step": 1090
    },
    {
      "epoch": 1.0083179297597042,
      "grad_norm": 2.296875,
      "learning_rate": 1.069502299553021e-05,
      "loss": 1.5688,
      "step": 1091
    },
    {
      "epoch": 1.009242144177449,
      "grad_norm": 2.25,
      "learning_rate": 1.0679771602311429e-05,
      "loss": 1.6627,
      "step": 1092
    },
    {
      "epoch": 1.010166358595194,
      "grad_norm": 2.21875,
      "learning_rate": 1.0664518620403143e-05,
      "loss": 1.7261,
      "step": 1093
    },
    {
      "epoch": 1.011090573012939,
      "grad_norm": 2.296875,
      "learning_rate": 1.0649264085452988e-05,
      "loss": 1.7151,
      "step": 1094
    },
    {
      "epoch": 1.012014787430684,
      "grad_norm": 2.203125,
      "learning_rate": 1.063400803311224e-05,
      "loss": 1.6986,
      "step": 1095
    },
    {
      "epoch": 1.0129390018484288,
      "grad_norm": 2.3125,
      "learning_rate": 1.0618750499035711e-05,
      "loss": 1.6223,
      "step": 1096
    },
    {
      "epoch": 1.0138632162661738,
      "grad_norm": 2.21875,
      "learning_rate": 1.060349151888167e-05,
      "loss": 1.723,
      "step": 1097
    },
    {
      "epoch": 1.0147874306839186,
      "grad_norm": 2.265625,
      "learning_rate": 1.0588231128311783e-05,
      "loss": 1.5908,
      "step": 1098
    },
    {
      "epoch": 1.0157116451016637,
      "grad_norm": 2.296875,
      "learning_rate": 1.0572969362991e-05,
      "loss": 1.6652,
      "step": 1099
    },
    {
      "epoch": 1.0166358595194085,
      "grad_norm": 2.25,
      "learning_rate": 1.0557706258587489e-05,
      "loss": 1.6731,
      "step": 1100
    },
    {
      "epoch": 1.0175600739371533,
      "grad_norm": 2.296875,
      "learning_rate": 1.0542441850772542e-05,
      "loss": 1.6589,
      "step": 1101
    },
    {
      "epoch": 1.0184842883548983,
      "grad_norm": 2.21875,
      "learning_rate": 1.0527176175220499e-05,
      "loss": 1.7045,
      "step": 1102
    },
    {
      "epoch": 1.0194085027726432,
      "grad_norm": 2.28125,
      "learning_rate": 1.0511909267608674e-05,
      "loss": 1.6563,
      "step": 1103
    },
    {
      "epoch": 1.0203327171903882,
      "grad_norm": 2.375,
      "learning_rate": 1.049664116361724e-05,
      "loss": 1.6612,
      "step": 1104
    },
    {
      "epoch": 1.021256931608133,
      "grad_norm": 2.25,
      "learning_rate": 1.0481371898929187e-05,
      "loss": 1.6899,
      "step": 1105
    },
    {
      "epoch": 1.022181146025878,
      "grad_norm": 2.171875,
      "learning_rate": 1.04661015092302e-05,
      "loss": 1.5654,
      "step": 1106
    },
    {
      "epoch": 1.0231053604436229,
      "grad_norm": 2.265625,
      "learning_rate": 1.0450830030208604e-05,
      "loss": 1.6492,
      "step": 1107
    },
    {
      "epoch": 1.024029574861368,
      "grad_norm": 2.234375,
      "learning_rate": 1.0435557497555267e-05,
      "loss": 1.6367,
      "step": 1108
    },
    {
      "epoch": 1.0249537892791127,
      "grad_norm": 2.28125,
      "learning_rate": 1.042028394696352e-05,
      "loss": 1.6599,
      "step": 1109
    },
    {
      "epoch": 1.0258780036968578,
      "grad_norm": 2.265625,
      "learning_rate": 1.0405009414129073e-05,
      "loss": 1.6892,
      "step": 1110
    },
    {
      "epoch": 1.0268022181146026,
      "grad_norm": 2.3125,
      "learning_rate": 1.0389733934749924e-05,
      "loss": 1.7044,
      "step": 1111
    },
    {
      "epoch": 1.0277264325323474,
      "grad_norm": 2.25,
      "learning_rate": 1.03744575445263e-05,
      "loss": 1.7083,
      "step": 1112
    },
    {
      "epoch": 1.0286506469500925,
      "grad_norm": 2.265625,
      "learning_rate": 1.0359180279160544e-05,
      "loss": 1.7404,
      "step": 1113
    },
    {
      "epoch": 1.0295748613678373,
      "grad_norm": 2.296875,
      "learning_rate": 1.034390217435704e-05,
      "loss": 1.6475,
      "step": 1114
    },
    {
      "epoch": 1.0304990757855823,
      "grad_norm": 2.171875,
      "learning_rate": 1.0328623265822148e-05,
      "loss": 1.6842,
      "step": 1115
    },
    {
      "epoch": 1.0314232902033271,
      "grad_norm": 2.1875,
      "learning_rate": 1.0313343589264097e-05,
      "loss": 1.6968,
      "step": 1116
    },
    {
      "epoch": 1.0323475046210722,
      "grad_norm": 2.25,
      "learning_rate": 1.0298063180392916e-05,
      "loss": 1.5967,
      "step": 1117
    },
    {
      "epoch": 1.033271719038817,
      "grad_norm": 2.265625,
      "learning_rate": 1.028278207492034e-05,
      "loss": 1.6485,
      "step": 1118
    },
    {
      "epoch": 1.034195933456562,
      "grad_norm": 2.328125,
      "learning_rate": 1.0267500308559733e-05,
      "loss": 1.7137,
      "step": 1119
    },
    {
      "epoch": 1.0351201478743068,
      "grad_norm": 2.140625,
      "learning_rate": 1.025221791702601e-05,
      "loss": 1.6419,
      "step": 1120
    },
    {
      "epoch": 1.0360443622920517,
      "grad_norm": 2.265625,
      "learning_rate": 1.0236934936035539e-05,
      "loss": 1.6602,
      "step": 1121
    },
    {
      "epoch": 1.0369685767097967,
      "grad_norm": 2.15625,
      "learning_rate": 1.022165140130607e-05,
      "loss": 1.6326,
      "step": 1122
    },
    {
      "epoch": 1.0378927911275415,
      "grad_norm": 2.125,
      "learning_rate": 1.0206367348556647e-05,
      "loss": 1.5986,
      "step": 1123
    },
    {
      "epoch": 1.0388170055452866,
      "grad_norm": 2.21875,
      "learning_rate": 1.0191082813507521e-05,
      "loss": 1.7337,
      "step": 1124
    },
    {
      "epoch": 1.0397412199630314,
      "grad_norm": 2.359375,
      "learning_rate": 1.0175797831880073e-05,
      "loss": 1.7409,
      "step": 1125
    },
    {
      "epoch": 1.0406654343807764,
      "grad_norm": 2.25,
      "learning_rate": 1.0160512439396735e-05,
      "loss": 1.6662,
      "step": 1126
    },
    {
      "epoch": 1.0415896487985212,
      "grad_norm": 2.1875,
      "learning_rate": 1.0145226671780885e-05,
      "loss": 1.655,
      "step": 1127
    },
    {
      "epoch": 1.0425138632162663,
      "grad_norm": 2.203125,
      "learning_rate": 1.0129940564756784e-05,
      "loss": 1.663,
      "step": 1128
    },
    {
      "epoch": 1.043438077634011,
      "grad_norm": 2.296875,
      "learning_rate": 1.011465415404949e-05,
      "loss": 1.7817,
      "step": 1129
    },
    {
      "epoch": 1.044362292051756,
      "grad_norm": 2.234375,
      "learning_rate": 1.009936747538477e-05,
      "loss": 1.7201,
      "step": 1130
    },
    {
      "epoch": 1.045286506469501,
      "grad_norm": 2.28125,
      "learning_rate": 1.0084080564489006e-05,
      "loss": 1.703,
      "step": 1131
    },
    {
      "epoch": 1.0462107208872458,
      "grad_norm": 2.21875,
      "learning_rate": 1.0068793457089141e-05,
      "loss": 1.7028,
      "step": 1132
    },
    {
      "epoch": 1.0471349353049908,
      "grad_norm": 2.265625,
      "learning_rate": 1.0053506188912563e-05,
      "loss": 1.6678,
      "step": 1133
    },
    {
      "epoch": 1.0480591497227356,
      "grad_norm": 2.21875,
      "learning_rate": 1.003821879568704e-05,
      "loss": 1.7261,
      "step": 1134
    },
    {
      "epoch": 1.0489833641404807,
      "grad_norm": 2.203125,
      "learning_rate": 1.0022931313140639e-05,
      "loss": 1.5828,
      "step": 1135
    },
    {
      "epoch": 1.0499075785582255,
      "grad_norm": 2.234375,
      "learning_rate": 1.0007643777001618e-05,
      "loss": 1.7158,
      "step": 1136
    },
    {
      "epoch": 1.0508317929759705,
      "grad_norm": 2.234375,
      "learning_rate": 9.992356222998383e-06,
      "loss": 1.7005,
      "step": 1137
    },
    {
      "epoch": 1.0517560073937153,
      "grad_norm": 2.203125,
      "learning_rate": 9.977068686859366e-06,
      "loss": 1.7228,
      "step": 1138
    },
    {
      "epoch": 1.0526802218114604,
      "grad_norm": 2.1875,
      "learning_rate": 9.961781204312961e-06,
      "loss": 1.6846,
      "step": 1139
    },
    {
      "epoch": 1.0536044362292052,
      "grad_norm": 2.171875,
      "learning_rate": 9.946493811087438e-06,
      "loss": 1.6803,
      "step": 1140
    },
    {
      "epoch": 1.05452865064695,
      "grad_norm": 2.296875,
      "learning_rate": 9.93120654291086e-06,
      "loss": 1.7282,
      "step": 1141
    },
    {
      "epoch": 1.055452865064695,
      "grad_norm": 2.203125,
      "learning_rate": 9.915919435510995e-06,
      "loss": 1.6489,
      "step": 1142
    },
    {
      "epoch": 1.0563770794824399,
      "grad_norm": 2.171875,
      "learning_rate": 9.900632524615235e-06,
      "loss": 1.7029,
      "step": 1143
    },
    {
      "epoch": 1.057301293900185,
      "grad_norm": 2.1875,
      "learning_rate": 9.88534584595051e-06,
      "loss": 1.6978,
      "step": 1144
    },
    {
      "epoch": 1.0582255083179297,
      "grad_norm": 2.171875,
      "learning_rate": 9.870059435243216e-06,
      "loss": 1.6167,
      "step": 1145
    },
    {
      "epoch": 1.0591497227356748,
      "grad_norm": 2.203125,
      "learning_rate": 9.854773328219118e-06,
      "loss": 1.752,
      "step": 1146
    },
    {
      "epoch": 1.0600739371534196,
      "grad_norm": 2.265625,
      "learning_rate": 9.839487560603266e-06,
      "loss": 1.688,
      "step": 1147
    },
    {
      "epoch": 1.0609981515711646,
      "grad_norm": 2.15625,
      "learning_rate": 9.824202168119928e-06,
      "loss": 1.636,
      "step": 1148
    },
    {
      "epoch": 1.0619223659889094,
      "grad_norm": 2.234375,
      "learning_rate": 9.80891718649248e-06,
      "loss": 1.7069,
      "step": 1149
    },
    {
      "epoch": 1.0628465804066543,
      "grad_norm": 2.234375,
      "learning_rate": 9.793632651443356e-06,
      "loss": 1.7015,
      "step": 1150
    },
    {
      "epoch": 1.0637707948243993,
      "grad_norm": 2.140625,
      "learning_rate": 9.778348598693933e-06,
      "loss": 1.6519,
      "step": 1151
    },
    {
      "epoch": 1.064695009242144,
      "grad_norm": 2.203125,
      "learning_rate": 9.763065063964465e-06,
      "loss": 1.6563,
      "step": 1152
    },
    {
      "epoch": 1.0656192236598891,
      "grad_norm": 2.203125,
      "learning_rate": 9.747782082973995e-06,
      "loss": 1.701,
      "step": 1153
    },
    {
      "epoch": 1.066543438077634,
      "grad_norm": 2.203125,
      "learning_rate": 9.732499691440267e-06,
      "loss": 1.696,
      "step": 1154
    },
    {
      "epoch": 1.067467652495379,
      "grad_norm": 2.1875,
      "learning_rate": 9.717217925079663e-06,
      "loss": 1.6334,
      "step": 1155
    },
    {
      "epoch": 1.0683918669131238,
      "grad_norm": 2.171875,
      "learning_rate": 9.701936819607086e-06,
      "loss": 1.6638,
      "step": 1156
    },
    {
      "epoch": 1.0693160813308689,
      "grad_norm": 2.3125,
      "learning_rate": 9.686656410735906e-06,
      "loss": 1.7437,
      "step": 1157
    },
    {
      "epoch": 1.0702402957486137,
      "grad_norm": 2.28125,
      "learning_rate": 9.671376734177857e-06,
      "loss": 1.7062,
      "step": 1158
    },
    {
      "epoch": 1.0711645101663585,
      "grad_norm": 2.1875,
      "learning_rate": 9.65609782564296e-06,
      "loss": 1.6728,
      "step": 1159
    },
    {
      "epoch": 1.0720887245841035,
      "grad_norm": 2.234375,
      "learning_rate": 9.64081972083946e-06,
      "loss": 1.7693,
      "step": 1160
    },
    {
      "epoch": 1.0730129390018484,
      "grad_norm": 2.21875,
      "learning_rate": 9.625542455473703e-06,
      "loss": 1.6582,
      "step": 1161
    },
    {
      "epoch": 1.0739371534195934,
      "grad_norm": 2.265625,
      "learning_rate": 9.610266065250077e-06,
      "loss": 1.6998,
      "step": 1162
    },
    {
      "epoch": 1.0748613678373382,
      "grad_norm": 2.171875,
      "learning_rate": 9.594990585870934e-06,
      "loss": 1.7135,
      "step": 1163
    },
    {
      "epoch": 1.0757855822550833,
      "grad_norm": 2.28125,
      "learning_rate": 9.57971605303648e-06,
      "loss": 1.7311,
      "step": 1164
    },
    {
      "epoch": 1.076709796672828,
      "grad_norm": 2.21875,
      "learning_rate": 9.564442502444734e-06,
      "loss": 1.6864,
      "step": 1165
    },
    {
      "epoch": 1.077634011090573,
      "grad_norm": 2.203125,
      "learning_rate": 9.5491699697914e-06,
      "loss": 1.7015,
      "step": 1166
    },
    {
      "epoch": 1.078558225508318,
      "grad_norm": 2.171875,
      "learning_rate": 9.533898490769804e-06,
      "loss": 1.6583,
      "step": 1167
    },
    {
      "epoch": 1.0794824399260627,
      "grad_norm": 2.171875,
      "learning_rate": 9.518628101070818e-06,
      "loss": 1.6718,
      "step": 1168
    },
    {
      "epoch": 1.0804066543438078,
      "grad_norm": 2.265625,
      "learning_rate": 9.503358836382761e-06,
      "loss": 1.7867,
      "step": 1169
    },
    {
      "epoch": 1.0813308687615526,
      "grad_norm": 2.21875,
      "learning_rate": 9.488090732391331e-06,
      "loss": 1.6962,
      "step": 1170
    },
    {
      "epoch": 1.0822550831792976,
      "grad_norm": 2.21875,
      "learning_rate": 9.472823824779504e-06,
      "loss": 1.7186,
      "step": 1171
    },
    {
      "epoch": 1.0831792975970425,
      "grad_norm": 2.34375,
      "learning_rate": 9.457558149227463e-06,
      "loss": 1.6749,
      "step": 1172
    },
    {
      "epoch": 1.0841035120147875,
      "grad_norm": 2.25,
      "learning_rate": 9.442293741412516e-06,
      "loss": 1.7142,
      "step": 1173
    },
    {
      "epoch": 1.0850277264325323,
      "grad_norm": 2.234375,
      "learning_rate": 9.427030637009002e-06,
      "loss": 1.7111,
      "step": 1174
    },
    {
      "epoch": 1.0859519408502774,
      "grad_norm": 2.328125,
      "learning_rate": 9.411768871688219e-06,
      "loss": 1.6949,
      "step": 1175
    },
    {
      "epoch": 1.0868761552680222,
      "grad_norm": 2.171875,
      "learning_rate": 9.396508481118333e-06,
      "loss": 1.7108,
      "step": 1176
    },
    {
      "epoch": 1.087800369685767,
      "grad_norm": 2.171875,
      "learning_rate": 9.381249500964294e-06,
      "loss": 1.663,
      "step": 1177
    },
    {
      "epoch": 1.088724584103512,
      "grad_norm": 2.203125,
      "learning_rate": 9.36599196688776e-06,
      "loss": 1.7298,
      "step": 1178
    },
    {
      "epoch": 1.0896487985212568,
      "grad_norm": 2.25,
      "learning_rate": 9.350735914547012e-06,
      "loss": 1.6764,
      "step": 1179
    },
    {
      "epoch": 1.0905730129390019,
      "grad_norm": 2.1875,
      "learning_rate": 9.33548137959686e-06,
      "loss": 1.6757,
      "step": 1180
    },
    {
      "epoch": 1.0914972273567467,
      "grad_norm": 2.3125,
      "learning_rate": 9.320228397688575e-06,
      "loss": 1.7313,
      "step": 1181
    },
    {
      "epoch": 1.0924214417744917,
      "grad_norm": 2.140625,
      "learning_rate": 9.304977004469791e-06,
      "loss": 1.6382,
      "step": 1182
    },
    {
      "epoch": 1.0933456561922366,
      "grad_norm": 2.203125,
      "learning_rate": 9.289727235584436e-06,
      "loss": 1.6057,
      "step": 1183
    },
    {
      "epoch": 1.0942698706099816,
      "grad_norm": 2.1875,
      "learning_rate": 9.27447912667264e-06,
      "loss": 1.7057,
      "step": 1184
    },
    {
      "epoch": 1.0951940850277264,
      "grad_norm": 2.28125,
      "learning_rate": 9.259232713370654e-06,
      "loss": 1.7117,
      "step": 1185
    },
    {
      "epoch": 1.0961182994454712,
      "grad_norm": 2.25,
      "learning_rate": 9.243988031310761e-06,
      "loss": 1.6676,
      "step": 1186
    },
    {
      "epoch": 1.0970425138632163,
      "grad_norm": 2.1875,
      "learning_rate": 9.228745116121205e-06,
      "loss": 1.6438,
      "step": 1187
    },
    {
      "epoch": 1.097966728280961,
      "grad_norm": 2.40625,
      "learning_rate": 9.213504003426094e-06,
      "loss": 1.6125,
      "step": 1188
    },
    {
      "epoch": 1.0988909426987061,
      "grad_norm": 2.328125,
      "learning_rate": 9.198264728845332e-06,
      "loss": 1.6903,
      "step": 1189
    },
    {
      "epoch": 1.099815157116451,
      "grad_norm": 2.265625,
      "learning_rate": 9.183027327994519e-06,
      "loss": 1.7749,
      "step": 1190
    },
    {
      "epoch": 1.100739371534196,
      "grad_norm": 2.234375,
      "learning_rate": 9.16779183648488e-06,
      "loss": 1.6739,
      "step": 1191
    },
    {
      "epoch": 1.1016635859519408,
      "grad_norm": 2.15625,
      "learning_rate": 9.152558289923179e-06,
      "loss": 1.6347,
      "step": 1192
    },
    {
      "epoch": 1.1025878003696858,
      "grad_norm": 2.3125,
      "learning_rate": 9.13732672391162e-06,
      "loss": 1.6656,
      "step": 1193
    },
    {
      "epoch": 1.1035120147874307,
      "grad_norm": 2.203125,
      "learning_rate": 9.122097174047805e-06,
      "loss": 1.7041,
      "step": 1194
    },
    {
      "epoch": 1.1044362292051757,
      "grad_norm": 2.15625,
      "learning_rate": 9.106869675924605e-06,
      "loss": 1.6654,
      "step": 1195
    },
    {
      "epoch": 1.1053604436229205,
      "grad_norm": 2.296875,
      "learning_rate": 9.091644265130101e-06,
      "loss": 1.7383,
      "step": 1196
    },
    {
      "epoch": 1.1062846580406653,
      "grad_norm": 2.21875,
      "learning_rate": 9.076420977247493e-06,
      "loss": 1.7195,
      "step": 1197
    },
    {
      "epoch": 1.1072088724584104,
      "grad_norm": 2.296875,
      "learning_rate": 9.061199847855018e-06,
      "loss": 1.6982,
      "step": 1198
    },
    {
      "epoch": 1.1081330868761552,
      "grad_norm": 2.3125,
      "learning_rate": 9.045980912525878e-06,
      "loss": 1.7115,
      "step": 1199
    },
    {
      "epoch": 1.1090573012939002,
      "grad_norm": 2.203125,
      "learning_rate": 9.030764206828146e-06,
      "loss": 1.7063,
      "step": 1200
    },
    {
      "epoch": 1.109981515711645,
      "grad_norm": 2.40625,
      "learning_rate": 9.015549766324668e-06,
      "loss": 1.7588,
      "step": 1201
    },
    {
      "epoch": 1.11090573012939,
      "grad_norm": 2.21875,
      "learning_rate": 9.000337626573014e-06,
      "loss": 1.7724,
      "step": 1202
    },
    {
      "epoch": 1.111829944547135,
      "grad_norm": 2.203125,
      "learning_rate": 8.985127823125365e-06,
      "loss": 1.7107,
      "step": 1203
    },
    {
      "epoch": 1.11275415896488,
      "grad_norm": 2.203125,
      "learning_rate": 8.969920391528459e-06,
      "loss": 1.7027,
      "step": 1204
    },
    {
      "epoch": 1.1136783733826248,
      "grad_norm": 2.203125,
      "learning_rate": 8.954715367323468e-06,
      "loss": 1.7218,
      "step": 1205
    },
    {
      "epoch": 1.1146025878003696,
      "grad_norm": 2.171875,
      "learning_rate": 8.939512786045953e-06,
      "loss": 1.7096,
      "step": 1206
    },
    {
      "epoch": 1.1155268022181146,
      "grad_norm": 2.25,
      "learning_rate": 8.92431268322576e-06,
      "loss": 1.6417,
      "step": 1207
    },
    {
      "epoch": 1.1164510166358594,
      "grad_norm": 2.203125,
      "learning_rate": 8.909115094386943e-06,
      "loss": 1.6267,
      "step": 1208
    },
    {
      "epoch": 1.1173752310536045,
      "grad_norm": 2.140625,
      "learning_rate": 8.893920055047686e-06,
      "loss": 1.6038,
      "step": 1209
    },
    {
      "epoch": 1.1182994454713493,
      "grad_norm": 2.390625,
      "learning_rate": 8.878727600720207e-06,
      "loss": 1.6731,
      "step": 1210
    },
    {
      "epoch": 1.1192236598890943,
      "grad_norm": 2.25,
      "learning_rate": 8.863537766910688e-06,
      "loss": 1.6885,
      "step": 1211
    },
    {
      "epoch": 1.1201478743068392,
      "grad_norm": 2.15625,
      "learning_rate": 8.848350589119178e-06,
      "loss": 1.6652,
      "step": 1212
    },
    {
      "epoch": 1.1210720887245842,
      "grad_norm": 2.21875,
      "learning_rate": 8.833166102839535e-06,
      "loss": 1.643,
      "step": 1213
    },
    {
      "epoch": 1.121996303142329,
      "grad_norm": 2.3125,
      "learning_rate": 8.81798434355931e-06,
      "loss": 1.7911,
      "step": 1214
    },
    {
      "epoch": 1.122920517560074,
      "grad_norm": 2.1875,
      "learning_rate": 8.80280534675969e-06,
      "loss": 1.6125,
      "step": 1215
    },
    {
      "epoch": 1.1238447319778189,
      "grad_norm": 2.203125,
      "learning_rate": 8.787629147915402e-06,
      "loss": 1.6708,
      "step": 1216
    },
    {
      "epoch": 1.1247689463955637,
      "grad_norm": 2.1875,
      "learning_rate": 8.772455782494632e-06,
      "loss": 1.7758,
      "step": 1217
    },
    {
      "epoch": 1.1256931608133087,
      "grad_norm": 2.1875,
      "learning_rate": 8.757285285958954e-06,
      "loss": 1.657,
      "step": 1218
    },
    {
      "epoch": 1.1266173752310535,
      "grad_norm": 2.203125,
      "learning_rate": 8.742117693763229e-06,
      "loss": 1.7069,
      "step": 1219
    },
    {
      "epoch": 1.1275415896487986,
      "grad_norm": 2.203125,
      "learning_rate": 8.726953041355528e-06,
      "loss": 1.6104,
      "step": 1220
    },
    {
      "epoch": 1.1284658040665434,
      "grad_norm": 2.21875,
      "learning_rate": 8.711791364177056e-06,
      "loss": 1.6719,
      "step": 1221
    },
    {
      "epoch": 1.1293900184842884,
      "grad_norm": 2.15625,
      "learning_rate": 8.696632697662065e-06,
      "loss": 1.6633,
      "step": 1222
    },
    {
      "epoch": 1.1303142329020333,
      "grad_norm": 2.15625,
      "learning_rate": 8.681477077237768e-06,
      "loss": 1.6205,
      "step": 1223
    },
    {
      "epoch": 1.1312384473197783,
      "grad_norm": 2.21875,
      "learning_rate": 8.666324538324264e-06,
      "loss": 1.7377,
      "step": 1224
    },
    {
      "epoch": 1.1321626617375231,
      "grad_norm": 2.265625,
      "learning_rate": 8.651175116334444e-06,
      "loss": 1.6687,
      "step": 1225
    },
    {
      "epoch": 1.133086876155268,
      "grad_norm": 2.265625,
      "learning_rate": 8.636028846673917e-06,
      "loss": 1.7027,
      "step": 1226
    },
    {
      "epoch": 1.134011090573013,
      "grad_norm": 2.21875,
      "learning_rate": 8.620885764740925e-06,
      "loss": 1.7604,
      "step": 1227
    },
    {
      "epoch": 1.1349353049907578,
      "grad_norm": 2.203125,
      "learning_rate": 8.605745905926261e-06,
      "loss": 1.7464,
      "step": 1228
    },
    {
      "epoch": 1.1358595194085028,
      "grad_norm": 2.21875,
      "learning_rate": 8.590609305613184e-06,
      "loss": 1.6775,
      "step": 1229
    },
    {
      "epoch": 1.1367837338262476,
      "grad_norm": 2.25,
      "learning_rate": 8.57547599917734e-06,
      "loss": 1.764,
      "step": 1230
    },
    {
      "epoch": 1.1377079482439927,
      "grad_norm": 2.09375,
      "learning_rate": 8.560346021986672e-06,
      "loss": 1.5902,
      "step": 1231
    },
    {
      "epoch": 1.1386321626617375,
      "grad_norm": 2.1875,
      "learning_rate": 8.545219409401338e-06,
      "loss": 1.6642,
      "step": 1232
    },
    {
      "epoch": 1.1395563770794825,
      "grad_norm": 2.28125,
      "learning_rate": 8.530096196773652e-06,
      "loss": 1.6365,
      "step": 1233
    },
    {
      "epoch": 1.1404805914972274,
      "grad_norm": 2.140625,
      "learning_rate": 8.514976419447963e-06,
      "loss": 1.5889,
      "step": 1234
    },
    {
      "epoch": 1.1414048059149722,
      "grad_norm": 2.234375,
      "learning_rate": 8.499860112760602e-06,
      "loss": 1.6588,
      "step": 1235
    },
    {
      "epoch": 1.1423290203327172,
      "grad_norm": 2.265625,
      "learning_rate": 8.484747312039779e-06,
      "loss": 1.7309,
      "step": 1236
    },
    {
      "epoch": 1.143253234750462,
      "grad_norm": 2.484375,
      "learning_rate": 8.469638052605513e-06,
      "loss": 1.7574,
      "step": 1237
    },
    {
      "epoch": 1.144177449168207,
      "grad_norm": 2.234375,
      "learning_rate": 8.454532369769561e-06,
      "loss": 1.6976,
      "step": 1238
    },
    {
      "epoch": 1.145101663585952,
      "grad_norm": 2.125,
      "learning_rate": 8.439430298835305e-06,
      "loss": 1.5859,
      "step": 1239
    },
    {
      "epoch": 1.146025878003697,
      "grad_norm": 2.203125,
      "learning_rate": 8.424331875097688e-06,
      "loss": 1.7248,
      "step": 1240
    },
    {
      "epoch": 1.1469500924214417,
      "grad_norm": 2.1875,
      "learning_rate": 8.40923713384313e-06,
      "loss": 1.663,
      "step": 1241
    },
    {
      "epoch": 1.1478743068391868,
      "grad_norm": 2.203125,
      "learning_rate": 8.394146110349449e-06,
      "loss": 1.6929,
      "step": 1242
    },
    {
      "epoch": 1.1487985212569316,
      "grad_norm": 2.203125,
      "learning_rate": 8.379058839885774e-06,
      "loss": 1.6407,
      "step": 1243
    },
    {
      "epoch": 1.1497227356746764,
      "grad_norm": 2.125,
      "learning_rate": 8.363975357712457e-06,
      "loss": 1.6776,
      "step": 1244
    },
    {
      "epoch": 1.1506469500924215,
      "grad_norm": 2.1875,
      "learning_rate": 8.348895699081e-06,
      "loss": 1.7499,
      "step": 1245
    },
    {
      "epoch": 1.1515711645101663,
      "grad_norm": 2.265625,
      "learning_rate": 8.33381989923397e-06,
      "loss": 1.6716,
      "step": 1246
    },
    {
      "epoch": 1.1524953789279113,
      "grad_norm": 2.15625,
      "learning_rate": 8.31874799340491e-06,
      "loss": 1.6767,
      "step": 1247
    },
    {
      "epoch": 1.1534195933456561,
      "grad_norm": 2.1875,
      "learning_rate": 8.303680016818275e-06,
      "loss": 1.6544,
      "step": 1248
    },
    {
      "epoch": 1.1543438077634012,
      "grad_norm": 2.1875,
      "learning_rate": 8.288616004689321e-06,
      "loss": 1.6572,
      "step": 1249
    },
    {
      "epoch": 1.155268022181146,
      "grad_norm": 2.1875,
      "learning_rate": 8.27355599222405e-06,
      "loss": 1.7152,
      "step": 1250
    },
    {
      "epoch": 1.156192236598891,
      "grad_norm": 2.21875,
      "learning_rate": 8.258500014619112e-06,
      "loss": 1.6723,
      "step": 1251
    },
    {
      "epoch": 1.1571164510166358,
      "grad_norm": 2.265625,
      "learning_rate": 8.243448107061728e-06,
      "loss": 1.7413,
      "step": 1252
    },
    {
      "epoch": 1.1580406654343807,
      "grad_norm": 2.1875,
      "learning_rate": 8.228400304729609e-06,
      "loss": 1.6629,
      "step": 1253
    },
    {
      "epoch": 1.1589648798521257,
      "grad_norm": 2.1875,
      "learning_rate": 8.213356642790867e-06,
      "loss": 1.7081,
      "step": 1254
    },
    {
      "epoch": 1.1598890942698705,
      "grad_norm": 2.28125,
      "learning_rate": 8.198317156403941e-06,
      "loss": 1.7335,
      "step": 1255
    },
    {
      "epoch": 1.1608133086876156,
      "grad_norm": 2.171875,
      "learning_rate": 8.18328188071751e-06,
      "loss": 1.6918,
      "step": 1256
    },
    {
      "epoch": 1.1617375231053604,
      "grad_norm": 2.203125,
      "learning_rate": 8.168250850870412e-06,
      "loss": 1.7469,
      "step": 1257
    },
    {
      "epoch": 1.1626617375231054,
      "grad_norm": 2.203125,
      "learning_rate": 8.153224101991565e-06,
      "loss": 1.6904,
      "step": 1258
    },
    {
      "epoch": 1.1635859519408502,
      "grad_norm": 2.1875,
      "learning_rate": 8.138201669199878e-06,
      "loss": 1.6588,
      "step": 1259
    },
    {
      "epoch": 1.1645101663585953,
      "grad_norm": 2.375,
      "learning_rate": 8.123183587604176e-06,
      "loss": 1.756,
      "step": 1260
    },
    {
      "epoch": 1.16543438077634,
      "grad_norm": 2.359375,
      "learning_rate": 8.108169892303111e-06,
      "loss": 1.686,
      "step": 1261
    },
    {
      "epoch": 1.166358595194085,
      "grad_norm": 2.25,
      "learning_rate": 8.093160618385088e-06,
      "loss": 1.7059,
      "step": 1262
    },
    {
      "epoch": 1.16728280961183,
      "grad_norm": 2.171875,
      "learning_rate": 8.078155800928178e-06,
      "loss": 1.6089,
      "step": 1263
    },
    {
      "epoch": 1.1682070240295748,
      "grad_norm": 2.234375,
      "learning_rate": 8.063155475000037e-06,
      "loss": 1.6881,
      "step": 1264
    },
    {
      "epoch": 1.1691312384473198,
      "grad_norm": 2.21875,
      "learning_rate": 8.048159675657819e-06,
      "loss": 1.7123,
      "step": 1265
    },
    {
      "epoch": 1.1700554528650646,
      "grad_norm": 2.234375,
      "learning_rate": 8.033168437948103e-06,
      "loss": 1.7298,
      "step": 1266
    },
    {
      "epoch": 1.1709796672828097,
      "grad_norm": 2.1875,
      "learning_rate": 8.01818179690681e-06,
      "loss": 1.7091,
      "step": 1267
    },
    {
      "epoch": 1.1719038817005545,
      "grad_norm": 2.25,
      "learning_rate": 8.003199787559112e-06,
      "loss": 1.6816,
      "step": 1268
    },
    {
      "epoch": 1.1728280961182995,
      "grad_norm": 2.234375,
      "learning_rate": 7.988222444919364e-06,
      "loss": 1.7283,
      "step": 1269
    },
    {
      "epoch": 1.1737523105360443,
      "grad_norm": 2.265625,
      "learning_rate": 7.973249803991006e-06,
      "loss": 1.7014,
      "step": 1270
    },
    {
      "epoch": 1.1746765249537892,
      "grad_norm": 2.265625,
      "learning_rate": 7.958281899766487e-06,
      "loss": 1.7745,
      "step": 1271
    },
    {
      "epoch": 1.1756007393715342,
      "grad_norm": 2.265625,
      "learning_rate": 7.943318767227205e-06,
      "loss": 1.6489,
      "step": 1272
    },
    {
      "epoch": 1.1765249537892792,
      "grad_norm": 2.21875,
      "learning_rate": 7.928360441343387e-06,
      "loss": 1.7413,
      "step": 1273
    },
    {
      "epoch": 1.177449168207024,
      "grad_norm": 2.15625,
      "learning_rate": 7.913406957074038e-06,
      "loss": 1.6366,
      "step": 1274
    },
    {
      "epoch": 1.1783733826247689,
      "grad_norm": 2.25,
      "learning_rate": 7.898458349366832e-06,
      "loss": 1.7147,
      "step": 1275
    },
    {
      "epoch": 1.179297597042514,
      "grad_norm": 2.15625,
      "learning_rate": 7.883514653158061e-06,
      "loss": 1.6752,
      "step": 1276
    },
    {
      "epoch": 1.1802218114602587,
      "grad_norm": 2.234375,
      "learning_rate": 7.868575903372539e-06,
      "loss": 1.7498,
      "step": 1277
    },
    {
      "epoch": 1.1811460258780038,
      "grad_norm": 2.171875,
      "learning_rate": 7.853642134923511e-06,
      "loss": 1.6209,
      "step": 1278
    },
    {
      "epoch": 1.1820702402957486,
      "grad_norm": 2.21875,
      "learning_rate": 7.838713382712583e-06,
      "loss": 1.7511,
      "step": 1279
    },
    {
      "epoch": 1.1829944547134936,
      "grad_norm": 2.203125,
      "learning_rate": 7.823789681629642e-06,
      "loss": 1.6565,
      "step": 1280
    },
    {
      "epoch": 1.1839186691312384,
      "grad_norm": 2.21875,
      "learning_rate": 7.808871066552756e-06,
      "loss": 1.761,
      "step": 1281
    },
    {
      "epoch": 1.1848428835489835,
      "grad_norm": 2.21875,
      "learning_rate": 7.793957572348132e-06,
      "loss": 1.7429,
      "step": 1282
    },
    {
      "epoch": 1.1857670979667283,
      "grad_norm": 2.375,
      "learning_rate": 7.779049233869985e-06,
      "loss": 1.6138,
      "step": 1283
    },
    {
      "epoch": 1.1866913123844731,
      "grad_norm": 2.21875,
      "learning_rate": 7.76414608596049e-06,
      "loss": 1.6289,
      "step": 1284
    },
    {
      "epoch": 1.1876155268022182,
      "grad_norm": 2.171875,
      "learning_rate": 7.749248163449693e-06,
      "loss": 1.7154,
      "step": 1285
    },
    {
      "epoch": 1.188539741219963,
      "grad_norm": 2.265625,
      "learning_rate": 7.734355501155422e-06,
      "loss": 1.6469,
      "step": 1286
    },
    {
      "epoch": 1.189463955637708,
      "grad_norm": 2.15625,
      "learning_rate": 7.719468133883219e-06,
      "loss": 1.7062,
      "step": 1287
    },
    {
      "epoch": 1.1903881700554528,
      "grad_norm": 2.140625,
      "learning_rate": 7.704586096426247e-06,
      "loss": 1.651,
      "step": 1288
    },
    {
      "epoch": 1.1913123844731979,
      "grad_norm": 2.203125,
      "learning_rate": 7.68970942356521e-06,
      "loss": 1.7599,
      "step": 1289
    },
    {
      "epoch": 1.1922365988909427,
      "grad_norm": 2.21875,
      "learning_rate": 7.674838150068275e-06,
      "loss": 1.7037,
      "step": 1290
    },
    {
      "epoch": 1.1931608133086877,
      "grad_norm": 2.1875,
      "learning_rate": 7.659972310691e-06,
      "loss": 1.6401,
      "step": 1291
    },
    {
      "epoch": 1.1940850277264325,
      "grad_norm": 2.234375,
      "learning_rate": 7.64511194017623e-06,
      "loss": 1.662,
      "step": 1292
    },
    {
      "epoch": 1.1950092421441774,
      "grad_norm": 2.1875,
      "learning_rate": 7.630257073254035e-06,
      "loss": 1.6613,
      "step": 1293
    },
    {
      "epoch": 1.1959334565619224,
      "grad_norm": 2.21875,
      "learning_rate": 7.615407744641618e-06,
      "loss": 1.7004,
      "step": 1294
    },
    {
      "epoch": 1.1968576709796672,
      "grad_norm": 2.15625,
      "learning_rate": 7.600563989043247e-06,
      "loss": 1.6239,
      "step": 1295
    },
    {
      "epoch": 1.1977818853974123,
      "grad_norm": 2.140625,
      "learning_rate": 7.5857258411501596e-06,
      "loss": 1.6252,
      "step": 1296
    },
    {
      "epoch": 1.198706099815157,
      "grad_norm": 2.234375,
      "learning_rate": 7.570893335640488e-06,
      "loss": 1.6812,
      "step": 1297
    },
    {
      "epoch": 1.1996303142329021,
      "grad_norm": 2.234375,
      "learning_rate": 7.556066507179177e-06,
      "loss": 1.7231,
      "step": 1298
    },
    {
      "epoch": 1.200554528650647,
      "grad_norm": 2.140625,
      "learning_rate": 7.541245390417906e-06,
      "loss": 1.7012,
      "step": 1299
    },
    {
      "epoch": 1.201478743068392,
      "grad_norm": 2.15625,
      "learning_rate": 7.526430019995001e-06,
      "loss": 1.6178,
      "step": 1300
    },
    {
      "epoch": 1.2024029574861368,
      "grad_norm": 2.1875,
      "learning_rate": 7.511620430535367e-06,
      "loss": 1.6295,
      "step": 1301
    },
    {
      "epoch": 1.2033271719038816,
      "grad_norm": 2.21875,
      "learning_rate": 7.496816656650389e-06,
      "loss": 1.6817,
      "step": 1302
    },
    {
      "epoch": 1.2042513863216266,
      "grad_norm": 2.15625,
      "learning_rate": 7.482018732937868e-06,
      "loss": 1.6994,
      "step": 1303
    },
    {
      "epoch": 1.2051756007393715,
      "grad_norm": 2.296875,
      "learning_rate": 7.467226693981926e-06,
      "loss": 1.7356,
      "step": 1304
    },
    {
      "epoch": 1.2060998151571165,
      "grad_norm": 2.1875,
      "learning_rate": 7.452440574352932e-06,
      "loss": 1.7073,
      "step": 1305
    },
    {
      "epoch": 1.2070240295748613,
      "grad_norm": 2.21875,
      "learning_rate": 7.437660408607432e-06,
      "loss": 1.7186,
      "step": 1306
    },
    {
      "epoch": 1.2079482439926064,
      "grad_norm": 2.1875,
      "learning_rate": 7.422886231288042e-06,
      "loss": 1.6317,
      "step": 1307
    },
    {
      "epoch": 1.2088724584103512,
      "grad_norm": 2.1875,
      "learning_rate": 7.408118076923394e-06,
      "loss": 1.6258,
      "step": 1308
    },
    {
      "epoch": 1.2097966728280962,
      "grad_norm": 2.203125,
      "learning_rate": 7.393355980028039e-06,
      "loss": 1.7462,
      "step": 1309
    },
    {
      "epoch": 1.210720887245841,
      "grad_norm": 2.109375,
      "learning_rate": 7.378599975102361e-06,
      "loss": 1.6689,
      "step": 1310
    },
    {
      "epoch": 1.2116451016635859,
      "grad_norm": 2.140625,
      "learning_rate": 7.36385009663253e-06,
      "loss": 1.6948,
      "step": 1311
    },
    {
      "epoch": 1.212569316081331,
      "grad_norm": 2.140625,
      "learning_rate": 7.349106379090382e-06,
      "loss": 1.6806,
      "step": 1312
    },
    {
      "epoch": 1.2134935304990757,
      "grad_norm": 2.1875,
      "learning_rate": 7.334368856933356e-06,
      "loss": 1.622,
      "step": 1313
    },
    {
      "epoch": 1.2144177449168208,
      "grad_norm": 2.265625,
      "learning_rate": 7.319637564604412e-06,
      "loss": 1.7048,
      "step": 1314
    },
    {
      "epoch": 1.2153419593345656,
      "grad_norm": 2.25,
      "learning_rate": 7.304912536531945e-06,
      "loss": 1.7063,
      "step": 1315
    },
    {
      "epoch": 1.2162661737523106,
      "grad_norm": 2.21875,
      "learning_rate": 7.290193807129728e-06,
      "loss": 1.6709,
      "step": 1316
    },
    {
      "epoch": 1.2171903881700554,
      "grad_norm": 2.21875,
      "learning_rate": 7.275481410796799e-06,
      "loss": 1.6511,
      "step": 1317
    },
    {
      "epoch": 1.2181146025878005,
      "grad_norm": 2.1875,
      "learning_rate": 7.260775381917392e-06,
      "loss": 1.6339,
      "step": 1318
    },
    {
      "epoch": 1.2190388170055453,
      "grad_norm": 2.3125,
      "learning_rate": 7.246075754860868e-06,
      "loss": 1.8067,
      "step": 1319
    },
    {
      "epoch": 1.21996303142329,
      "grad_norm": 2.234375,
      "learning_rate": 7.23138256398162e-06,
      "loss": 1.6221,
      "step": 1320
    },
    {
      "epoch": 1.2208872458410351,
      "grad_norm": 2.171875,
      "learning_rate": 7.216695843619013e-06,
      "loss": 1.5519,
      "step": 1321
    },
    {
      "epoch": 1.22181146025878,
      "grad_norm": 2.1875,
      "learning_rate": 7.20201562809727e-06,
      "loss": 1.6872,
      "step": 1322
    },
    {
      "epoch": 1.222735674676525,
      "grad_norm": 2.21875,
      "learning_rate": 7.187341951725424e-06,
      "loss": 1.6383,
      "step": 1323
    },
    {
      "epoch": 1.2236598890942698,
      "grad_norm": 2.25,
      "learning_rate": 7.172674848797218e-06,
      "loss": 1.6802,
      "step": 1324
    },
    {
      "epoch": 1.2245841035120149,
      "grad_norm": 2.1875,
      "learning_rate": 7.1580143535910475e-06,
      "loss": 1.6682,
      "step": 1325
    },
    {
      "epoch": 1.2255083179297597,
      "grad_norm": 2.1875,
      "learning_rate": 7.143360500369845e-06,
      "loss": 1.7064,
      "step": 1326
    },
    {
      "epoch": 1.2264325323475047,
      "grad_norm": 2.234375,
      "learning_rate": 7.128713323381032e-06,
      "loss": 1.5928,
      "step": 1327
    },
    {
      "epoch": 1.2273567467652495,
      "grad_norm": 2.203125,
      "learning_rate": 7.114072856856424e-06,
      "loss": 1.7363,
      "step": 1328
    },
    {
      "epoch": 1.2282809611829943,
      "grad_norm": 2.203125,
      "learning_rate": 7.099439135012154e-06,
      "loss": 1.7233,
      "step": 1329
    },
    {
      "epoch": 1.2292051756007394,
      "grad_norm": 2.1875,
      "learning_rate": 7.084812192048593e-06,
      "loss": 1.7236,
      "step": 1330
    },
    {
      "epoch": 1.2301293900184842,
      "grad_norm": 2.125,
      "learning_rate": 7.070192062150269e-06,
      "loss": 1.5934,
      "step": 1331
    },
    {
      "epoch": 1.2310536044362292,
      "grad_norm": 2.234375,
      "learning_rate": 7.055578779485784e-06,
      "loss": 1.621,
      "step": 1332
    },
    {
      "epoch": 1.231977818853974,
      "grad_norm": 2.1875,
      "learning_rate": 7.040972378207739e-06,
      "loss": 1.7263,
      "step": 1333
    },
    {
      "epoch": 1.232902033271719,
      "grad_norm": 2.234375,
      "learning_rate": 7.026372892452653e-06,
      "loss": 1.7424,
      "step": 1334
    },
    {
      "epoch": 1.233826247689464,
      "grad_norm": 2.140625,
      "learning_rate": 7.011780356340888e-06,
      "loss": 1.6029,
      "step": 1335
    },
    {
      "epoch": 1.234750462107209,
      "grad_norm": 2.25,
      "learning_rate": 6.997194803976556e-06,
      "loss": 1.693,
      "step": 1336
    },
    {
      "epoch": 1.2356746765249538,
      "grad_norm": 2.21875,
      "learning_rate": 6.982616269447449e-06,
      "loss": 1.6927,
      "step": 1337
    },
    {
      "epoch": 1.2365988909426986,
      "grad_norm": 2.171875,
      "learning_rate": 6.9680447868249615e-06,
      "loss": 1.7218,
      "step": 1338
    },
    {
      "epoch": 1.2375231053604436,
      "grad_norm": 2.359375,
      "learning_rate": 6.953480390164001e-06,
      "loss": 1.5644,
      "step": 1339
    },
    {
      "epoch": 1.2384473197781884,
      "grad_norm": 2.265625,
      "learning_rate": 6.9389231135029225e-06,
      "loss": 1.6842,
      "step": 1340
    },
    {
      "epoch": 1.2393715341959335,
      "grad_norm": 2.1875,
      "learning_rate": 6.924372990863434e-06,
      "loss": 1.6772,
      "step": 1341
    },
    {
      "epoch": 1.2402957486136783,
      "grad_norm": 2.21875,
      "learning_rate": 6.909830056250527e-06,
      "loss": 1.7295,
      "step": 1342
    },
    {
      "epoch": 1.2412199630314233,
      "grad_norm": 2.171875,
      "learning_rate": 6.8952943436523934e-06,
      "loss": 1.7222,
      "step": 1343
    },
    {
      "epoch": 1.2421441774491682,
      "grad_norm": 2.171875,
      "learning_rate": 6.880765887040344e-06,
      "loss": 1.6912,
      "step": 1344
    },
    {
      "epoch": 1.2430683918669132,
      "grad_norm": 2.21875,
      "learning_rate": 6.866244720368738e-06,
      "loss": 1.8309,
      "step": 1345
    },
    {
      "epoch": 1.243992606284658,
      "grad_norm": 2.25,
      "learning_rate": 6.8517308775748915e-06,
      "loss": 1.6578,
      "step": 1346
    },
    {
      "epoch": 1.2449168207024028,
      "grad_norm": 2.1875,
      "learning_rate": 6.837224392579007e-06,
      "loss": 1.6651,
      "step": 1347
    },
    {
      "epoch": 1.2458410351201479,
      "grad_norm": 2.296875,
      "learning_rate": 6.822725299284091e-06,
      "loss": 1.6297,
      "step": 1348
    },
    {
      "epoch": 1.2467652495378927,
      "grad_norm": 2.15625,
      "learning_rate": 6.808233631575867e-06,
      "loss": 1.6741,
      "step": 1349
    },
    {
      "epoch": 1.2476894639556377,
      "grad_norm": 2.203125,
      "learning_rate": 6.793749423322719e-06,
      "loss": 1.7022,
      "step": 1350
    },
    {
      "epoch": 1.2486136783733826,
      "grad_norm": 2.3125,
      "learning_rate": 6.779272708375588e-06,
      "loss": 1.676,
      "step": 1351
    },
    {
      "epoch": 1.2495378927911276,
      "grad_norm": 2.125,
      "learning_rate": 6.764803520567905e-06,
      "loss": 1.6752,
      "step": 1352
    },
    {
      "epoch": 1.2504621072088724,
      "grad_norm": 2.21875,
      "learning_rate": 6.750341893715506e-06,
      "loss": 1.6942,
      "step": 1353
    },
    {
      "epoch": 1.2513863216266174,
      "grad_norm": 2.171875,
      "learning_rate": 6.735887861616555e-06,
      "loss": 1.641,
      "step": 1354
    },
    {
      "epoch": 1.2523105360443623,
      "grad_norm": 2.203125,
      "learning_rate": 6.721441458051481e-06,
      "loss": 1.6589,
      "step": 1355
    },
    {
      "epoch": 1.253234750462107,
      "grad_norm": 2.171875,
      "learning_rate": 6.70700271678287e-06,
      "loss": 1.6802,
      "step": 1356
    },
    {
      "epoch": 1.2541589648798521,
      "grad_norm": 2.21875,
      "learning_rate": 6.692571671555399e-06,
      "loss": 1.6807,
      "step": 1357
    },
    {
      "epoch": 1.2550831792975972,
      "grad_norm": 2.15625,
      "learning_rate": 6.678148356095768e-06,
      "loss": 1.6753,
      "step": 1358
    },
    {
      "epoch": 1.256007393715342,
      "grad_norm": 2.375,
      "learning_rate": 6.663732804112604e-06,
      "loss": 1.6712,
      "step": 1359
    },
    {
      "epoch": 1.2569316081330868,
      "grad_norm": 2.171875,
      "learning_rate": 6.649325049296401e-06,
      "loss": 1.7105,
      "step": 1360
    },
    {
      "epoch": 1.2578558225508318,
      "grad_norm": 2.234375,
      "learning_rate": 6.634925125319418e-06,
      "loss": 1.6819,
      "step": 1361
    },
    {
      "epoch": 1.2587800369685767,
      "grad_norm": 2.171875,
      "learning_rate": 6.620533065835617e-06,
      "loss": 1.6486,
      "step": 1362
    },
    {
      "epoch": 1.2597042513863217,
      "grad_norm": 2.109375,
      "learning_rate": 6.606148904480577e-06,
      "loss": 1.5767,
      "step": 1363
    },
    {
      "epoch": 1.2606284658040665,
      "grad_norm": 2.21875,
      "learning_rate": 6.591772674871434e-06,
      "loss": 1.66,
      "step": 1364
    },
    {
      "epoch": 1.2615526802218113,
      "grad_norm": 2.1875,
      "learning_rate": 6.577404410606765e-06,
      "loss": 1.7251,
      "step": 1365
    },
    {
      "epoch": 1.2624768946395564,
      "grad_norm": 2.171875,
      "learning_rate": 6.563044145266543e-06,
      "loss": 1.742,
      "step": 1366
    },
    {
      "epoch": 1.2634011090573014,
      "grad_norm": 2.15625,
      "learning_rate": 6.548691912412044e-06,
      "loss": 1.6591,
      "step": 1367
    },
    {
      "epoch": 1.2643253234750462,
      "grad_norm": 2.15625,
      "learning_rate": 6.534347745585772e-06,
      "loss": 1.6356,
      "step": 1368
    },
    {
      "epoch": 1.265249537892791,
      "grad_norm": 2.140625,
      "learning_rate": 6.520011678311382e-06,
      "loss": 1.7256,
      "step": 1369
    },
    {
      "epoch": 1.266173752310536,
      "grad_norm": 2.328125,
      "learning_rate": 6.505683744093597e-06,
      "loss": 1.7519,
      "step": 1370
    },
    {
      "epoch": 1.267097966728281,
      "grad_norm": 2.171875,
      "learning_rate": 6.491363976418131e-06,
      "loss": 1.6225,
      "step": 1371
    },
    {
      "epoch": 1.268022181146026,
      "grad_norm": 2.15625,
      "learning_rate": 6.477052408751616e-06,
      "loss": 1.6605,
      "step": 1372
    },
    {
      "epoch": 1.2689463955637708,
      "grad_norm": 2.203125,
      "learning_rate": 6.462749074541515e-06,
      "loss": 1.693,
      "step": 1373
    },
    {
      "epoch": 1.2698706099815156,
      "grad_norm": 2.125,
      "learning_rate": 6.448454007216054e-06,
      "loss": 1.6445,
      "step": 1374
    },
    {
      "epoch": 1.2707948243992606,
      "grad_norm": 2.125,
      "learning_rate": 6.434167240184135e-06,
      "loss": 1.6697,
      "step": 1375
    },
    {
      "epoch": 1.2717190388170057,
      "grad_norm": 2.203125,
      "learning_rate": 6.419888806835261e-06,
      "loss": 1.6962,
      "step": 1376
    },
    {
      "epoch": 1.2726432532347505,
      "grad_norm": 2.15625,
      "learning_rate": 6.405618740539463e-06,
      "loss": 1.632,
      "step": 1377
    },
    {
      "epoch": 1.2735674676524953,
      "grad_norm": 2.265625,
      "learning_rate": 6.391357074647209e-06,
      "loss": 1.634,
      "step": 1378
    },
    {
      "epoch": 1.2744916820702403,
      "grad_norm": 2.265625,
      "learning_rate": 6.377103842489344e-06,
      "loss": 1.6808,
      "step": 1379
    },
    {
      "epoch": 1.2754158964879851,
      "grad_norm": 2.25,
      "learning_rate": 6.362859077376997e-06,
      "loss": 1.659,
      "step": 1380
    },
    {
      "epoch": 1.2763401109057302,
      "grad_norm": 2.21875,
      "learning_rate": 6.348622812601508e-06,
      "loss": 1.7258,
      "step": 1381
    },
    {
      "epoch": 1.277264325323475,
      "grad_norm": 2.125,
      "learning_rate": 6.334395081434357e-06,
      "loss": 1.6225,
      "step": 1382
    },
    {
      "epoch": 1.27818853974122,
      "grad_norm": 2.265625,
      "learning_rate": 6.320175917127069e-06,
      "loss": 1.638,
      "step": 1383
    },
    {
      "epoch": 1.2791127541589649,
      "grad_norm": 2.125,
      "learning_rate": 6.305965352911162e-06,
      "loss": 1.6406,
      "step": 1384
    },
    {
      "epoch": 1.28003696857671,
      "grad_norm": 2.109375,
      "learning_rate": 6.291763421998043e-06,
      "loss": 1.6966,
      "step": 1385
    },
    {
      "epoch": 1.2809611829944547,
      "grad_norm": 2.21875,
      "learning_rate": 6.277570157578947e-06,
      "loss": 1.7182,
      "step": 1386
    },
    {
      "epoch": 1.2818853974121995,
      "grad_norm": 2.171875,
      "learning_rate": 6.263385592824858e-06,
      "loss": 1.6446,
      "step": 1387
    },
    {
      "epoch": 1.2828096118299446,
      "grad_norm": 2.1875,
      "learning_rate": 6.249209760886413e-06,
      "loss": 1.6408,
      "step": 1388
    },
    {
      "epoch": 1.2837338262476894,
      "grad_norm": 2.15625,
      "learning_rate": 6.2350426948938625e-06,
      "loss": 1.5935,
      "step": 1389
    },
    {
      "epoch": 1.2846580406654344,
      "grad_norm": 2.1875,
      "learning_rate": 6.220884427956953e-06,
      "loss": 1.6687,
      "step": 1390
    },
    {
      "epoch": 1.2855822550831792,
      "grad_norm": 2.09375,
      "learning_rate": 6.206734993164875e-06,
      "loss": 1.6454,
      "step": 1391
    },
    {
      "epoch": 1.2865064695009243,
      "grad_norm": 2.234375,
      "learning_rate": 6.192594423586171e-06,
      "loss": 1.6601,
      "step": 1392
    },
    {
      "epoch": 1.287430683918669,
      "grad_norm": 2.234375,
      "learning_rate": 6.178462752268666e-06,
      "loss": 1.7254,
      "step": 1393
    },
    {
      "epoch": 1.2883548983364141,
      "grad_norm": 2.171875,
      "learning_rate": 6.164340012239397e-06,
      "loss": 1.683,
      "step": 1394
    },
    {
      "epoch": 1.289279112754159,
      "grad_norm": 2.171875,
      "learning_rate": 6.150226236504523e-06,
      "loss": 1.668,
      "step": 1395
    },
    {
      "epoch": 1.2902033271719038,
      "grad_norm": 2.109375,
      "learning_rate": 6.136121458049247e-06,
      "loss": 1.6105,
      "step": 1396
    },
    {
      "epoch": 1.2911275415896488,
      "grad_norm": 2.15625,
      "learning_rate": 6.122025709837749e-06,
      "loss": 1.6908,
      "step": 1397
    },
    {
      "epoch": 1.2920517560073936,
      "grad_norm": 2.140625,
      "learning_rate": 6.107939024813101e-06,
      "loss": 1.6723,
      "step": 1398
    },
    {
      "epoch": 1.2929759704251387,
      "grad_norm": 2.171875,
      "learning_rate": 6.093861435897208e-06,
      "loss": 1.7832,
      "step": 1399
    },
    {
      "epoch": 1.2939001848428835,
      "grad_norm": 2.234375,
      "learning_rate": 6.079792975990699e-06,
      "loss": 1.7412,
      "step": 1400
    },
    {
      "epoch": 1.2948243992606285,
      "grad_norm": 2.25,
      "learning_rate": 6.0657336779728735e-06,
      "loss": 1.7124,
      "step": 1401
    },
    {
      "epoch": 1.2957486136783734,
      "grad_norm": 2.234375,
      "learning_rate": 6.051683574701617e-06,
      "loss": 1.7146,
      "step": 1402
    },
    {
      "epoch": 1.2966728280961184,
      "grad_norm": 2.203125,
      "learning_rate": 6.037642699013338e-06,
      "loss": 1.644,
      "step": 1403
    },
    {
      "epoch": 1.2975970425138632,
      "grad_norm": 2.21875,
      "learning_rate": 6.02361108372286e-06,
      "loss": 1.7089,
      "step": 1404
    },
    {
      "epoch": 1.298521256931608,
      "grad_norm": 2.1875,
      "learning_rate": 6.009588761623379e-06,
      "loss": 1.7159,
      "step": 1405
    },
    {
      "epoch": 1.299445471349353,
      "grad_norm": 2.140625,
      "learning_rate": 5.995575765486368e-06,
      "loss": 1.6629,
      "step": 1406
    },
    {
      "epoch": 1.300369685767098,
      "grad_norm": 2.203125,
      "learning_rate": 5.981572128061497e-06,
      "loss": 1.7625,
      "step": 1407
    },
    {
      "epoch": 1.301293900184843,
      "grad_norm": 2.203125,
      "learning_rate": 5.9675778820765754e-06,
      "loss": 1.6923,
      "step": 1408
    },
    {
      "epoch": 1.3022181146025877,
      "grad_norm": 2.234375,
      "learning_rate": 5.953593060237457e-06,
      "loss": 1.7193,
      "step": 1409
    },
    {
      "epoch": 1.3031423290203328,
      "grad_norm": 2.203125,
      "learning_rate": 5.939617695227974e-06,
      "loss": 1.5905,
      "step": 1410
    },
    {
      "epoch": 1.3040665434380776,
      "grad_norm": 2.234375,
      "learning_rate": 5.925651819709852e-06,
      "loss": 1.7245,
      "step": 1411
    },
    {
      "epoch": 1.3049907578558226,
      "grad_norm": 2.125,
      "learning_rate": 5.911695466322641e-06,
      "loss": 1.6268,
      "step": 1412
    },
    {
      "epoch": 1.3059149722735675,
      "grad_norm": 2.203125,
      "learning_rate": 5.897748667683643e-06,
      "loss": 1.6749,
      "step": 1413
    },
    {
      "epoch": 1.3068391866913123,
      "grad_norm": 2.140625,
      "learning_rate": 5.883811456387821e-06,
      "loss": 1.6378,
      "step": 1414
    },
    {
      "epoch": 1.3077634011090573,
      "grad_norm": 2.171875,
      "learning_rate": 5.8698838650077365e-06,
      "loss": 1.6425,
      "step": 1415
    },
    {
      "epoch": 1.3086876155268024,
      "grad_norm": 2.15625,
      "learning_rate": 5.855965926093466e-06,
      "loss": 1.7253,
      "step": 1416
    },
    {
      "epoch": 1.3096118299445472,
      "grad_norm": 2.1875,
      "learning_rate": 5.8420576721725255e-06,
      "loss": 1.7506,
      "step": 1417
    },
    {
      "epoch": 1.310536044362292,
      "grad_norm": 2.171875,
      "learning_rate": 5.828159135749802e-06,
      "loss": 1.6706,
      "step": 1418
    },
    {
      "epoch": 1.311460258780037,
      "grad_norm": 2.15625,
      "learning_rate": 5.8142703493074716e-06,
      "loss": 1.6328,
      "step": 1419
    },
    {
      "epoch": 1.3123844731977818,
      "grad_norm": 2.15625,
      "learning_rate": 5.800391345304915e-06,
      "loss": 1.7046,
      "step": 1420
    },
    {
      "epoch": 1.3133086876155269,
      "grad_norm": 2.1875,
      "learning_rate": 5.78652215617866e-06,
      "loss": 1.7224,
      "step": 1421
    },
    {
      "epoch": 1.3142329020332717,
      "grad_norm": 2.1875,
      "learning_rate": 5.772662814342292e-06,
      "loss": 1.788,
      "step": 1422
    },
    {
      "epoch": 1.3151571164510165,
      "grad_norm": 2.125,
      "learning_rate": 5.758813352186379e-06,
      "loss": 1.6454,
      "step": 1423
    },
    {
      "epoch": 1.3160813308687616,
      "grad_norm": 2.15625,
      "learning_rate": 5.744973802078409e-06,
      "loss": 1.6167,
      "step": 1424
    },
    {
      "epoch": 1.3170055452865066,
      "grad_norm": 2.171875,
      "learning_rate": 5.731144196362696e-06,
      "loss": 1.7531,
      "step": 1425
    },
    {
      "epoch": 1.3179297597042514,
      "grad_norm": 2.234375,
      "learning_rate": 5.717324567360314e-06,
      "loss": 1.7143,
      "step": 1426
    },
    {
      "epoch": 1.3188539741219962,
      "grad_norm": 2.15625,
      "learning_rate": 5.703514947369021e-06,
      "loss": 1.6752,
      "step": 1427
    },
    {
      "epoch": 1.3197781885397413,
      "grad_norm": 2.171875,
      "learning_rate": 5.689715368663191e-06,
      "loss": 1.6187,
      "step": 1428
    },
    {
      "epoch": 1.320702402957486,
      "grad_norm": 2.09375,
      "learning_rate": 5.675925863493721e-06,
      "loss": 1.6027,
      "step": 1429
    },
    {
      "epoch": 1.3216266173752311,
      "grad_norm": 2.1875,
      "learning_rate": 5.662146464087966e-06,
      "loss": 1.7121,
      "step": 1430
    },
    {
      "epoch": 1.322550831792976,
      "grad_norm": 2.1875,
      "learning_rate": 5.648377202649664e-06,
      "loss": 1.6924,
      "step": 1431
    },
    {
      "epoch": 1.3234750462107208,
      "grad_norm": 2.140625,
      "learning_rate": 5.6346181113588645e-06,
      "loss": 1.6083,
      "step": 1432
    },
    {
      "epoch": 1.3243992606284658,
      "grad_norm": 2.171875,
      "learning_rate": 5.6208692223718434e-06,
      "loss": 1.7109,
      "step": 1433
    },
    {
      "epoch": 1.3253234750462108,
      "grad_norm": 2.140625,
      "learning_rate": 5.60713056782103e-06,
      "loss": 1.7177,
      "step": 1434
    },
    {
      "epoch": 1.3262476894639557,
      "grad_norm": 2.15625,
      "learning_rate": 5.593402179814944e-06,
      "loss": 1.6718,
      "step": 1435
    },
    {
      "epoch": 1.3271719038817005,
      "grad_norm": 2.203125,
      "learning_rate": 5.579684090438099e-06,
      "loss": 1.7007,
      "step": 1436
    },
    {
      "epoch": 1.3280961182994455,
      "grad_norm": 2.203125,
      "learning_rate": 5.565976331750955e-06,
      "loss": 1.7268,
      "step": 1437
    },
    {
      "epoch": 1.3290203327171903,
      "grad_norm": 2.234375,
      "learning_rate": 5.552278935789816e-06,
      "loss": 1.6633,
      "step": 1438
    },
    {
      "epoch": 1.3299445471349354,
      "grad_norm": 2.171875,
      "learning_rate": 5.538591934566771e-06,
      "loss": 1.6503,
      "step": 1439
    },
    {
      "epoch": 1.3308687615526802,
      "grad_norm": 2.15625,
      "learning_rate": 5.524915360069617e-06,
      "loss": 1.7402,
      "step": 1440
    },
    {
      "epoch": 1.331792975970425,
      "grad_norm": 2.140625,
      "learning_rate": 5.511249244261781e-06,
      "loss": 1.6785,
      "step": 1441
    },
    {
      "epoch": 1.33271719038817,
      "grad_norm": 2.234375,
      "learning_rate": 5.4975936190822456e-06,
      "loss": 1.7743,
      "step": 1442
    },
    {
      "epoch": 1.333641404805915,
      "grad_norm": 2.203125,
      "learning_rate": 5.4839485164454796e-06,
      "loss": 1.7053,
      "step": 1443
    },
    {
      "epoch": 1.33456561922366,
      "grad_norm": 2.15625,
      "learning_rate": 5.4703139682413585e-06,
      "loss": 1.7412,
      "step": 1444
    },
    {
      "epoch": 1.3354898336414047,
      "grad_norm": 2.15625,
      "learning_rate": 5.456690006335091e-06,
      "loss": 1.6954,
      "step": 1445
    },
    {
      "epoch": 1.3364140480591498,
      "grad_norm": 2.171875,
      "learning_rate": 5.443076662567136e-06,
      "loss": 1.7228,
      "step": 1446
    },
    {
      "epoch": 1.3373382624768946,
      "grad_norm": 2.21875,
      "learning_rate": 5.429473968753157e-06,
      "loss": 1.7219,
      "step": 1447
    },
    {
      "epoch": 1.3382624768946396,
      "grad_norm": 2.125,
      "learning_rate": 5.415881956683911e-06,
      "loss": 1.6544,
      "step": 1448
    },
    {
      "epoch": 1.3391866913123844,
      "grad_norm": 2.140625,
      "learning_rate": 5.402300658125197e-06,
      "loss": 1.6401,
      "step": 1449
    },
    {
      "epoch": 1.3401109057301293,
      "grad_norm": 2.171875,
      "learning_rate": 5.3887301048177685e-06,
      "loss": 1.7249,
      "step": 1450
    },
    {
      "epoch": 1.3410351201478743,
      "grad_norm": 2.15625,
      "learning_rate": 5.375170328477271e-06,
      "loss": 1.6682,
      "step": 1451
    },
    {
      "epoch": 1.3419593345656193,
      "grad_norm": 2.171875,
      "learning_rate": 5.361621360794169e-06,
      "loss": 1.6601,
      "step": 1452
    },
    {
      "epoch": 1.3428835489833642,
      "grad_norm": 2.21875,
      "learning_rate": 5.3480832334336615e-06,
      "loss": 1.7277,
      "step": 1453
    },
    {
      "epoch": 1.343807763401109,
      "grad_norm": 2.203125,
      "learning_rate": 5.334555978035609e-06,
      "loss": 1.6536,
      "step": 1454
    },
    {
      "epoch": 1.344731977818854,
      "grad_norm": 2.125,
      "learning_rate": 5.321039626214468e-06,
      "loss": 1.6839,
      "step": 1455
    },
    {
      "epoch": 1.3456561922365988,
      "grad_norm": 2.171875,
      "learning_rate": 5.307534209559206e-06,
      "loss": 1.7224,
      "step": 1456
    },
    {
      "epoch": 1.3465804066543439,
      "grad_norm": 2.21875,
      "learning_rate": 5.2940397596332515e-06,
      "loss": 1.7435,
      "step": 1457
    },
    {
      "epoch": 1.3475046210720887,
      "grad_norm": 2.234375,
      "learning_rate": 5.280556307974376e-06,
      "loss": 1.7032,
      "step": 1458
    },
    {
      "epoch": 1.3484288354898337,
      "grad_norm": 2.25,
      "learning_rate": 5.267083886094668e-06,
      "loss": 1.8032,
      "step": 1459
    },
    {
      "epoch": 1.3493530499075785,
      "grad_norm": 2.203125,
      "learning_rate": 5.253622525480429e-06,
      "loss": 1.721,
      "step": 1460
    },
    {
      "epoch": 1.3502772643253236,
      "grad_norm": 2.140625,
      "learning_rate": 5.240172257592109e-06,
      "loss": 1.6883,
      "step": 1461
    },
    {
      "epoch": 1.3512014787430684,
      "grad_norm": 2.140625,
      "learning_rate": 5.226733113864242e-06,
      "loss": 1.6391,
      "step": 1462
    },
    {
      "epoch": 1.3521256931608132,
      "grad_norm": 2.296875,
      "learning_rate": 5.213305125705354e-06,
      "loss": 1.6023,
      "step": 1463
    },
    {
      "epoch": 1.3530499075785583,
      "grad_norm": 2.171875,
      "learning_rate": 5.199888324497907e-06,
      "loss": 1.7731,
      "step": 1464
    },
    {
      "epoch": 1.353974121996303,
      "grad_norm": 2.1875,
      "learning_rate": 5.18648274159821e-06,
      "loss": 1.6239,
      "step": 1465
    },
    {
      "epoch": 1.354898336414048,
      "grad_norm": 2.21875,
      "learning_rate": 5.17308840833636e-06,
      "loss": 1.6655,
      "step": 1466
    },
    {
      "epoch": 1.355822550831793,
      "grad_norm": 2.09375,
      "learning_rate": 5.159705356016158e-06,
      "loss": 1.6086,
      "step": 1467
    },
    {
      "epoch": 1.356746765249538,
      "grad_norm": 2.171875,
      "learning_rate": 5.146333615915047e-06,
      "loss": 1.7517,
      "step": 1468
    },
    {
      "epoch": 1.3576709796672828,
      "grad_norm": 2.171875,
      "learning_rate": 5.132973219284023e-06,
      "loss": 1.735,
      "step": 1469
    },
    {
      "epoch": 1.3585951940850278,
      "grad_norm": 2.140625,
      "learning_rate": 5.119624197347581e-06,
      "loss": 1.7187,
      "step": 1470
    },
    {
      "epoch": 1.3595194085027726,
      "grad_norm": 2.09375,
      "learning_rate": 5.106286581303623e-06,
      "loss": 1.6843,
      "step": 1471
    },
    {
      "epoch": 1.3604436229205175,
      "grad_norm": 2.078125,
      "learning_rate": 5.092960402323404e-06,
      "loss": 1.6426,
      "step": 1472
    },
    {
      "epoch": 1.3613678373382625,
      "grad_norm": 2.1875,
      "learning_rate": 5.079645691551445e-06,
      "loss": 1.6992,
      "step": 1473
    },
    {
      "epoch": 1.3622920517560073,
      "grad_norm": 2.203125,
      "learning_rate": 5.06634248010546e-06,
      "loss": 1.6783,
      "step": 1474
    },
    {
      "epoch": 1.3632162661737524,
      "grad_norm": 2.25,
      "learning_rate": 5.0530507990762955e-06,
      "loss": 1.6866,
      "step": 1475
    },
    {
      "epoch": 1.3641404805914972,
      "grad_norm": 2.140625,
      "learning_rate": 5.039770679527844e-06,
      "loss": 1.6912,
      "step": 1476
    },
    {
      "epoch": 1.3650646950092422,
      "grad_norm": 2.171875,
      "learning_rate": 5.026502152496986e-06,
      "loss": 1.7571,
      "step": 1477
    },
    {
      "epoch": 1.365988909426987,
      "grad_norm": 2.1875,
      "learning_rate": 5.013245248993499e-06,
      "loss": 1.6757,
      "step": 1478
    },
    {
      "epoch": 1.366913123844732,
      "grad_norm": 2.1875,
      "learning_rate": 5.000000000000003e-06,
      "loss": 1.6955,
      "step": 1479
    },
    {
      "epoch": 1.3678373382624769,
      "grad_norm": 2.296875,
      "learning_rate": 4.986766436471873e-06,
      "loss": 1.6643,
      "step": 1480
    },
    {
      "epoch": 1.3687615526802217,
      "grad_norm": 2.140625,
      "learning_rate": 4.973544589337186e-06,
      "loss": 1.6371,
      "step": 1481
    },
    {
      "epoch": 1.3696857670979667,
      "grad_norm": 2.203125,
      "learning_rate": 4.960334489496626e-06,
      "loss": 1.6576,
      "step": 1482
    },
    {
      "epoch": 1.3706099815157118,
      "grad_norm": 2.234375,
      "learning_rate": 4.947136167823428e-06,
      "loss": 1.7847,
      "step": 1483
    },
    {
      "epoch": 1.3715341959334566,
      "grad_norm": 2.171875,
      "learning_rate": 4.933949655163295e-06,
      "loss": 1.7052,
      "step": 1484
    },
    {
      "epoch": 1.3724584103512014,
      "grad_norm": 2.109375,
      "learning_rate": 4.920774982334335e-06,
      "loss": 1.6767,
      "step": 1485
    },
    {
      "epoch": 1.3733826247689465,
      "grad_norm": 2.1875,
      "learning_rate": 4.907612180126986e-06,
      "loss": 1.6474,
      "step": 1486
    },
    {
      "epoch": 1.3743068391866913,
      "grad_norm": 2.171875,
      "learning_rate": 4.89446127930394e-06,
      "loss": 1.768,
      "step": 1487
    },
    {
      "epoch": 1.3752310536044363,
      "grad_norm": 2.234375,
      "learning_rate": 4.881322310600079e-06,
      "loss": 1.6284,
      "step": 1488
    },
    {
      "epoch": 1.3761552680221811,
      "grad_norm": 2.140625,
      "learning_rate": 4.868195304722391e-06,
      "loss": 1.6341,
      "step": 1489
    },
    {
      "epoch": 1.377079482439926,
      "grad_norm": 2.15625,
      "learning_rate": 4.855080292349913e-06,
      "loss": 1.6644,
      "step": 1490
    },
    {
      "epoch": 1.378003696857671,
      "grad_norm": 2.140625,
      "learning_rate": 4.841977304133653e-06,
      "loss": 1.7202,
      "step": 1491
    },
    {
      "epoch": 1.378927911275416,
      "grad_norm": 2.265625,
      "learning_rate": 4.828886370696511e-06,
      "loss": 1.7045,
      "step": 1492
    },
    {
      "epoch": 1.3798521256931608,
      "grad_norm": 2.1875,
      "learning_rate": 4.81580752263322e-06,
      "loss": 1.6782,
      "step": 1493
    },
    {
      "epoch": 1.3807763401109057,
      "grad_norm": 2.21875,
      "learning_rate": 4.802740790510258e-06,
      "loss": 1.772,
      "step": 1494
    },
    {
      "epoch": 1.3817005545286507,
      "grad_norm": 2.328125,
      "learning_rate": 4.789686204865797e-06,
      "loss": 1.6978,
      "step": 1495
    },
    {
      "epoch": 1.3826247689463955,
      "grad_norm": 2.15625,
      "learning_rate": 4.776643796209621e-06,
      "loss": 1.6673,
      "step": 1496
    },
    {
      "epoch": 1.3835489833641406,
      "grad_norm": 2.15625,
      "learning_rate": 4.7636135950230544e-06,
      "loss": 1.6506,
      "step": 1497
    },
    {
      "epoch": 1.3844731977818854,
      "grad_norm": 2.171875,
      "learning_rate": 4.750595631758886e-06,
      "loss": 1.6678,
      "step": 1498
    },
    {
      "epoch": 1.3853974121996302,
      "grad_norm": 2.1875,
      "learning_rate": 4.7375899368413105e-06,
      "loss": 1.6911,
      "step": 1499
    },
    {
      "epoch": 1.3863216266173752,
      "grad_norm": 2.15625,
      "learning_rate": 4.724596540665843e-06,
      "loss": 1.6708,
      "step": 1500
    },
    {
      "epoch": 1.3872458410351203,
      "grad_norm": 2.125,
      "learning_rate": 4.7116154735992695e-06,
      "loss": 1.7159,
      "step": 1501
    },
    {
      "epoch": 1.388170055452865,
      "grad_norm": 2.15625,
      "learning_rate": 4.698646765979541e-06,
      "loss": 1.6999,
      "step": 1502
    },
    {
      "epoch": 1.38909426987061,
      "grad_norm": 2.171875,
      "learning_rate": 4.685690448115739e-06,
      "loss": 1.6688,
      "step": 1503
    },
    {
      "epoch": 1.390018484288355,
      "grad_norm": 2.109375,
      "learning_rate": 4.672746550287985e-06,
      "loss": 1.6871,
      "step": 1504
    },
    {
      "epoch": 1.3909426987060998,
      "grad_norm": 2.15625,
      "learning_rate": 4.6598151027473674e-06,
      "loss": 1.6373,
      "step": 1505
    },
    {
      "epoch": 1.3918669131238448,
      "grad_norm": 2.15625,
      "learning_rate": 4.6468961357158915e-06,
      "loss": 1.6941,
      "step": 1506
    },
    {
      "epoch": 1.3927911275415896,
      "grad_norm": 2.09375,
      "learning_rate": 4.63398967938638e-06,
      "loss": 1.5945,
      "step": 1507
    },
    {
      "epoch": 1.3937153419593344,
      "grad_norm": 2.1875,
      "learning_rate": 4.621095763922425e-06,
      "loss": 1.7059,
      "step": 1508
    },
    {
      "epoch": 1.3946395563770795,
      "grad_norm": 2.28125,
      "learning_rate": 4.608214419458306e-06,
      "loss": 1.6923,
      "step": 1509
    },
    {
      "epoch": 1.3955637707948245,
      "grad_norm": 2.140625,
      "learning_rate": 4.595345676098923e-06,
      "loss": 1.6638,
      "step": 1510
    },
    {
      "epoch": 1.3964879852125693,
      "grad_norm": 2.25,
      "learning_rate": 4.582489563919729e-06,
      "loss": 1.7736,
      "step": 1511
    },
    {
      "epoch": 1.3974121996303142,
      "grad_norm": 2.234375,
      "learning_rate": 4.5696461129666525e-06,
      "loss": 1.7227,
      "step": 1512
    },
    {
      "epoch": 1.3983364140480592,
      "grad_norm": 2.203125,
      "learning_rate": 4.5568153532560354e-06,
      "loss": 1.7075,
      "step": 1513
    },
    {
      "epoch": 1.399260628465804,
      "grad_norm": 2.21875,
      "learning_rate": 4.5439973147745534e-06,
      "loss": 1.664,
      "step": 1514
    },
    {
      "epoch": 1.400184842883549,
      "grad_norm": 2.1875,
      "learning_rate": 4.531192027479161e-06,
      "loss": 1.6294,
      "step": 1515
    },
    {
      "epoch": 1.4011090573012939,
      "grad_norm": 2.25,
      "learning_rate": 4.518399521297005e-06,
      "loss": 1.7111,
      "step": 1516
    },
    {
      "epoch": 1.4020332717190387,
      "grad_norm": 2.203125,
      "learning_rate": 4.505619826125364e-06,
      "loss": 1.6501,
      "step": 1517
    },
    {
      "epoch": 1.4029574861367837,
      "grad_norm": 2.203125,
      "learning_rate": 4.492852971831572e-06,
      "loss": 1.724,
      "step": 1518
    },
    {
      "epoch": 1.4038817005545288,
      "grad_norm": 2.140625,
      "learning_rate": 4.480098988252958e-06,
      "loss": 1.6937,
      "step": 1519
    },
    {
      "epoch": 1.4048059149722736,
      "grad_norm": 2.125,
      "learning_rate": 4.4673579051967665e-06,
      "loss": 1.6457,
      "step": 1520
    },
    {
      "epoch": 1.4057301293900184,
      "grad_norm": 2.109375,
      "learning_rate": 4.454629752440096e-06,
      "loss": 1.6571,
      "step": 1521
    },
    {
      "epoch": 1.4066543438077634,
      "grad_norm": 2.125,
      "learning_rate": 4.441914559729825e-06,
      "loss": 1.6457,
      "step": 1522
    },
    {
      "epoch": 1.4075785582255083,
      "grad_norm": 2.1875,
      "learning_rate": 4.429212356782541e-06,
      "loss": 1.627,
      "step": 1523
    },
    {
      "epoch": 1.4085027726432533,
      "grad_norm": 2.265625,
      "learning_rate": 4.41652317328447e-06,
      "loss": 1.7244,
      "step": 1524
    },
    {
      "epoch": 1.4094269870609981,
      "grad_norm": 2.140625,
      "learning_rate": 4.403847038891425e-06,
      "loss": 1.6988,
      "step": 1525
    },
    {
      "epoch": 1.410351201478743,
      "grad_norm": 2.046875,
      "learning_rate": 4.391183983228705e-06,
      "loss": 1.6255,
      "step": 1526
    },
    {
      "epoch": 1.411275415896488,
      "grad_norm": 2.1875,
      "learning_rate": 4.378534035891051e-06,
      "loss": 1.7229,
      "step": 1527
    },
    {
      "epoch": 1.412199630314233,
      "grad_norm": 2.125,
      "learning_rate": 4.365897226442573e-06,
      "loss": 1.6781,
      "step": 1528
    },
    {
      "epoch": 1.4131238447319778,
      "grad_norm": 2.125,
      "learning_rate": 4.353273584416657e-06,
      "loss": 1.6935,
      "step": 1529
    },
    {
      "epoch": 1.4140480591497226,
      "grad_norm": 2.078125,
      "learning_rate": 4.340663139315942e-06,
      "loss": 1.6428,
      "step": 1530
    },
    {
      "epoch": 1.4149722735674677,
      "grad_norm": 2.125,
      "learning_rate": 4.328065920612207e-06,
      "loss": 1.6272,
      "step": 1531
    },
    {
      "epoch": 1.4158964879852125,
      "grad_norm": 2.15625,
      "learning_rate": 4.315481957746325e-06,
      "loss": 1.6781,
      "step": 1532
    },
    {
      "epoch": 1.4168207024029575,
      "grad_norm": 2.109375,
      "learning_rate": 4.302911280128189e-06,
      "loss": 1.677,
      "step": 1533
    },
    {
      "epoch": 1.4177449168207024,
      "grad_norm": 2.15625,
      "learning_rate": 4.290353917136639e-06,
      "loss": 1.7141,
      "step": 1534
    },
    {
      "epoch": 1.4186691312384474,
      "grad_norm": 2.125,
      "learning_rate": 4.277809898119407e-06,
      "loss": 1.6917,
      "step": 1535
    },
    {
      "epoch": 1.4195933456561922,
      "grad_norm": 2.171875,
      "learning_rate": 4.265279252393035e-06,
      "loss": 1.6623,
      "step": 1536
    },
    {
      "epoch": 1.4205175600739373,
      "grad_norm": 2.21875,
      "learning_rate": 4.2527620092428e-06,
      "loss": 1.708,
      "step": 1537
    },
    {
      "epoch": 1.421441774491682,
      "grad_norm": 2.109375,
      "learning_rate": 4.240258197922668e-06,
      "loss": 1.6086,
      "step": 1538
    },
    {
      "epoch": 1.422365988909427,
      "grad_norm": 2.171875,
      "learning_rate": 4.227767847655205e-06,
      "loss": 1.7323,
      "step": 1539
    },
    {
      "epoch": 1.423290203327172,
      "grad_norm": 2.1875,
      "learning_rate": 4.215290987631531e-06,
      "loss": 1.7266,
      "step": 1540
    },
    {
      "epoch": 1.4242144177449167,
      "grad_norm": 2.171875,
      "learning_rate": 4.202827647011226e-06,
      "loss": 1.7032,
      "step": 1541
    },
    {
      "epoch": 1.4251386321626618,
      "grad_norm": 2.1875,
      "learning_rate": 4.190377854922275e-06,
      "loss": 1.7317,
      "step": 1542
    },
    {
      "epoch": 1.4260628465804066,
      "grad_norm": 2.1875,
      "learning_rate": 4.177941640461003e-06,
      "loss": 1.6222,
      "step": 1543
    },
    {
      "epoch": 1.4269870609981516,
      "grad_norm": 2.21875,
      "learning_rate": 4.165519032691998e-06,
      "loss": 1.7467,
      "step": 1544
    },
    {
      "epoch": 1.4279112754158965,
      "grad_norm": 2.171875,
      "learning_rate": 4.153110060648054e-06,
      "loss": 1.6809,
      "step": 1545
    },
    {
      "epoch": 1.4288354898336415,
      "grad_norm": 2.15625,
      "learning_rate": 4.140714753330091e-06,
      "loss": 1.6463,
      "step": 1546
    },
    {
      "epoch": 1.4297597042513863,
      "grad_norm": 2.125,
      "learning_rate": 4.1283331397070954e-06,
      "loss": 1.644,
      "step": 1547
    },
    {
      "epoch": 1.4306839186691311,
      "grad_norm": 2.140625,
      "learning_rate": 4.1159652487160505e-06,
      "loss": 1.6242,
      "step": 1548
    },
    {
      "epoch": 1.4316081330868762,
      "grad_norm": 2.171875,
      "learning_rate": 4.1036111092618725e-06,
      "loss": 1.6891,
      "step": 1549
    },
    {
      "epoch": 1.432532347504621,
      "grad_norm": 2.203125,
      "learning_rate": 4.0912707502173335e-06,
      "loss": 1.7121,
      "step": 1550
    },
    {
      "epoch": 1.433456561922366,
      "grad_norm": 2.125,
      "learning_rate": 4.078944200422999e-06,
      "loss": 1.6429,
      "step": 1551
    },
    {
      "epoch": 1.4343807763401109,
      "grad_norm": 2.234375,
      "learning_rate": 4.066631488687166e-06,
      "loss": 1.6839,
      "step": 1552
    },
    {
      "epoch": 1.435304990757856,
      "grad_norm": 2.1875,
      "learning_rate": 4.054332643785786e-06,
      "loss": 1.6877,
      "step": 1553
    },
    {
      "epoch": 1.4362292051756007,
      "grad_norm": 2.171875,
      "learning_rate": 4.042047694462405e-06,
      "loss": 1.6882,
      "step": 1554
    },
    {
      "epoch": 1.4371534195933457,
      "grad_norm": 2.15625,
      "learning_rate": 4.029776669428091e-06,
      "loss": 1.5973,
      "step": 1555
    },
    {
      "epoch": 1.4380776340110906,
      "grad_norm": 2.109375,
      "learning_rate": 4.017519597361375e-06,
      "loss": 1.7257,
      "step": 1556
    },
    {
      "epoch": 1.4390018484288354,
      "grad_norm": 2.1875,
      "learning_rate": 4.005276506908172e-06,
      "loss": 1.7546,
      "step": 1557
    },
    {
      "epoch": 1.4399260628465804,
      "grad_norm": 2.1875,
      "learning_rate": 3.993047426681723e-06,
      "loss": 1.702,
      "step": 1558
    },
    {
      "epoch": 1.4408502772643252,
      "grad_norm": 2.140625,
      "learning_rate": 3.980832385262532e-06,
      "loss": 1.6732,
      "step": 1559
    },
    {
      "epoch": 1.4417744916820703,
      "grad_norm": 2.21875,
      "learning_rate": 3.968631411198285e-06,
      "loss": 1.7638,
      "step": 1560
    },
    {
      "epoch": 1.442698706099815,
      "grad_norm": 2.21875,
      "learning_rate": 3.9564445330037934e-06,
      "loss": 1.7503,
      "step": 1561
    },
    {
      "epoch": 1.4436229205175601,
      "grad_norm": 2.171875,
      "learning_rate": 3.944271779160928e-06,
      "loss": 1.7149,
      "step": 1562
    },
    {
      "epoch": 1.444547134935305,
      "grad_norm": 2.171875,
      "learning_rate": 3.932113178118543e-06,
      "loss": 1.7417,
      "step": 1563
    },
    {
      "epoch": 1.44547134935305,
      "grad_norm": 2.15625,
      "learning_rate": 3.919968758292425e-06,
      "loss": 1.6055,
      "step": 1564
    },
    {
      "epoch": 1.4463955637707948,
      "grad_norm": 2.171875,
      "learning_rate": 3.907838548065211e-06,
      "loss": 1.7189,
      "step": 1565
    },
    {
      "epoch": 1.4473197781885396,
      "grad_norm": 2.140625,
      "learning_rate": 3.895722575786333e-06,
      "loss": 1.7194,
      "step": 1566
    },
    {
      "epoch": 1.4482439926062847,
      "grad_norm": 2.09375,
      "learning_rate": 3.883620869771943e-06,
      "loss": 1.6323,
      "step": 1567
    },
    {
      "epoch": 1.4491682070240297,
      "grad_norm": 2.203125,
      "learning_rate": 3.871533458304855e-06,
      "loss": 1.6635,
      "step": 1568
    },
    {
      "epoch": 1.4500924214417745,
      "grad_norm": 2.140625,
      "learning_rate": 3.859460369634479e-06,
      "loss": 1.6471,
      "step": 1569
    },
    {
      "epoch": 1.4510166358595193,
      "grad_norm": 2.109375,
      "learning_rate": 3.847401631976744e-06,
      "loss": 1.6401,
      "step": 1570
    },
    {
      "epoch": 1.4519408502772644,
      "grad_norm": 2.1875,
      "learning_rate": 3.835357273514044e-06,
      "loss": 1.6384,
      "step": 1571
    },
    {
      "epoch": 1.4528650646950092,
      "grad_norm": 2.125,
      "learning_rate": 3.823327322395161e-06,
      "loss": 1.666,
      "step": 1572
    },
    {
      "epoch": 1.4537892791127542,
      "grad_norm": 2.140625,
      "learning_rate": 3.8113118067352116e-06,
      "loss": 1.6375,
      "step": 1573
    },
    {
      "epoch": 1.454713493530499,
      "grad_norm": 2.171875,
      "learning_rate": 3.7993107546155783e-06,
      "loss": 1.7317,
      "step": 1574
    },
    {
      "epoch": 1.4556377079482439,
      "grad_norm": 2.140625,
      "learning_rate": 3.7873241940838347e-06,
      "loss": 1.7396,
      "step": 1575
    },
    {
      "epoch": 1.456561922365989,
      "grad_norm": 2.15625,
      "learning_rate": 3.775352153153687e-06,
      "loss": 1.7018,
      "step": 1576
    },
    {
      "epoch": 1.457486136783734,
      "grad_norm": 2.203125,
      "learning_rate": 3.763394659804912e-06,
      "loss": 1.6507,
      "step": 1577
    },
    {
      "epoch": 1.4584103512014788,
      "grad_norm": 2.171875,
      "learning_rate": 3.7514517419832784e-06,
      "loss": 1.6875,
      "step": 1578
    },
    {
      "epoch": 1.4593345656192236,
      "grad_norm": 2.171875,
      "learning_rate": 3.739523427600509e-06,
      "loss": 1.7956,
      "step": 1579
    },
    {
      "epoch": 1.4602587800369686,
      "grad_norm": 2.15625,
      "learning_rate": 3.727609744534173e-06,
      "loss": 1.7088,
      "step": 1580
    },
    {
      "epoch": 1.4611829944547134,
      "grad_norm": 2.109375,
      "learning_rate": 3.715710720627661e-06,
      "loss": 1.6138,
      "step": 1581
    },
    {
      "epoch": 1.4621072088724585,
      "grad_norm": 2.109375,
      "learning_rate": 3.703826383690099e-06,
      "loss": 1.637,
      "step": 1582
    },
    {
      "epoch": 1.4630314232902033,
      "grad_norm": 2.15625,
      "learning_rate": 3.691956761496285e-06,
      "loss": 1.5822,
      "step": 1583
    },
    {
      "epoch": 1.4639556377079481,
      "grad_norm": 2.140625,
      "learning_rate": 3.6801018817866375e-06,
      "loss": 1.7177,
      "step": 1584
    },
    {
      "epoch": 1.4648798521256932,
      "grad_norm": 2.15625,
      "learning_rate": 3.66826177226711e-06,
      "loss": 1.5937,
      "step": 1585
    },
    {
      "epoch": 1.4658040665434382,
      "grad_norm": 2.1875,
      "learning_rate": 3.656436460609142e-06,
      "loss": 1.6385,
      "step": 1586
    },
    {
      "epoch": 1.466728280961183,
      "grad_norm": 2.21875,
      "learning_rate": 3.644625974449586e-06,
      "loss": 1.7138,
      "step": 1587
    },
    {
      "epoch": 1.4676524953789278,
      "grad_norm": 2.234375,
      "learning_rate": 3.6328303413906474e-06,
      "loss": 1.714,
      "step": 1588
    },
    {
      "epoch": 1.4685767097966729,
      "grad_norm": 2.1875,
      "learning_rate": 3.62104958899982e-06,
      "loss": 1.7448,
      "step": 1589
    },
    {
      "epoch": 1.4695009242144177,
      "grad_norm": 2.203125,
      "learning_rate": 3.609283744809817e-06,
      "loss": 1.637,
      "step": 1590
    },
    {
      "epoch": 1.4704251386321627,
      "grad_norm": 2.15625,
      "learning_rate": 3.5975328363185123e-06,
      "loss": 1.6341,
      "step": 1591
    },
    {
      "epoch": 1.4713493530499075,
      "grad_norm": 2.125,
      "learning_rate": 3.5857968909888694e-06,
      "loss": 1.6803,
      "step": 1592
    },
    {
      "epoch": 1.4722735674676524,
      "grad_norm": 2.1875,
      "learning_rate": 3.574075936248892e-06,
      "loss": 1.7021,
      "step": 1593
    },
    {
      "epoch": 1.4731977818853974,
      "grad_norm": 2.1875,
      "learning_rate": 3.5623699994915363e-06,
      "loss": 1.6389,
      "step": 1594
    },
    {
      "epoch": 1.4741219963031424,
      "grad_norm": 2.234375,
      "learning_rate": 3.5506791080746695e-06,
      "loss": 1.7286,
      "step": 1595
    },
    {
      "epoch": 1.4750462107208873,
      "grad_norm": 2.171875,
      "learning_rate": 3.5390032893209902e-06,
      "loss": 1.6244,
      "step": 1596
    },
    {
      "epoch": 1.475970425138632,
      "grad_norm": 2.140625,
      "learning_rate": 3.527342570517975e-06,
      "loss": 1.6732,
      "step": 1597
    },
    {
      "epoch": 1.4768946395563771,
      "grad_norm": 2.234375,
      "learning_rate": 3.5156969789178097e-06,
      "loss": 1.7415,
      "step": 1598
    },
    {
      "epoch": 1.477818853974122,
      "grad_norm": 2.171875,
      "learning_rate": 3.5040665417373232e-06,
      "loss": 1.668,
      "step": 1599
    },
    {
      "epoch": 1.478743068391867,
      "grad_norm": 2.1875,
      "learning_rate": 3.492451286157932e-06,
      "loss": 1.766,
      "step": 1600
    },
    {
      "epoch": 1.4796672828096118,
      "grad_norm": 2.1875,
      "learning_rate": 3.480851239325569e-06,
      "loss": 1.6679,
      "step": 1601
    },
    {
      "epoch": 1.4805914972273566,
      "grad_norm": 2.1875,
      "learning_rate": 3.4692664283506193e-06,
      "loss": 1.7481,
      "step": 1602
    },
    {
      "epoch": 1.4815157116451017,
      "grad_norm": 2.140625,
      "learning_rate": 3.4576968803078724e-06,
      "loss": 1.634,
      "step": 1603
    },
    {
      "epoch": 1.4824399260628467,
      "grad_norm": 2.3125,
      "learning_rate": 3.4461426222364337e-06,
      "loss": 1.7525,
      "step": 1604
    },
    {
      "epoch": 1.4833641404805915,
      "grad_norm": 2.125,
      "learning_rate": 3.434603681139681e-06,
      "loss": 1.684,
      "step": 1605
    },
    {
      "epoch": 1.4842883548983363,
      "grad_norm": 2.125,
      "learning_rate": 3.4230800839851965e-06,
      "loss": 1.6552,
      "step": 1606
    },
    {
      "epoch": 1.4852125693160814,
      "grad_norm": 2.203125,
      "learning_rate": 3.41157185770469e-06,
      "loss": 1.7101,
      "step": 1607
    },
    {
      "epoch": 1.4861367837338262,
      "grad_norm": 2.265625,
      "learning_rate": 3.400079029193966e-06,
      "loss": 1.6776,
      "step": 1608
    },
    {
      "epoch": 1.4870609981515712,
      "grad_norm": 2.234375,
      "learning_rate": 3.388601625312833e-06,
      "loss": 1.6726,
      "step": 1609
    },
    {
      "epoch": 1.487985212569316,
      "grad_norm": 2.140625,
      "learning_rate": 3.3771396728850505e-06,
      "loss": 1.7325,
      "step": 1610
    },
    {
      "epoch": 1.488909426987061,
      "grad_norm": 2.125,
      "learning_rate": 3.3656931986982676e-06,
      "loss": 1.6178,
      "step": 1611
    },
    {
      "epoch": 1.489833641404806,
      "grad_norm": 2.15625,
      "learning_rate": 3.3542622295039595e-06,
      "loss": 1.7695,
      "step": 1612
    },
    {
      "epoch": 1.490757855822551,
      "grad_norm": 2.1875,
      "learning_rate": 3.342846792017369e-06,
      "loss": 1.7747,
      "step": 1613
    },
    {
      "epoch": 1.4916820702402958,
      "grad_norm": 2.203125,
      "learning_rate": 3.3314469129174365e-06,
      "loss": 1.7212,
      "step": 1614
    },
    {
      "epoch": 1.4926062846580406,
      "grad_norm": 2.15625,
      "learning_rate": 3.3200626188467343e-06,
      "loss": 1.6473,
      "step": 1615
    },
    {
      "epoch": 1.4935304990757856,
      "grad_norm": 2.203125,
      "learning_rate": 3.308693936411421e-06,
      "loss": 1.6612,
      "step": 1616
    },
    {
      "epoch": 1.4944547134935304,
      "grad_norm": 2.15625,
      "learning_rate": 3.297340892181161e-06,
      "loss": 1.6447,
      "step": 1617
    },
    {
      "epoch": 1.4953789279112755,
      "grad_norm": 2.3125,
      "learning_rate": 3.2860035126890833e-06,
      "loss": 1.6532,
      "step": 1618
    },
    {
      "epoch": 1.4963031423290203,
      "grad_norm": 2.125,
      "learning_rate": 3.2746818244316957e-06,
      "loss": 1.742,
      "step": 1619
    },
    {
      "epoch": 1.4972273567467653,
      "grad_norm": 2.203125,
      "learning_rate": 3.2633758538688355e-06,
      "loss": 1.7345,
      "step": 1620
    },
    {
      "epoch": 1.4981515711645101,
      "grad_norm": 2.234375,
      "learning_rate": 3.2520856274236103e-06,
      "loss": 1.7484,
      "step": 1621
    },
    {
      "epoch": 1.4990757855822552,
      "grad_norm": 2.078125,
      "learning_rate": 3.240811171482329e-06,
      "loss": 1.6099,
      "step": 1622
    },
    {
      "epoch": 1.5,
      "grad_norm": 2.203125,
      "learning_rate": 3.229552512394445e-06,
      "loss": 1.6716,
      "step": 1623
    },
    {
      "epoch": 1.5009242144177448,
      "grad_norm": 2.1875,
      "learning_rate": 3.218309676472492e-06,
      "loss": 1.7036,
      "step": 1624
    },
    {
      "epoch": 1.5018484288354899,
      "grad_norm": 2.171875,
      "learning_rate": 3.207082689992025e-06,
      "loss": 1.787,
      "step": 1625
    },
    {
      "epoch": 1.502772643253235,
      "grad_norm": 2.265625,
      "learning_rate": 3.195871579191552e-06,
      "loss": 1.7575,
      "step": 1626
    },
    {
      "epoch": 1.5036968576709797,
      "grad_norm": 2.171875,
      "learning_rate": 3.184676370272488e-06,
      "loss": 1.7195,
      "step": 1627
    },
    {
      "epoch": 1.5046210720887245,
      "grad_norm": 2.140625,
      "learning_rate": 3.173497089399078e-06,
      "loss": 1.6199,
      "step": 1628
    },
    {
      "epoch": 1.5055452865064693,
      "grad_norm": 2.171875,
      "learning_rate": 3.1623337626983398e-06,
      "loss": 1.7029,
      "step": 1629
    },
    {
      "epoch": 1.5064695009242144,
      "grad_norm": 2.140625,
      "learning_rate": 3.151186416260006e-06,
      "loss": 1.5596,
      "step": 1630
    },
    {
      "epoch": 1.5073937153419594,
      "grad_norm": 2.125,
      "learning_rate": 3.140055076136466e-06,
      "loss": 1.646,
      "step": 1631
    },
    {
      "epoch": 1.5083179297597042,
      "grad_norm": 2.203125,
      "learning_rate": 3.1289397683426935e-06,
      "loss": 1.6982,
      "step": 1632
    },
    {
      "epoch": 1.509242144177449,
      "grad_norm": 2.109375,
      "learning_rate": 3.117840518856201e-06,
      "loss": 1.6321,
      "step": 1633
    },
    {
      "epoch": 1.510166358595194,
      "grad_norm": 2.1875,
      "learning_rate": 3.106757353616966e-06,
      "loss": 1.7068,
      "step": 1634
    },
    {
      "epoch": 1.5110905730129391,
      "grad_norm": 2.171875,
      "learning_rate": 3.095690298527377e-06,
      "loss": 1.6792,
      "step": 1635
    },
    {
      "epoch": 1.512014787430684,
      "grad_norm": 2.09375,
      "learning_rate": 3.084639379452169e-06,
      "loss": 1.5915,
      "step": 1636
    },
    {
      "epoch": 1.5129390018484288,
      "grad_norm": 2.234375,
      "learning_rate": 3.073604622218376e-06,
      "loss": 1.7646,
      "step": 1637
    },
    {
      "epoch": 1.5138632162661736,
      "grad_norm": 2.171875,
      "learning_rate": 3.062586052615246e-06,
      "loss": 1.6813,
      "step": 1638
    },
    {
      "epoch": 1.5147874306839186,
      "grad_norm": 2.171875,
      "learning_rate": 3.0515836963942056e-06,
      "loss": 1.6172,
      "step": 1639
    },
    {
      "epoch": 1.5157116451016637,
      "grad_norm": 2.1875,
      "learning_rate": 3.040597579268784e-06,
      "loss": 1.7103,
      "step": 1640
    },
    {
      "epoch": 1.5166358595194085,
      "grad_norm": 2.171875,
      "learning_rate": 3.029627726914559e-06,
      "loss": 1.7618,
      "step": 1641
    },
    {
      "epoch": 1.5175600739371533,
      "grad_norm": 2.203125,
      "learning_rate": 3.0186741649690966e-06,
      "loss": 1.781,
      "step": 1642
    },
    {
      "epoch": 1.5184842883548983,
      "grad_norm": 2.09375,
      "learning_rate": 3.0077369190318906e-06,
      "loss": 1.5187,
      "step": 1643
    },
    {
      "epoch": 1.5194085027726434,
      "grad_norm": 2.109375,
      "learning_rate": 2.9968160146643022e-06,
      "loss": 1.6056,
      "step": 1644
    },
    {
      "epoch": 1.5203327171903882,
      "grad_norm": 2.1875,
      "learning_rate": 2.9859114773895027e-06,
      "loss": 1.684,
      "step": 1645
    },
    {
      "epoch": 1.521256931608133,
      "grad_norm": 2.140625,
      "learning_rate": 2.9750233326924073e-06,
      "loss": 1.668,
      "step": 1646
    },
    {
      "epoch": 1.5221811460258778,
      "grad_norm": 2.140625,
      "learning_rate": 2.9641516060196286e-06,
      "loss": 1.6224,
      "step": 1647
    },
    {
      "epoch": 1.5231053604436229,
      "grad_norm": 2.15625,
      "learning_rate": 2.953296322779401e-06,
      "loss": 1.7032,
      "step": 1648
    },
    {
      "epoch": 1.524029574861368,
      "grad_norm": 2.140625,
      "learning_rate": 2.942457508341536e-06,
      "loss": 1.6487,
      "step": 1649
    },
    {
      "epoch": 1.5249537892791127,
      "grad_norm": 2.109375,
      "learning_rate": 2.931635188037345e-06,
      "loss": 1.6681,
      "step": 1650
    },
    {
      "epoch": 1.5258780036968576,
      "grad_norm": 2.078125,
      "learning_rate": 2.920829387159596e-06,
      "loss": 1.6449,
      "step": 1651
    },
    {
      "epoch": 1.5268022181146026,
      "grad_norm": 2.125,
      "learning_rate": 2.9100401309624604e-06,
      "loss": 1.6665,
      "step": 1652
    },
    {
      "epoch": 1.5277264325323476,
      "grad_norm": 2.203125,
      "learning_rate": 2.89926744466143e-06,
      "loss": 1.7807,
      "step": 1653
    },
    {
      "epoch": 1.5286506469500925,
      "grad_norm": 2.09375,
      "learning_rate": 2.8885113534332742e-06,
      "loss": 1.6116,
      "step": 1654
    },
    {
      "epoch": 1.5295748613678373,
      "grad_norm": 2.09375,
      "learning_rate": 2.8777718824159818e-06,
      "loss": 1.607,
      "step": 1655
    },
    {
      "epoch": 1.5304990757855823,
      "grad_norm": 2.15625,
      "learning_rate": 2.8670490567086895e-06,
      "loss": 1.7245,
      "step": 1656
    },
    {
      "epoch": 1.5314232902033271,
      "grad_norm": 2.15625,
      "learning_rate": 2.8563429013716515e-06,
      "loss": 1.667,
      "step": 1657
    },
    {
      "epoch": 1.5323475046210722,
      "grad_norm": 2.125,
      "learning_rate": 2.845653441426138e-06,
      "loss": 1.6678,
      "step": 1658
    },
    {
      "epoch": 1.533271719038817,
      "grad_norm": 2.1875,
      "learning_rate": 2.834980701854417e-06,
      "loss": 1.7711,
      "step": 1659
    },
    {
      "epoch": 1.5341959334565618,
      "grad_norm": 2.15625,
      "learning_rate": 2.8243247075996694e-06,
      "loss": 1.6594,
      "step": 1660
    },
    {
      "epoch": 1.5351201478743068,
      "grad_norm": 2.1875,
      "learning_rate": 2.8136854835659512e-06,
      "loss": 1.738,
      "step": 1661
    },
    {
      "epoch": 1.5360443622920519,
      "grad_norm": 2.09375,
      "learning_rate": 2.803063054618118e-06,
      "loss": 1.6341,
      "step": 1662
    },
    {
      "epoch": 1.5369685767097967,
      "grad_norm": 2.15625,
      "learning_rate": 2.792457445581773e-06,
      "loss": 1.6879,
      "step": 1663
    },
    {
      "epoch": 1.5378927911275415,
      "grad_norm": 2.234375,
      "learning_rate": 2.781868681243214e-06,
      "loss": 1.715,
      "step": 1664
    },
    {
      "epoch": 1.5388170055452866,
      "grad_norm": 2.140625,
      "learning_rate": 2.771296786349367e-06,
      "loss": 1.6355,
      "step": 1665
    },
    {
      "epoch": 1.5397412199630314,
      "grad_norm": 2.15625,
      "learning_rate": 2.7607417856077334e-06,
      "loss": 1.6677,
      "step": 1666
    },
    {
      "epoch": 1.5406654343807764,
      "grad_norm": 2.1875,
      "learning_rate": 2.750203703686335e-06,
      "loss": 1.7833,
      "step": 1667
    },
    {
      "epoch": 1.5415896487985212,
      "grad_norm": 2.125,
      "learning_rate": 2.7396825652136484e-06,
      "loss": 1.6678,
      "step": 1668
    },
    {
      "epoch": 1.542513863216266,
      "grad_norm": 2.125,
      "learning_rate": 2.7291783947785544e-06,
      "loss": 1.7076,
      "step": 1669
    },
    {
      "epoch": 1.543438077634011,
      "grad_norm": 2.203125,
      "learning_rate": 2.718691216930273e-06,
      "loss": 1.5984,
      "step": 1670
    },
    {
      "epoch": 1.5443622920517561,
      "grad_norm": 2.203125,
      "learning_rate": 2.7082210561783207e-06,
      "loss": 1.734,
      "step": 1671
    },
    {
      "epoch": 1.545286506469501,
      "grad_norm": 2.125,
      "learning_rate": 2.697767936992436e-06,
      "loss": 1.6952,
      "step": 1672
    },
    {
      "epoch": 1.5462107208872458,
      "grad_norm": 2.140625,
      "learning_rate": 2.687331883802531e-06,
      "loss": 1.6976,
      "step": 1673
    },
    {
      "epoch": 1.5471349353049908,
      "grad_norm": 2.15625,
      "learning_rate": 2.6769129209986324e-06,
      "loss": 1.6475,
      "step": 1674
    },
    {
      "epoch": 1.5480591497227358,
      "grad_norm": 2.15625,
      "learning_rate": 2.6665110729308263e-06,
      "loss": 1.6831,
      "step": 1675
    },
    {
      "epoch": 1.5489833641404807,
      "grad_norm": 2.140625,
      "learning_rate": 2.6561263639091995e-06,
      "loss": 1.6075,
      "step": 1676
    },
    {
      "epoch": 1.5499075785582255,
      "grad_norm": 2.09375,
      "learning_rate": 2.645758818203785e-06,
      "loss": 1.5824,
      "step": 1677
    },
    {
      "epoch": 1.5508317929759703,
      "grad_norm": 2.140625,
      "learning_rate": 2.635408460044502e-06,
      "loss": 1.5836,
      "step": 1678
    },
    {
      "epoch": 1.5517560073937153,
      "grad_norm": 2.109375,
      "learning_rate": 2.625075313621098e-06,
      "loss": 1.6713,
      "step": 1679
    },
    {
      "epoch": 1.5526802218114604,
      "grad_norm": 2.1875,
      "learning_rate": 2.614759403083098e-06,
      "loss": 1.7297,
      "step": 1680
    },
    {
      "epoch": 1.5536044362292052,
      "grad_norm": 2.109375,
      "learning_rate": 2.6044607525397503e-06,
      "loss": 1.5955,
      "step": 1681
    },
    {
      "epoch": 1.55452865064695,
      "grad_norm": 2.140625,
      "learning_rate": 2.5941793860599564e-06,
      "loss": 1.7423,
      "step": 1682
    },
    {
      "epoch": 1.555452865064695,
      "grad_norm": 2.1875,
      "learning_rate": 2.5839153276722263e-06,
      "loss": 1.6421,
      "step": 1683
    },
    {
      "epoch": 1.55637707948244,
      "grad_norm": 2.1875,
      "learning_rate": 2.5736686013646226e-06,
      "loss": 1.6339,
      "step": 1684
    },
    {
      "epoch": 1.557301293900185,
      "grad_norm": 2.125,
      "learning_rate": 2.563439231084691e-06,
      "loss": 1.6867,
      "step": 1685
    },
    {
      "epoch": 1.5582255083179297,
      "grad_norm": 2.390625,
      "learning_rate": 2.55322724073943e-06,
      "loss": 1.7123,
      "step": 1686
    },
    {
      "epoch": 1.5591497227356745,
      "grad_norm": 2.21875,
      "learning_rate": 2.543032654195209e-06,
      "loss": 1.7549,
      "step": 1687
    },
    {
      "epoch": 1.5600739371534196,
      "grad_norm": 2.203125,
      "learning_rate": 2.532855495277724e-06,
      "loss": 1.6939,
      "step": 1688
    },
    {
      "epoch": 1.5609981515711646,
      "grad_norm": 2.15625,
      "learning_rate": 2.522695787771944e-06,
      "loss": 1.7357,
      "step": 1689
    },
    {
      "epoch": 1.5619223659889094,
      "grad_norm": 2.125,
      "learning_rate": 2.512553555422048e-06,
      "loss": 1.7114,
      "step": 1690
    },
    {
      "epoch": 1.5628465804066543,
      "grad_norm": 2.171875,
      "learning_rate": 2.502428821931382e-06,
      "loss": 1.7505,
      "step": 1691
    },
    {
      "epoch": 1.5637707948243993,
      "grad_norm": 2.15625,
      "learning_rate": 2.4923216109623917e-06,
      "loss": 1.6676,
      "step": 1692
    },
    {
      "epoch": 1.5646950092421443,
      "grad_norm": 2.15625,
      "learning_rate": 2.482231946136563e-06,
      "loss": 1.7044,
      "step": 1693
    },
    {
      "epoch": 1.5656192236598891,
      "grad_norm": 2.203125,
      "learning_rate": 2.4721598510343857e-06,
      "loss": 1.6285,
      "step": 1694
    },
    {
      "epoch": 1.566543438077634,
      "grad_norm": 2.203125,
      "learning_rate": 2.4621053491952796e-06,
      "loss": 1.7315,
      "step": 1695
    },
    {
      "epoch": 1.5674676524953788,
      "grad_norm": 2.09375,
      "learning_rate": 2.4520684641175573e-06,
      "loss": 1.6311,
      "step": 1696
    },
    {
      "epoch": 1.5683918669131238,
      "grad_norm": 2.171875,
      "learning_rate": 2.442049219258351e-06,
      "loss": 1.6828,
      "step": 1697
    },
    {
      "epoch": 1.5693160813308689,
      "grad_norm": 2.171875,
      "learning_rate": 2.432047638033568e-06,
      "loss": 1.7121,
      "step": 1698
    },
    {
      "epoch": 1.5702402957486137,
      "grad_norm": 2.203125,
      "learning_rate": 2.422063743817832e-06,
      "loss": 1.6962,
      "step": 1699
    },
    {
      "epoch": 1.5711645101663585,
      "grad_norm": 2.125,
      "learning_rate": 2.412097559944442e-06,
      "loss": 1.7056,
      "step": 1700
    },
    {
      "epoch": 1.5720887245841035,
      "grad_norm": 2.140625,
      "learning_rate": 2.4021491097052862e-06,
      "loss": 1.673,
      "step": 1701
    },
    {
      "epoch": 1.5730129390018486,
      "grad_norm": 2.1875,
      "learning_rate": 2.3922184163508254e-06,
      "loss": 1.5784,
      "step": 1702
    },
    {
      "epoch": 1.5739371534195934,
      "grad_norm": 2.125,
      "learning_rate": 2.382305503090011e-06,
      "loss": 1.5782,
      "step": 1703
    },
    {
      "epoch": 1.5748613678373382,
      "grad_norm": 2.171875,
      "learning_rate": 2.372410393090243e-06,
      "loss": 1.6696,
      "step": 1704
    },
    {
      "epoch": 1.575785582255083,
      "grad_norm": 2.1875,
      "learning_rate": 2.362533109477321e-06,
      "loss": 1.7108,
      "step": 1705
    },
    {
      "epoch": 1.576709796672828,
      "grad_norm": 2.203125,
      "learning_rate": 2.3526736753353694e-06,
      "loss": 1.7086,
      "step": 1706
    },
    {
      "epoch": 1.577634011090573,
      "grad_norm": 2.15625,
      "learning_rate": 2.3428321137068077e-06,
      "loss": 1.6954,
      "step": 1707
    },
    {
      "epoch": 1.578558225508318,
      "grad_norm": 2.203125,
      "learning_rate": 2.33300844759228e-06,
      "loss": 1.701,
      "step": 1708
    },
    {
      "epoch": 1.5794824399260627,
      "grad_norm": 2.125,
      "learning_rate": 2.3232026999506062e-06,
      "loss": 1.6875,
      "step": 1709
    },
    {
      "epoch": 1.5804066543438078,
      "grad_norm": 2.15625,
      "learning_rate": 2.3134148936987343e-06,
      "loss": 1.769,
      "step": 1710
    },
    {
      "epoch": 1.5813308687615528,
      "grad_norm": 2.171875,
      "learning_rate": 2.303645051711677e-06,
      "loss": 1.6297,
      "step": 1711
    },
    {
      "epoch": 1.5822550831792976,
      "grad_norm": 2.15625,
      "learning_rate": 2.2938931968224644e-06,
      "loss": 1.6628,
      "step": 1712
    },
    {
      "epoch": 1.5831792975970425,
      "grad_norm": 2.1875,
      "learning_rate": 2.2841593518220884e-06,
      "loss": 1.7038,
      "step": 1713
    },
    {
      "epoch": 1.5841035120147873,
      "grad_norm": 2.1875,
      "learning_rate": 2.27444353945945e-06,
      "loss": 1.7573,
      "step": 1714
    },
    {
      "epoch": 1.5850277264325323,
      "grad_norm": 2.265625,
      "learning_rate": 2.26474578244131e-06,
      "loss": 1.6885,
      "step": 1715
    },
    {
      "epoch": 1.5859519408502774,
      "grad_norm": 2.140625,
      "learning_rate": 2.2550661034322284e-06,
      "loss": 1.6782,
      "step": 1716
    },
    {
      "epoch": 1.5868761552680222,
      "grad_norm": 2.140625,
      "learning_rate": 2.2454045250545154e-06,
      "loss": 1.7038,
      "step": 1717
    },
    {
      "epoch": 1.587800369685767,
      "grad_norm": 2.125,
      "learning_rate": 2.2357610698881784e-06,
      "loss": 1.614,
      "step": 1718
    },
    {
      "epoch": 1.588724584103512,
      "grad_norm": 2.09375,
      "learning_rate": 2.22613576047087e-06,
      "loss": 1.6649,
      "step": 1719
    },
    {
      "epoch": 1.589648798521257,
      "grad_norm": 2.140625,
      "learning_rate": 2.216528619297834e-06,
      "loss": 1.6781,
      "step": 1720
    },
    {
      "epoch": 1.5905730129390019,
      "grad_norm": 2.171875,
      "learning_rate": 2.2069396688218535e-06,
      "loss": 1.7225,
      "step": 1721
    },
    {
      "epoch": 1.5914972273567467,
      "grad_norm": 2.1875,
      "learning_rate": 2.1973689314531986e-06,
      "loss": 1.7373,
      "step": 1722
    },
    {
      "epoch": 1.5924214417744915,
      "grad_norm": 2.15625,
      "learning_rate": 2.187816429559574e-06,
      "loss": 1.6918,
      "step": 1723
    },
    {
      "epoch": 1.5933456561922366,
      "grad_norm": 2.125,
      "learning_rate": 2.178282185466061e-06,
      "loss": 1.6775,
      "step": 1724
    },
    {
      "epoch": 1.5942698706099816,
      "grad_norm": 2.046875,
      "learning_rate": 2.168766221455082e-06,
      "loss": 1.5588,
      "step": 1725
    },
    {
      "epoch": 1.5951940850277264,
      "grad_norm": 2.140625,
      "learning_rate": 2.1592685597663287e-06,
      "loss": 1.7085,
      "step": 1726
    },
    {
      "epoch": 1.5961182994454712,
      "grad_norm": 2.15625,
      "learning_rate": 2.1497892225967222e-06,
      "loss": 1.65,
      "step": 1727
    },
    {
      "epoch": 1.5970425138632163,
      "grad_norm": 2.3125,
      "learning_rate": 2.1403282321003493e-06,
      "loss": 1.7147,
      "step": 1728
    },
    {
      "epoch": 1.5979667282809613,
      "grad_norm": 2.09375,
      "learning_rate": 2.130885610388428e-06,
      "loss": 1.6512,
      "step": 1729
    },
    {
      "epoch": 1.5988909426987061,
      "grad_norm": 2.203125,
      "learning_rate": 2.121461379529247e-06,
      "loss": 1.7235,
      "step": 1730
    },
    {
      "epoch": 1.599815157116451,
      "grad_norm": 2.140625,
      "learning_rate": 2.1120555615481096e-06,
      "loss": 1.6793,
      "step": 1731
    },
    {
      "epoch": 1.600739371534196,
      "grad_norm": 2.109375,
      "learning_rate": 2.1026681784272872e-06,
      "loss": 1.6066,
      "step": 1732
    },
    {
      "epoch": 1.6016635859519408,
      "grad_norm": 2.203125,
      "learning_rate": 2.093299252105969e-06,
      "loss": 1.7119,
      "step": 1733
    },
    {
      "epoch": 1.6025878003696858,
      "grad_norm": 2.15625,
      "learning_rate": 2.083948804480204e-06,
      "loss": 1.7392,
      "step": 1734
    },
    {
      "epoch": 1.6035120147874307,
      "grad_norm": 2.171875,
      "learning_rate": 2.0746168574028667e-06,
      "loss": 1.6103,
      "step": 1735
    },
    {
      "epoch": 1.6044362292051755,
      "grad_norm": 2.15625,
      "learning_rate": 2.0653034326835786e-06,
      "loss": 1.6784,
      "step": 1736
    },
    {
      "epoch": 1.6053604436229205,
      "grad_norm": 2.140625,
      "learning_rate": 2.056008552088682e-06,
      "loss": 1.6799,
      "step": 1737
    },
    {
      "epoch": 1.6062846580406656,
      "grad_norm": 2.140625,
      "learning_rate": 2.0467322373411757e-06,
      "loss": 1.7153,
      "step": 1738
    },
    {
      "epoch": 1.6072088724584104,
      "grad_norm": 2.140625,
      "learning_rate": 2.037474510120676e-06,
      "loss": 1.6437,
      "step": 1739
    },
    {
      "epoch": 1.6081330868761552,
      "grad_norm": 2.09375,
      "learning_rate": 2.0282353920633513e-06,
      "loss": 1.5968,
      "step": 1740
    },
    {
      "epoch": 1.6090573012939002,
      "grad_norm": 2.109375,
      "learning_rate": 2.019014904761879e-06,
      "loss": 1.6299,
      "step": 1741
    },
    {
      "epoch": 1.609981515711645,
      "grad_norm": 2.140625,
      "learning_rate": 2.009813069765396e-06,
      "loss": 1.6786,
      "step": 1742
    },
    {
      "epoch": 1.61090573012939,
      "grad_norm": 2.234375,
      "learning_rate": 2.000629908579449e-06,
      "loss": 1.668,
      "step": 1743
    },
    {
      "epoch": 1.611829944547135,
      "grad_norm": 2.125,
      "learning_rate": 1.9914654426659374e-06,
      "loss": 1.6618,
      "step": 1744
    },
    {
      "epoch": 1.6127541589648797,
      "grad_norm": 2.140625,
      "learning_rate": 1.9823196934430743e-06,
      "loss": 1.5754,
      "step": 1745
    },
    {
      "epoch": 1.6136783733826248,
      "grad_norm": 2.125,
      "learning_rate": 1.9731926822853255e-06,
      "loss": 1.6699,
      "step": 1746
    },
    {
      "epoch": 1.6146025878003698,
      "grad_norm": 2.109375,
      "learning_rate": 1.9640844305233643e-06,
      "loss": 1.6882,
      "step": 1747
    },
    {
      "epoch": 1.6155268022181146,
      "grad_norm": 2.125,
      "learning_rate": 1.954994959444021e-06,
      "loss": 1.6647,
      "step": 1748
    },
    {
      "epoch": 1.6164510166358594,
      "grad_norm": 2.140625,
      "learning_rate": 1.945924290290242e-06,
      "loss": 1.6628,
      "step": 1749
    },
    {
      "epoch": 1.6173752310536045,
      "grad_norm": 2.265625,
      "learning_rate": 1.936872444261022e-06,
      "loss": 1.7196,
      "step": 1750
    },
    {
      "epoch": 1.6182994454713495,
      "grad_norm": 2.15625,
      "learning_rate": 1.9278394425113666e-06,
      "loss": 1.6188,
      "step": 1751
    },
    {
      "epoch": 1.6192236598890943,
      "grad_norm": 2.125,
      "learning_rate": 1.9188253061522434e-06,
      "loss": 1.6413,
      "step": 1752
    },
    {
      "epoch": 1.6201478743068392,
      "grad_norm": 2.140625,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 1.6915,
      "step": 1753
    },
    {
      "epoch": 1.621072088724584,
      "grad_norm": 2.125,
      "learning_rate": 1.9008537138289528e-06,
      "loss": 1.6813,
      "step": 1754
    },
    {
      "epoch": 1.621996303142329,
      "grad_norm": 2.25,
      "learning_rate": 1.8918962998660705e-06,
      "loss": 1.5854,
      "step": 1755
    },
    {
      "epoch": 1.622920517560074,
      "grad_norm": 2.234375,
      "learning_rate": 1.8829578352961886e-06,
      "loss": 1.6958,
      "step": 1756
    },
    {
      "epoch": 1.6238447319778189,
      "grad_norm": 2.203125,
      "learning_rate": 1.874038341009331e-06,
      "loss": 1.7275,
      "step": 1757
    },
    {
      "epoch": 1.6247689463955637,
      "grad_norm": 2.15625,
      "learning_rate": 1.8651378378511841e-06,
      "loss": 1.6768,
      "step": 1758
    },
    {
      "epoch": 1.6256931608133087,
      "grad_norm": 2.140625,
      "learning_rate": 1.8562563466230577e-06,
      "loss": 1.6385,
      "step": 1759
    },
    {
      "epoch": 1.6266173752310538,
      "grad_norm": 2.125,
      "learning_rate": 1.84739388808182e-06,
      "loss": 1.6735,
      "step": 1760
    },
    {
      "epoch": 1.6275415896487986,
      "grad_norm": 2.0625,
      "learning_rate": 1.8385504829398614e-06,
      "loss": 1.603,
      "step": 1761
    },
    {
      "epoch": 1.6284658040665434,
      "grad_norm": 2.140625,
      "learning_rate": 1.8297261518650454e-06,
      "loss": 1.6916,
      "step": 1762
    },
    {
      "epoch": 1.6293900184842882,
      "grad_norm": 2.140625,
      "learning_rate": 1.8209209154806484e-06,
      "loss": 1.7388,
      "step": 1763
    },
    {
      "epoch": 1.6303142329020333,
      "grad_norm": 2.078125,
      "learning_rate": 1.8121347943653333e-06,
      "loss": 1.5955,
      "step": 1764
    },
    {
      "epoch": 1.6312384473197783,
      "grad_norm": 2.1875,
      "learning_rate": 1.8033678090530814e-06,
      "loss": 1.7096,
      "step": 1765
    },
    {
      "epoch": 1.6321626617375231,
      "grad_norm": 2.15625,
      "learning_rate": 1.7946199800331533e-06,
      "loss": 1.701,
      "step": 1766
    },
    {
      "epoch": 1.633086876155268,
      "grad_norm": 2.15625,
      "learning_rate": 1.78589132775004e-06,
      "loss": 1.7094,
      "step": 1767
    },
    {
      "epoch": 1.634011090573013,
      "grad_norm": 2.109375,
      "learning_rate": 1.7771818726034106e-06,
      "loss": 1.6379,
      "step": 1768
    },
    {
      "epoch": 1.634935304990758,
      "grad_norm": 2.125,
      "learning_rate": 1.7684916349480796e-06,
      "loss": 1.6712,
      "step": 1769
    },
    {
      "epoch": 1.6358595194085028,
      "grad_norm": 2.109375,
      "learning_rate": 1.7598206350939384e-06,
      "loss": 1.7486,
      "step": 1770
    },
    {
      "epoch": 1.6367837338262476,
      "grad_norm": 2.21875,
      "learning_rate": 1.7511688933059178e-06,
      "loss": 1.581,
      "step": 1771
    },
    {
      "epoch": 1.6377079482439925,
      "grad_norm": 2.21875,
      "learning_rate": 1.7425364298039426e-06,
      "loss": 1.6845,
      "step": 1772
    },
    {
      "epoch": 1.6386321626617375,
      "grad_norm": 2.109375,
      "learning_rate": 1.733923264762889e-06,
      "loss": 1.6657,
      "step": 1773
    },
    {
      "epoch": 1.6395563770794825,
      "grad_norm": 2.140625,
      "learning_rate": 1.7253294183125223e-06,
      "loss": 1.6386,
      "step": 1774
    },
    {
      "epoch": 1.6404805914972274,
      "grad_norm": 2.171875,
      "learning_rate": 1.7167549105374614e-06,
      "loss": 1.6283,
      "step": 1775
    },
    {
      "epoch": 1.6414048059149722,
      "grad_norm": 2.140625,
      "learning_rate": 1.7081997614771306e-06,
      "loss": 1.6964,
      "step": 1776
    },
    {
      "epoch": 1.6423290203327172,
      "grad_norm": 2.140625,
      "learning_rate": 1.6996639911257051e-06,
      "loss": 1.6741,
      "step": 1777
    },
    {
      "epoch": 1.6432532347504623,
      "grad_norm": 2.109375,
      "learning_rate": 1.691147619432083e-06,
      "loss": 1.6756,
      "step": 1778
    },
    {
      "epoch": 1.644177449168207,
      "grad_norm": 2.125,
      "learning_rate": 1.6826506662998098e-06,
      "loss": 1.7271,
      "step": 1779
    },
    {
      "epoch": 1.645101663585952,
      "grad_norm": 2.15625,
      "learning_rate": 1.6741731515870596e-06,
      "loss": 1.6986,
      "step": 1780
    },
    {
      "epoch": 1.6460258780036967,
      "grad_norm": 2.21875,
      "learning_rate": 1.6657150951065716e-06,
      "loss": 1.7367,
      "step": 1781
    },
    {
      "epoch": 1.6469500924214417,
      "grad_norm": 2.171875,
      "learning_rate": 1.65727651662561e-06,
      "loss": 1.6797,
      "step": 1782
    },
    {
      "epoch": 1.6478743068391868,
      "grad_norm": 2.09375,
      "learning_rate": 1.6488574358659214e-06,
      "loss": 1.6388,
      "step": 1783
    },
    {
      "epoch": 1.6487985212569316,
      "grad_norm": 2.15625,
      "learning_rate": 1.6404578725036802e-06,
      "loss": 1.6462,
      "step": 1784
    },
    {
      "epoch": 1.6497227356746764,
      "grad_norm": 2.171875,
      "learning_rate": 1.6320778461694465e-06,
      "loss": 1.6633,
      "step": 1785
    },
    {
      "epoch": 1.6506469500924215,
      "grad_norm": 2.140625,
      "learning_rate": 1.623717376448123e-06,
      "loss": 1.6329,
      "step": 1786
    },
    {
      "epoch": 1.6515711645101665,
      "grad_norm": 2.171875,
      "learning_rate": 1.6153764828789043e-06,
      "loss": 1.6788,
      "step": 1787
    },
    {
      "epoch": 1.6524953789279113,
      "grad_norm": 2.125,
      "learning_rate": 1.6070551849552362e-06,
      "loss": 1.7128,
      "step": 1788
    },
    {
      "epoch": 1.6534195933456561,
      "grad_norm": 2.171875,
      "learning_rate": 1.5987535021247668e-06,
      "loss": 1.6424,
      "step": 1789
    },
    {
      "epoch": 1.654343807763401,
      "grad_norm": 2.109375,
      "learning_rate": 1.5904714537892996e-06,
      "loss": 1.6429,
      "step": 1790
    },
    {
      "epoch": 1.655268022181146,
      "grad_norm": 2.140625,
      "learning_rate": 1.5822090593047534e-06,
      "loss": 1.6624,
      "step": 1791
    },
    {
      "epoch": 1.656192236598891,
      "grad_norm": 2.140625,
      "learning_rate": 1.5739663379811122e-06,
      "loss": 1.6841,
      "step": 1792
    },
    {
      "epoch": 1.6571164510166358,
      "grad_norm": 2.140625,
      "learning_rate": 1.5657433090823847e-06,
      "loss": 1.6478,
      "step": 1793
    },
    {
      "epoch": 1.6580406654343807,
      "grad_norm": 2.15625,
      "learning_rate": 1.5575399918265543e-06,
      "loss": 1.7295,
      "step": 1794
    },
    {
      "epoch": 1.6589648798521257,
      "grad_norm": 2.0625,
      "learning_rate": 1.5493564053855381e-06,
      "loss": 1.648,
      "step": 1795
    },
    {
      "epoch": 1.6598890942698707,
      "grad_norm": 2.109375,
      "learning_rate": 1.541192568885136e-06,
      "loss": 1.5844,
      "step": 1796
    },
    {
      "epoch": 1.6608133086876156,
      "grad_norm": 2.15625,
      "learning_rate": 1.5330485014049968e-06,
      "loss": 1.6718,
      "step": 1797
    },
    {
      "epoch": 1.6617375231053604,
      "grad_norm": 2.1875,
      "learning_rate": 1.524924221978563e-06,
      "loss": 1.7539,
      "step": 1798
    },
    {
      "epoch": 1.6626617375231052,
      "grad_norm": 2.203125,
      "learning_rate": 1.5168197495930315e-06,
      "loss": 1.6613,
      "step": 1799
    },
    {
      "epoch": 1.6635859519408502,
      "grad_norm": 2.1875,
      "learning_rate": 1.5087351031893083e-06,
      "loss": 1.6992,
      "step": 1800
    },
    {
      "epoch": 1.6645101663585953,
      "grad_norm": 2.125,
      "learning_rate": 1.5006703016619662e-06,
      "loss": 1.6668,
      "step": 1801
    },
    {
      "epoch": 1.66543438077634,
      "grad_norm": 2.1875,
      "learning_rate": 1.4926253638591926e-06,
      "loss": 1.7006,
      "step": 1802
    },
    {
      "epoch": 1.666358595194085,
      "grad_norm": 2.15625,
      "learning_rate": 1.484600308582762e-06,
      "loss": 1.7117,
      "step": 1803
    },
    {
      "epoch": 1.66728280961183,
      "grad_norm": 2.140625,
      "learning_rate": 1.4765951545879732e-06,
      "loss": 1.6933,
      "step": 1804
    },
    {
      "epoch": 1.668207024029575,
      "grad_norm": 2.171875,
      "learning_rate": 1.468609920583617e-06,
      "loss": 1.6115,
      "step": 1805
    },
    {
      "epoch": 1.6691312384473198,
      "grad_norm": 2.15625,
      "learning_rate": 1.460644625231925e-06,
      "loss": 1.7159,
      "step": 1806
    },
    {
      "epoch": 1.6700554528650646,
      "grad_norm": 2.125,
      "learning_rate": 1.4526992871485346e-06,
      "loss": 1.6245,
      "step": 1807
    },
    {
      "epoch": 1.6709796672828097,
      "grad_norm": 2.0625,
      "learning_rate": 1.4447739249024429e-06,
      "loss": 1.7,
      "step": 1808
    },
    {
      "epoch": 1.6719038817005545,
      "grad_norm": 2.15625,
      "learning_rate": 1.436868557015959e-06,
      "loss": 1.6692,
      "step": 1809
    },
    {
      "epoch": 1.6728280961182995,
      "grad_norm": 2.171875,
      "learning_rate": 1.428983201964662e-06,
      "loss": 1.6697,
      "step": 1810
    },
    {
      "epoch": 1.6737523105360443,
      "grad_norm": 2.171875,
      "learning_rate": 1.4211178781773582e-06,
      "loss": 1.7418,
      "step": 1811
    },
    {
      "epoch": 1.6746765249537892,
      "grad_norm": 2.203125,
      "learning_rate": 1.4132726040360468e-06,
      "loss": 1.6979,
      "step": 1812
    },
    {
      "epoch": 1.6756007393715342,
      "grad_norm": 2.28125,
      "learning_rate": 1.4054473978758632e-06,
      "loss": 1.6794,
      "step": 1813
    },
    {
      "epoch": 1.6765249537892792,
      "grad_norm": 2.15625,
      "learning_rate": 1.3976422779850385e-06,
      "loss": 1.7166,
      "step": 1814
    },
    {
      "epoch": 1.677449168207024,
      "grad_norm": 2.109375,
      "learning_rate": 1.3898572626048656e-06,
      "loss": 1.6725,
      "step": 1815
    },
    {
      "epoch": 1.6783733826247689,
      "grad_norm": 2.15625,
      "learning_rate": 1.3820923699296485e-06,
      "loss": 1.7308,
      "step": 1816
    },
    {
      "epoch": 1.679297597042514,
      "grad_norm": 2.140625,
      "learning_rate": 1.3743476181066684e-06,
      "loss": 1.6168,
      "step": 1817
    },
    {
      "epoch": 1.6802218114602587,
      "grad_norm": 2.203125,
      "learning_rate": 1.366623025236129e-06,
      "loss": 1.757,
      "step": 1818
    },
    {
      "epoch": 1.6811460258780038,
      "grad_norm": 2.171875,
      "learning_rate": 1.3589186093711227e-06,
      "loss": 1.7461,
      "step": 1819
    },
    {
      "epoch": 1.6820702402957486,
      "grad_norm": 2.234375,
      "learning_rate": 1.351234388517586e-06,
      "loss": 1.696,
      "step": 1820
    },
    {
      "epoch": 1.6829944547134934,
      "grad_norm": 2.140625,
      "learning_rate": 1.3435703806342592e-06,
      "loss": 1.6843,
      "step": 1821
    },
    {
      "epoch": 1.6839186691312384,
      "grad_norm": 2.15625,
      "learning_rate": 1.3359266036326413e-06,
      "loss": 1.6991,
      "step": 1822
    },
    {
      "epoch": 1.6848428835489835,
      "grad_norm": 2.09375,
      "learning_rate": 1.328303075376951e-06,
      "loss": 1.725,
      "step": 1823
    },
    {
      "epoch": 1.6857670979667283,
      "grad_norm": 2.234375,
      "learning_rate": 1.3206998136840833e-06,
      "loss": 1.6876,
      "step": 1824
    },
    {
      "epoch": 1.6866913123844731,
      "grad_norm": 2.15625,
      "learning_rate": 1.313116836323568e-06,
      "loss": 1.7342,
      "step": 1825
    },
    {
      "epoch": 1.6876155268022182,
      "grad_norm": 2.125,
      "learning_rate": 1.3055541610175282e-06,
      "loss": 1.6485,
      "step": 1826
    },
    {
      "epoch": 1.6885397412199632,
      "grad_norm": 2.140625,
      "learning_rate": 1.298011805440642e-06,
      "loss": 1.6714,
      "step": 1827
    },
    {
      "epoch": 1.689463955637708,
      "grad_norm": 2.140625,
      "learning_rate": 1.290489787220096e-06,
      "loss": 1.6743,
      "step": 1828
    },
    {
      "epoch": 1.6903881700554528,
      "grad_norm": 2.125,
      "learning_rate": 1.2829881239355469e-06,
      "loss": 1.6445,
      "step": 1829
    },
    {
      "epoch": 1.6913123844731976,
      "grad_norm": 2.15625,
      "learning_rate": 1.2755068331190789e-06,
      "loss": 1.6244,
      "step": 1830
    },
    {
      "epoch": 1.6922365988909427,
      "grad_norm": 2.171875,
      "learning_rate": 1.2680459322551652e-06,
      "loss": 1.724,
      "step": 1831
    },
    {
      "epoch": 1.6931608133086877,
      "grad_norm": 2.109375,
      "learning_rate": 1.2606054387806265e-06,
      "loss": 1.6201,
      "step": 1832
    },
    {
      "epoch": 1.6940850277264325,
      "grad_norm": 2.171875,
      "learning_rate": 1.2531853700845875e-06,
      "loss": 1.6364,
      "step": 1833
    },
    {
      "epoch": 1.6950092421441774,
      "grad_norm": 2.125,
      "learning_rate": 1.245785743508441e-06,
      "loss": 1.6783,
      "step": 1834
    },
    {
      "epoch": 1.6959334565619224,
      "grad_norm": 2.15625,
      "learning_rate": 1.2384065763457997e-06,
      "loss": 1.6219,
      "step": 1835
    },
    {
      "epoch": 1.6968576709796674,
      "grad_norm": 2.1875,
      "learning_rate": 1.2310478858424656e-06,
      "loss": 1.7286,
      "step": 1836
    },
    {
      "epoch": 1.6977818853974123,
      "grad_norm": 2.140625,
      "learning_rate": 1.2237096891963862e-06,
      "loss": 1.682,
      "step": 1837
    },
    {
      "epoch": 1.698706099815157,
      "grad_norm": 2.140625,
      "learning_rate": 1.216392003557607e-06,
      "loss": 1.6028,
      "step": 1838
    },
    {
      "epoch": 1.699630314232902,
      "grad_norm": 2.21875,
      "learning_rate": 1.2090948460282415e-06,
      "loss": 1.681,
      "step": 1839
    },
    {
      "epoch": 1.700554528650647,
      "grad_norm": 2.140625,
      "learning_rate": 1.2018182336624272e-06,
      "loss": 1.6552,
      "step": 1840
    },
    {
      "epoch": 1.701478743068392,
      "grad_norm": 2.1875,
      "learning_rate": 1.194562183466279e-06,
      "loss": 1.6955,
      "step": 1841
    },
    {
      "epoch": 1.7024029574861368,
      "grad_norm": 2.203125,
      "learning_rate": 1.1873267123978683e-06,
      "loss": 1.7359,
      "step": 1842
    },
    {
      "epoch": 1.7033271719038816,
      "grad_norm": 2.21875,
      "learning_rate": 1.1801118373671617e-06,
      "loss": 1.6675,
      "step": 1843
    },
    {
      "epoch": 1.7042513863216266,
      "grad_norm": 2.1875,
      "learning_rate": 1.1729175752359922e-06,
      "loss": 1.7108,
      "step": 1844
    },
    {
      "epoch": 1.7051756007393717,
      "grad_norm": 2.140625,
      "learning_rate": 1.1657439428180228e-06,
      "loss": 1.6802,
      "step": 1845
    },
    {
      "epoch": 1.7060998151571165,
      "grad_norm": 2.171875,
      "learning_rate": 1.1585909568786957e-06,
      "loss": 1.7424,
      "step": 1846
    },
    {
      "epoch": 1.7070240295748613,
      "grad_norm": 2.109375,
      "learning_rate": 1.1514586341352108e-06,
      "loss": 1.6835,
      "step": 1847
    },
    {
      "epoch": 1.7079482439926061,
      "grad_norm": 2.203125,
      "learning_rate": 1.1443469912564685e-06,
      "loss": 1.7367,
      "step": 1848
    },
    {
      "epoch": 1.7088724584103512,
      "grad_norm": 2.171875,
      "learning_rate": 1.1372560448630377e-06,
      "loss": 1.7265,
      "step": 1849
    },
    {
      "epoch": 1.7097966728280962,
      "grad_norm": 2.15625,
      "learning_rate": 1.130185811527119e-06,
      "loss": 1.7146,
      "step": 1850
    },
    {
      "epoch": 1.710720887245841,
      "grad_norm": 2.171875,
      "learning_rate": 1.12313630777251e-06,
      "loss": 1.6858,
      "step": 1851
    },
    {
      "epoch": 1.7116451016635859,
      "grad_norm": 2.171875,
      "learning_rate": 1.1161075500745544e-06,
      "loss": 1.8063,
      "step": 1852
    },
    {
      "epoch": 1.712569316081331,
      "grad_norm": 2.171875,
      "learning_rate": 1.1090995548601147e-06,
      "loss": 1.7348,
      "step": 1853
    },
    {
      "epoch": 1.713493530499076,
      "grad_norm": 2.140625,
      "learning_rate": 1.1021123385075261e-06,
      "loss": 1.696,
      "step": 1854
    },
    {
      "epoch": 1.7144177449168208,
      "grad_norm": 2.109375,
      "learning_rate": 1.0951459173465628e-06,
      "loss": 1.7036,
      "step": 1855
    },
    {
      "epoch": 1.7153419593345656,
      "grad_norm": 2.15625,
      "learning_rate": 1.0882003076584059e-06,
      "loss": 1.6283,
      "step": 1856
    },
    {
      "epoch": 1.7162661737523104,
      "grad_norm": 2.234375,
      "learning_rate": 1.0812755256755848e-06,
      "loss": 1.7004,
      "step": 1857
    },
    {
      "epoch": 1.7171903881700554,
      "grad_norm": 2.125,
      "learning_rate": 1.0743715875819616e-06,
      "loss": 1.7005,
      "step": 1858
    },
    {
      "epoch": 1.7181146025878005,
      "grad_norm": 2.140625,
      "learning_rate": 1.067488509512683e-06,
      "loss": 1.7246,
      "step": 1859
    },
    {
      "epoch": 1.7190388170055453,
      "grad_norm": 2.15625,
      "learning_rate": 1.060626307554139e-06,
      "loss": 1.6638,
      "step": 1860
    },
    {
      "epoch": 1.71996303142329,
      "grad_norm": 2.15625,
      "learning_rate": 1.0537849977439406e-06,
      "loss": 1.6188,
      "step": 1861
    },
    {
      "epoch": 1.7208872458410351,
      "grad_norm": 2.15625,
      "learning_rate": 1.046964596070863e-06,
      "loss": 1.7115,
      "step": 1862
    },
    {
      "epoch": 1.7218114602587802,
      "grad_norm": 2.109375,
      "learning_rate": 1.0401651184748184e-06,
      "loss": 1.6051,
      "step": 1863
    },
    {
      "epoch": 1.722735674676525,
      "grad_norm": 2.15625,
      "learning_rate": 1.0333865808468203e-06,
      "loss": 1.6756,
      "step": 1864
    },
    {
      "epoch": 1.7236598890942698,
      "grad_norm": 2.21875,
      "learning_rate": 1.0266289990289413e-06,
      "loss": 1.7087,
      "step": 1865
    },
    {
      "epoch": 1.7245841035120146,
      "grad_norm": 2.15625,
      "learning_rate": 1.0198923888142798e-06,
      "loss": 1.677,
      "step": 1866
    },
    {
      "epoch": 1.7255083179297597,
      "grad_norm": 2.171875,
      "learning_rate": 1.0131767659469206e-06,
      "loss": 1.6409,
      "step": 1867
    },
    {
      "epoch": 1.7264325323475047,
      "grad_norm": 2.234375,
      "learning_rate": 1.0064821461218976e-06,
      "loss": 1.7091,
      "step": 1868
    },
    {
      "epoch": 1.7273567467652495,
      "grad_norm": 2.234375,
      "learning_rate": 9.998085449851635e-07,
      "loss": 1.7075,
      "step": 1869
    },
    {
      "epoch": 1.7282809611829943,
      "grad_norm": 2.140625,
      "learning_rate": 9.93155978133541e-07,
      "loss": 1.6839,
      "step": 1870
    },
    {
      "epoch": 1.7292051756007394,
      "grad_norm": 2.109375,
      "learning_rate": 9.86524461114703e-07,
      "loss": 1.6832,
      "step": 1871
    },
    {
      "epoch": 1.7301293900184844,
      "grad_norm": 2.234375,
      "learning_rate": 9.799140094271198e-07,
      "loss": 1.6676,
      "step": 1872
    },
    {
      "epoch": 1.7310536044362292,
      "grad_norm": 2.203125,
      "learning_rate": 9.733246385200313e-07,
      "loss": 1.711,
      "step": 1873
    },
    {
      "epoch": 1.731977818853974,
      "grad_norm": 2.140625,
      "learning_rate": 9.66756363793413e-07,
      "loss": 1.6818,
      "step": 1874
    },
    {
      "epoch": 1.7329020332717189,
      "grad_norm": 2.140625,
      "learning_rate": 9.602092005979314e-07,
      "loss": 1.6639,
      "step": 1875
    },
    {
      "epoch": 1.733826247689464,
      "grad_norm": 2.15625,
      "learning_rate": 9.536831642349186e-07,
      "loss": 1.654,
      "step": 1876
    },
    {
      "epoch": 1.734750462107209,
      "grad_norm": 2.109375,
      "learning_rate": 9.471782699563281e-07,
      "loss": 1.6257,
      "step": 1877
    },
    {
      "epoch": 1.7356746765249538,
      "grad_norm": 2.15625,
      "learning_rate": 9.406945329647021e-07,
      "loss": 1.6474,
      "step": 1878
    },
    {
      "epoch": 1.7365988909426986,
      "grad_norm": 2.15625,
      "learning_rate": 9.342319684131396e-07,
      "loss": 1.6825,
      "step": 1879
    },
    {
      "epoch": 1.7375231053604436,
      "grad_norm": 2.203125,
      "learning_rate": 9.277905914052521e-07,
      "loss": 1.6631,
      "step": 1880
    },
    {
      "epoch": 1.7384473197781887,
      "grad_norm": 2.1875,
      "learning_rate": 9.213704169951421e-07,
      "loss": 1.6688,
      "step": 1881
    },
    {
      "epoch": 1.7393715341959335,
      "grad_norm": 2.171875,
      "learning_rate": 9.149714601873516e-07,
      "loss": 1.6986,
      "step": 1882
    },
    {
      "epoch": 1.7402957486136783,
      "grad_norm": 2.1875,
      "learning_rate": 9.085937359368402e-07,
      "loss": 1.7323,
      "step": 1883
    },
    {
      "epoch": 1.7412199630314233,
      "grad_norm": 2.15625,
      "learning_rate": 9.022372591489381e-07,
      "loss": 1.6647,
      "step": 1884
    },
    {
      "epoch": 1.7421441774491682,
      "grad_norm": 2.1875,
      "learning_rate": 8.959020446793288e-07,
      "loss": 1.7657,
      "step": 1885
    },
    {
      "epoch": 1.7430683918669132,
      "grad_norm": 2.1875,
      "learning_rate": 8.895881073339963e-07,
      "loss": 1.7017,
      "step": 1886
    },
    {
      "epoch": 1.743992606284658,
      "grad_norm": 2.1875,
      "learning_rate": 8.832954618691991e-07,
      "loss": 1.683,
      "step": 1887
    },
    {
      "epoch": 1.7449168207024028,
      "grad_norm": 2.234375,
      "learning_rate": 8.770241229914356e-07,
      "loss": 1.6885,
      "step": 1888
    },
    {
      "epoch": 1.7458410351201479,
      "grad_norm": 2.109375,
      "learning_rate": 8.70774105357407e-07,
      "loss": 1.6864,
      "step": 1889
    },
    {
      "epoch": 1.746765249537893,
      "grad_norm": 2.15625,
      "learning_rate": 8.645454235739903e-07,
      "loss": 1.7023,
      "step": 1890
    },
    {
      "epoch": 1.7476894639556377,
      "grad_norm": 2.09375,
      "learning_rate": 8.583380921981932e-07,
      "loss": 1.6499,
      "step": 1891
    },
    {
      "epoch": 1.7486136783733826,
      "grad_norm": 2.171875,
      "learning_rate": 8.521521257371245e-07,
      "loss": 1.6935,
      "step": 1892
    },
    {
      "epoch": 1.7495378927911276,
      "grad_norm": 2.1875,
      "learning_rate": 8.459875386479654e-07,
      "loss": 1.6654,
      "step": 1893
    },
    {
      "epoch": 1.7504621072088724,
      "grad_norm": 2.203125,
      "learning_rate": 8.398443453379268e-07,
      "loss": 1.7092,
      "step": 1894
    },
    {
      "epoch": 1.7513863216266174,
      "grad_norm": 2.09375,
      "learning_rate": 8.337225601642274e-07,
      "loss": 1.6648,
      "step": 1895
    },
    {
      "epoch": 1.7523105360443623,
      "grad_norm": 2.171875,
      "learning_rate": 8.276221974340481e-07,
      "loss": 1.7173,
      "step": 1896
    },
    {
      "epoch": 1.753234750462107,
      "grad_norm": 2.28125,
      "learning_rate": 8.215432714045024e-07,
      "loss": 1.7339,
      "step": 1897
    },
    {
      "epoch": 1.7541589648798521,
      "grad_norm": 2.265625,
      "learning_rate": 8.154857962826079e-07,
      "loss": 1.7517,
      "step": 1898
    },
    {
      "epoch": 1.7550831792975972,
      "grad_norm": 2.140625,
      "learning_rate": 8.09449786225247e-07,
      "loss": 1.6352,
      "step": 1899
    },
    {
      "epoch": 1.756007393715342,
      "grad_norm": 2.125,
      "learning_rate": 8.034352553391367e-07,
      "loss": 1.7038,
      "step": 1900
    },
    {
      "epoch": 1.7569316081330868,
      "grad_norm": 2.28125,
      "learning_rate": 7.974422176807961e-07,
      "loss": 1.7253,
      "step": 1901
    },
    {
      "epoch": 1.7578558225508318,
      "grad_norm": 2.140625,
      "learning_rate": 7.914706872565115e-07,
      "loss": 1.6413,
      "step": 1902
    },
    {
      "epoch": 1.7587800369685767,
      "grad_norm": 2.15625,
      "learning_rate": 7.855206780223057e-07,
      "loss": 1.6303,
      "step": 1903
    },
    {
      "epoch": 1.7597042513863217,
      "grad_norm": 2.171875,
      "learning_rate": 7.795922038839032e-07,
      "loss": 1.7049,
      "step": 1904
    },
    {
      "epoch": 1.7606284658040665,
      "grad_norm": 2.140625,
      "learning_rate": 7.736852786967019e-07,
      "loss": 1.6553,
      "step": 1905
    },
    {
      "epoch": 1.7615526802218113,
      "grad_norm": 2.09375,
      "learning_rate": 7.677999162657335e-07,
      "loss": 1.6725,
      "step": 1906
    },
    {
      "epoch": 1.7624768946395564,
      "grad_norm": 2.125,
      "learning_rate": 7.619361303456396e-07,
      "loss": 1.7171,
      "step": 1907
    },
    {
      "epoch": 1.7634011090573014,
      "grad_norm": 2.15625,
      "learning_rate": 7.560939346406326e-07,
      "loss": 1.6409,
      "step": 1908
    },
    {
      "epoch": 1.7643253234750462,
      "grad_norm": 2.109375,
      "learning_rate": 7.502733428044684e-07,
      "loss": 1.6762,
      "step": 1909
    },
    {
      "epoch": 1.765249537892791,
      "grad_norm": 2.1875,
      "learning_rate": 7.44474368440411e-07,
      "loss": 1.7212,
      "step": 1910
    },
    {
      "epoch": 1.766173752310536,
      "grad_norm": 2.15625,
      "learning_rate": 7.386970251012037e-07,
      "loss": 1.6699,
      "step": 1911
    },
    {
      "epoch": 1.7670979667282811,
      "grad_norm": 2.140625,
      "learning_rate": 7.329413262890351e-07,
      "loss": 1.6666,
      "step": 1912
    },
    {
      "epoch": 1.768022181146026,
      "grad_norm": 2.171875,
      "learning_rate": 7.27207285455509e-07,
      "loss": 1.7303,
      "step": 1913
    },
    {
      "epoch": 1.7689463955637708,
      "grad_norm": 2.140625,
      "learning_rate": 7.214949160016116e-07,
      "loss": 1.739,
      "step": 1914
    },
    {
      "epoch": 1.7698706099815156,
      "grad_norm": 2.1875,
      "learning_rate": 7.158042312776847e-07,
      "loss": 1.6231,
      "step": 1915
    },
    {
      "epoch": 1.7707948243992606,
      "grad_norm": 2.140625,
      "learning_rate": 7.101352445833865e-07,
      "loss": 1.7046,
      "step": 1916
    },
    {
      "epoch": 1.7717190388170057,
      "grad_norm": 2.140625,
      "learning_rate": 7.044879691676676e-07,
      "loss": 1.7177,
      "step": 1917
    },
    {
      "epoch": 1.7726432532347505,
      "grad_norm": 2.265625,
      "learning_rate": 6.988624182287362e-07,
      "loss": 1.7265,
      "step": 1918
    },
    {
      "epoch": 1.7735674676524953,
      "grad_norm": 2.109375,
      "learning_rate": 6.932586049140255e-07,
      "loss": 1.6002,
      "step": 1919
    },
    {
      "epoch": 1.7744916820702403,
      "grad_norm": 2.15625,
      "learning_rate": 6.876765423201715e-07,
      "loss": 1.6873,
      "step": 1920
    },
    {
      "epoch": 1.7754158964879854,
      "grad_norm": 2.203125,
      "learning_rate": 6.82116243492974e-07,
      "loss": 1.7209,
      "step": 1921
    },
    {
      "epoch": 1.7763401109057302,
      "grad_norm": 2.109375,
      "learning_rate": 6.765777214273683e-07,
      "loss": 1.6982,
      "step": 1922
    },
    {
      "epoch": 1.777264325323475,
      "grad_norm": 2.125,
      "learning_rate": 6.710609890673936e-07,
      "loss": 1.6647,
      "step": 1923
    },
    {
      "epoch": 1.7781885397412198,
      "grad_norm": 2.125,
      "learning_rate": 6.655660593061719e-07,
      "loss": 1.6599,
      "step": 1924
    },
    {
      "epoch": 1.7791127541589649,
      "grad_norm": 2.171875,
      "learning_rate": 6.600929449858629e-07,
      "loss": 1.731,
      "step": 1925
    },
    {
      "epoch": 1.78003696857671,
      "grad_norm": 2.171875,
      "learning_rate": 6.546416588976445e-07,
      "loss": 1.6241,
      "step": 1926
    },
    {
      "epoch": 1.7809611829944547,
      "grad_norm": 2.109375,
      "learning_rate": 6.492122137816792e-07,
      "loss": 1.6618,
      "step": 1927
    },
    {
      "epoch": 1.7818853974121995,
      "grad_norm": 2.171875,
      "learning_rate": 6.438046223270833e-07,
      "loss": 1.6418,
      "step": 1928
    },
    {
      "epoch": 1.7828096118299446,
      "grad_norm": 2.15625,
      "learning_rate": 6.384188971719052e-07,
      "loss": 1.6577,
      "step": 1929
    },
    {
      "epoch": 1.7837338262476896,
      "grad_norm": 2.09375,
      "learning_rate": 6.330550509030853e-07,
      "loss": 1.6285,
      "step": 1930
    },
    {
      "epoch": 1.7846580406654344,
      "grad_norm": 2.125,
      "learning_rate": 6.277130960564292e-07,
      "loss": 1.6872,
      "step": 1931
    },
    {
      "epoch": 1.7855822550831792,
      "grad_norm": 2.109375,
      "learning_rate": 6.223930451165849e-07,
      "loss": 1.6117,
      "step": 1932
    },
    {
      "epoch": 1.786506469500924,
      "grad_norm": 2.109375,
      "learning_rate": 6.170949105170043e-07,
      "loss": 1.6835,
      "step": 1933
    },
    {
      "epoch": 1.787430683918669,
      "grad_norm": 2.203125,
      "learning_rate": 6.11818704639926e-07,
      "loss": 1.6912,
      "step": 1934
    },
    {
      "epoch": 1.7883548983364141,
      "grad_norm": 2.109375,
      "learning_rate": 6.065644398163284e-07,
      "loss": 1.5874,
      "step": 1935
    },
    {
      "epoch": 1.789279112754159,
      "grad_norm": 2.171875,
      "learning_rate": 6.013321283259199e-07,
      "loss": 1.6781,
      "step": 1936
    },
    {
      "epoch": 1.7902033271719038,
      "grad_norm": 2.109375,
      "learning_rate": 5.961217823970988e-07,
      "loss": 1.6647,
      "step": 1937
    },
    {
      "epoch": 1.7911275415896488,
      "grad_norm": 2.125,
      "learning_rate": 5.909334142069278e-07,
      "loss": 1.5979,
      "step": 1938
    },
    {
      "epoch": 1.7920517560073939,
      "grad_norm": 2.15625,
      "learning_rate": 5.857670358811096e-07,
      "loss": 1.6654,
      "step": 1939
    },
    {
      "epoch": 1.7929759704251387,
      "grad_norm": 2.125,
      "learning_rate": 5.80622659493949e-07,
      "loss": 1.722,
      "step": 1940
    },
    {
      "epoch": 1.7939001848428835,
      "grad_norm": 2.21875,
      "learning_rate": 5.755002970683321e-07,
      "loss": 1.8237,
      "step": 1941
    },
    {
      "epoch": 1.7948243992606283,
      "grad_norm": 2.15625,
      "learning_rate": 5.70399960575696e-07,
      "loss": 1.696,
      "step": 1942
    },
    {
      "epoch": 1.7957486136783734,
      "grad_norm": 2.1875,
      "learning_rate": 5.653216619360047e-07,
      "loss": 1.6727,
      "step": 1943
    },
    {
      "epoch": 1.7966728280961184,
      "grad_norm": 2.125,
      "learning_rate": 5.60265413017711e-07,
      "loss": 1.6217,
      "step": 1944
    },
    {
      "epoch": 1.7975970425138632,
      "grad_norm": 2.171875,
      "learning_rate": 5.552312256377423e-07,
      "loss": 1.7085,
      "step": 1945
    },
    {
      "epoch": 1.798521256931608,
      "grad_norm": 2.09375,
      "learning_rate": 5.50219111561462e-07,
      "loss": 1.6472,
      "step": 1946
    },
    {
      "epoch": 1.799445471349353,
      "grad_norm": 2.15625,
      "learning_rate": 5.452290825026463e-07,
      "loss": 1.5847,
      "step": 1947
    },
    {
      "epoch": 1.800369685767098,
      "grad_norm": 2.109375,
      "learning_rate": 5.402611501234578e-07,
      "loss": 1.6924,
      "step": 1948
    },
    {
      "epoch": 1.801293900184843,
      "grad_norm": 2.171875,
      "learning_rate": 5.353153260344179e-07,
      "loss": 1.676,
      "step": 1949
    },
    {
      "epoch": 1.8022181146025877,
      "grad_norm": 2.1875,
      "learning_rate": 5.303916217943772e-07,
      "loss": 1.7218,
      "step": 1950
    },
    {
      "epoch": 1.8031423290203326,
      "grad_norm": 2.140625,
      "learning_rate": 5.254900489104919e-07,
      "loss": 1.6064,
      "step": 1951
    },
    {
      "epoch": 1.8040665434380776,
      "grad_norm": 2.109375,
      "learning_rate": 5.206106188381921e-07,
      "loss": 1.7153,
      "step": 1952
    },
    {
      "epoch": 1.8049907578558226,
      "grad_norm": 2.15625,
      "learning_rate": 5.157533429811612e-07,
      "loss": 1.6863,
      "step": 1953
    },
    {
      "epoch": 1.8059149722735675,
      "grad_norm": 2.1875,
      "learning_rate": 5.109182326913053e-07,
      "loss": 1.6782,
      "step": 1954
    },
    {
      "epoch": 1.8068391866913123,
      "grad_norm": 2.1875,
      "learning_rate": 5.061052992687254e-07,
      "loss": 1.696,
      "step": 1955
    },
    {
      "epoch": 1.8077634011090573,
      "grad_norm": 2.125,
      "learning_rate": 5.01314553961697e-07,
      "loss": 1.6038,
      "step": 1956
    },
    {
      "epoch": 1.8086876155268024,
      "grad_norm": 2.109375,
      "learning_rate": 4.965460079666362e-07,
      "loss": 1.6244,
      "step": 1957
    },
    {
      "epoch": 1.8096118299445472,
      "grad_norm": 2.078125,
      "learning_rate": 4.917996724280782e-07,
      "loss": 1.5711,
      "step": 1958
    },
    {
      "epoch": 1.810536044362292,
      "grad_norm": 2.125,
      "learning_rate": 4.870755584386544e-07,
      "loss": 1.6501,
      "step": 1959
    },
    {
      "epoch": 1.8114602587800368,
      "grad_norm": 2.140625,
      "learning_rate": 4.823736770390553e-07,
      "loss": 1.6201,
      "step": 1960
    },
    {
      "epoch": 1.8123844731977818,
      "grad_norm": 2.140625,
      "learning_rate": 4.776940392180185e-07,
      "loss": 1.6897,
      "step": 1961
    },
    {
      "epoch": 1.8133086876155269,
      "grad_norm": 2.15625,
      "learning_rate": 4.73036655912289e-07,
      "loss": 1.742,
      "step": 1962
    },
    {
      "epoch": 1.8142329020332717,
      "grad_norm": 2.15625,
      "learning_rate": 4.684015380066087e-07,
      "loss": 1.7342,
      "step": 1963
    },
    {
      "epoch": 1.8151571164510165,
      "grad_norm": 2.234375,
      "learning_rate": 4.6378869633367797e-07,
      "loss": 1.6994,
      "step": 1964
    },
    {
      "epoch": 1.8160813308687616,
      "grad_norm": 2.171875,
      "learning_rate": 4.5919814167413755e-07,
      "loss": 1.6879,
      "step": 1965
    },
    {
      "epoch": 1.8170055452865066,
      "grad_norm": 2.203125,
      "learning_rate": 4.5462988475654113e-07,
      "loss": 1.7065,
      "step": 1966
    },
    {
      "epoch": 1.8179297597042514,
      "grad_norm": 2.109375,
      "learning_rate": 4.500839362573284e-07,
      "loss": 1.7116,
      "step": 1967
    },
    {
      "epoch": 1.8188539741219962,
      "grad_norm": 2.15625,
      "learning_rate": 4.455603068008063e-07,
      "loss": 1.6573,
      "step": 1968
    },
    {
      "epoch": 1.8197781885397413,
      "grad_norm": 2.078125,
      "learning_rate": 4.410590069591192e-07,
      "loss": 1.579,
      "step": 1969
    },
    {
      "epoch": 1.820702402957486,
      "grad_norm": 2.21875,
      "learning_rate": 4.365800472522208e-07,
      "loss": 1.7174,
      "step": 1970
    },
    {
      "epoch": 1.8216266173752311,
      "grad_norm": 2.15625,
      "learning_rate": 4.321234381478578e-07,
      "loss": 1.6624,
      "step": 1971
    },
    {
      "epoch": 1.822550831792976,
      "grad_norm": 2.125,
      "learning_rate": 4.2768919006153876e-07,
      "loss": 1.7577,
      "step": 1972
    },
    {
      "epoch": 1.8234750462107208,
      "grad_norm": 2.109375,
      "learning_rate": 4.2327731335651824e-07,
      "loss": 1.6311,
      "step": 1973
    },
    {
      "epoch": 1.8243992606284658,
      "grad_norm": 2.15625,
      "learning_rate": 4.1888781834375947e-07,
      "loss": 1.6642,
      "step": 1974
    },
    {
      "epoch": 1.8253234750462108,
      "grad_norm": 2.125,
      "learning_rate": 4.1452071528192304e-07,
      "loss": 1.6249,
      "step": 1975
    },
    {
      "epoch": 1.8262476894639557,
      "grad_norm": 2.140625,
      "learning_rate": 4.1017601437733256e-07,
      "loss": 1.6727,
      "step": 1976
    },
    {
      "epoch": 1.8271719038817005,
      "grad_norm": 2.140625,
      "learning_rate": 4.058537257839612e-07,
      "loss": 1.6696,
      "step": 1977
    },
    {
      "epoch": 1.8280961182994455,
      "grad_norm": 2.171875,
      "learning_rate": 4.0155385960339743e-07,
      "loss": 1.6321,
      "step": 1978
    },
    {
      "epoch": 1.8290203327171903,
      "grad_norm": 2.140625,
      "learning_rate": 3.972764258848305e-07,
      "loss": 1.6942,
      "step": 1979
    },
    {
      "epoch": 1.8299445471349354,
      "grad_norm": 2.09375,
      "learning_rate": 3.930214346250205e-07,
      "loss": 1.6518,
      "step": 1980
    },
    {
      "epoch": 1.8308687615526802,
      "grad_norm": 2.203125,
      "learning_rate": 3.887888957682773e-07,
      "loss": 1.7098,
      "step": 1981
    },
    {
      "epoch": 1.831792975970425,
      "grad_norm": 2.15625,
      "learning_rate": 3.8457881920643815e-07,
      "loss": 1.7276,
      "step": 1982
    },
    {
      "epoch": 1.83271719038817,
      "grad_norm": 2.15625,
      "learning_rate": 3.8039121477884467e-07,
      "loss": 1.6752,
      "step": 1983
    },
    {
      "epoch": 1.833641404805915,
      "grad_norm": 2.109375,
      "learning_rate": 3.762260922723182e-07,
      "loss": 1.6225,
      "step": 1984
    },
    {
      "epoch": 1.83456561922366,
      "grad_norm": 2.15625,
      "learning_rate": 3.7208346142113773e-07,
      "loss": 1.6557,
      "step": 1985
    },
    {
      "epoch": 1.8354898336414047,
      "grad_norm": 2.21875,
      "learning_rate": 3.6796333190701636e-07,
      "loss": 1.7453,
      "step": 1986
    },
    {
      "epoch": 1.8364140480591498,
      "grad_norm": 2.125,
      "learning_rate": 3.638657133590817e-07,
      "loss": 1.6337,
      "step": 1987
    },
    {
      "epoch": 1.8373382624768948,
      "grad_norm": 2.15625,
      "learning_rate": 3.597906153538477e-07,
      "loss": 1.7055,
      "step": 1988
    },
    {
      "epoch": 1.8382624768946396,
      "grad_norm": 2.171875,
      "learning_rate": 3.5573804741519834e-07,
      "loss": 1.6566,
      "step": 1989
    },
    {
      "epoch": 1.8391866913123844,
      "grad_norm": 2.140625,
      "learning_rate": 3.517080190143629e-07,
      "loss": 1.6798,
      "step": 1990
    },
    {
      "epoch": 1.8401109057301293,
      "grad_norm": 2.125,
      "learning_rate": 3.477005395698929e-07,
      "loss": 1.6773,
      "step": 1991
    },
    {
      "epoch": 1.8410351201478743,
      "grad_norm": 2.109375,
      "learning_rate": 3.437156184476387e-07,
      "loss": 1.6994,
      "step": 1992
    },
    {
      "epoch": 1.8419593345656193,
      "grad_norm": 2.15625,
      "learning_rate": 3.397532649607338e-07,
      "loss": 1.7365,
      "step": 1993
    },
    {
      "epoch": 1.8428835489833642,
      "grad_norm": 2.1875,
      "learning_rate": 3.358134883695674e-07,
      "loss": 1.6523,
      "step": 1994
    },
    {
      "epoch": 1.843807763401109,
      "grad_norm": 2.140625,
      "learning_rate": 3.31896297881763e-07,
      "loss": 1.6887,
      "step": 1995
    },
    {
      "epoch": 1.844731977818854,
      "grad_norm": 2.09375,
      "learning_rate": 3.280017026521598e-07,
      "loss": 1.5847,
      "step": 1996
    },
    {
      "epoch": 1.845656192236599,
      "grad_norm": 2.125,
      "learning_rate": 3.241297117827891e-07,
      "loss": 1.6851,
      "step": 1997
    },
    {
      "epoch": 1.8465804066543439,
      "grad_norm": 2.09375,
      "learning_rate": 3.2028033432285444e-07,
      "loss": 1.6794,
      "step": 1998
    },
    {
      "epoch": 1.8475046210720887,
      "grad_norm": 2.140625,
      "learning_rate": 3.1645357926870957e-07,
      "loss": 1.6648,
      "step": 1999
    },
    {
      "epoch": 1.8484288354898335,
      "grad_norm": 2.15625,
      "learning_rate": 3.1264945556383486e-07,
      "loss": 1.6273,
      "step": 2000
    },
    {
      "epoch": 1.8493530499075785,
      "grad_norm": 2.21875,
      "learning_rate": 3.08867972098823e-07,
      "loss": 1.7627,
      "step": 2001
    },
    {
      "epoch": 1.8502772643253236,
      "grad_norm": 2.203125,
      "learning_rate": 3.0510913771135463e-07,
      "loss": 1.6879,
      "step": 2002
    },
    {
      "epoch": 1.8512014787430684,
      "grad_norm": 2.15625,
      "learning_rate": 3.0137296118617265e-07,
      "loss": 1.6635,
      "step": 2003
    },
    {
      "epoch": 1.8521256931608132,
      "grad_norm": 2.15625,
      "learning_rate": 2.976594512550723e-07,
      "loss": 1.6677,
      "step": 2004
    },
    {
      "epoch": 1.8530499075785583,
      "grad_norm": 2.140625,
      "learning_rate": 2.9396861659686916e-07,
      "loss": 1.672,
      "step": 2005
    },
    {
      "epoch": 1.8539741219963033,
      "grad_norm": 2.171875,
      "learning_rate": 2.9030046583738756e-07,
      "loss": 1.6957,
      "step": 2006
    },
    {
      "epoch": 1.854898336414048,
      "grad_norm": 2.171875,
      "learning_rate": 2.8665500754943787e-07,
      "loss": 1.6912,
      "step": 2007
    },
    {
      "epoch": 1.855822550831793,
      "grad_norm": 2.109375,
      "learning_rate": 2.830322502527971e-07,
      "loss": 1.6495,
      "step": 2008
    },
    {
      "epoch": 1.8567467652495377,
      "grad_norm": 2.1875,
      "learning_rate": 2.7943220241418376e-07,
      "loss": 1.7346,
      "step": 2009
    },
    {
      "epoch": 1.8576709796672828,
      "grad_norm": 2.125,
      "learning_rate": 2.758548724472465e-07,
      "loss": 1.6525,
      "step": 2010
    },
    {
      "epoch": 1.8585951940850278,
      "grad_norm": 2.234375,
      "learning_rate": 2.7230026871253533e-07,
      "loss": 1.7678,
      "step": 2011
    },
    {
      "epoch": 1.8595194085027726,
      "grad_norm": 2.15625,
      "learning_rate": 2.68768399517495e-07,
      "loss": 1.709,
      "step": 2012
    },
    {
      "epoch": 1.8604436229205175,
      "grad_norm": 2.15625,
      "learning_rate": 2.652592731164294e-07,
      "loss": 1.6727,
      "step": 2013
    },
    {
      "epoch": 1.8613678373382625,
      "grad_norm": 2.109375,
      "learning_rate": 2.6177289771049274e-07,
      "loss": 1.635,
      "step": 2014
    },
    {
      "epoch": 1.8622920517560075,
      "grad_norm": 2.171875,
      "learning_rate": 2.5830928144767065e-07,
      "loss": 1.6493,
      "step": 2015
    },
    {
      "epoch": 1.8632162661737524,
      "grad_norm": 2.140625,
      "learning_rate": 2.548684324227557e-07,
      "loss": 1.7734,
      "step": 2016
    },
    {
      "epoch": 1.8641404805914972,
      "grad_norm": 2.171875,
      "learning_rate": 2.5145035867733315e-07,
      "loss": 1.6968,
      "step": 2017
    },
    {
      "epoch": 1.865064695009242,
      "grad_norm": 2.203125,
      "learning_rate": 2.4805506819975955e-07,
      "loss": 1.7287,
      "step": 2018
    },
    {
      "epoch": 1.865988909426987,
      "grad_norm": 2.140625,
      "learning_rate": 2.446825689251442e-07,
      "loss": 1.7558,
      "step": 2019
    },
    {
      "epoch": 1.866913123844732,
      "grad_norm": 2.125,
      "learning_rate": 2.4133286873533114e-07,
      "loss": 1.6396,
      "step": 2020
    },
    {
      "epoch": 1.8678373382624769,
      "grad_norm": 2.125,
      "learning_rate": 2.380059754588826e-07,
      "loss": 1.6737,
      "step": 2021
    },
    {
      "epoch": 1.8687615526802217,
      "grad_norm": 2.125,
      "learning_rate": 2.347018968710568e-07,
      "loss": 1.6379,
      "step": 2022
    },
    {
      "epoch": 1.8696857670979667,
      "grad_norm": 2.109375,
      "learning_rate": 2.3142064069379466e-07,
      "loss": 1.5698,
      "step": 2023
    },
    {
      "epoch": 1.8706099815157118,
      "grad_norm": 2.140625,
      "learning_rate": 2.281622145956952e-07,
      "loss": 1.6499,
      "step": 2024
    },
    {
      "epoch": 1.8715341959334566,
      "grad_norm": 2.140625,
      "learning_rate": 2.2492662619200578e-07,
      "loss": 1.6965,
      "step": 2025
    },
    {
      "epoch": 1.8724584103512014,
      "grad_norm": 2.125,
      "learning_rate": 2.217138830445942e-07,
      "loss": 1.7385,
      "step": 2026
    },
    {
      "epoch": 1.8733826247689462,
      "grad_norm": 2.140625,
      "learning_rate": 2.1852399266194312e-07,
      "loss": 1.6848,
      "step": 2027
    },
    {
      "epoch": 1.8743068391866913,
      "grad_norm": 2.140625,
      "learning_rate": 2.1535696249912362e-07,
      "loss": 1.672,
      "step": 2028
    },
    {
      "epoch": 1.8752310536044363,
      "grad_norm": 2.1875,
      "learning_rate": 2.1221279995777833e-07,
      "loss": 1.6331,
      "step": 2029
    },
    {
      "epoch": 1.8761552680221811,
      "grad_norm": 2.1875,
      "learning_rate": 2.0909151238610703e-07,
      "loss": 1.7452,
      "step": 2030
    },
    {
      "epoch": 1.877079482439926,
      "grad_norm": 2.203125,
      "learning_rate": 2.0599310707885234e-07,
      "loss": 1.7548,
      "step": 2031
    },
    {
      "epoch": 1.878003696857671,
      "grad_norm": 2.125,
      "learning_rate": 2.0291759127727294e-07,
      "loss": 1.6543,
      "step": 2032
    },
    {
      "epoch": 1.878927911275416,
      "grad_norm": 2.140625,
      "learning_rate": 1.9986497216913702e-07,
      "loss": 1.6178,
      "step": 2033
    },
    {
      "epoch": 1.8798521256931608,
      "grad_norm": 2.171875,
      "learning_rate": 1.9683525688869776e-07,
      "loss": 1.6636,
      "step": 2034
    },
    {
      "epoch": 1.8807763401109057,
      "grad_norm": 2.125,
      "learning_rate": 1.9382845251668335e-07,
      "loss": 1.699,
      "step": 2035
    },
    {
      "epoch": 1.8817005545286505,
      "grad_norm": 2.140625,
      "learning_rate": 1.90844566080276e-07,
      "loss": 1.7252,
      "step": 2036
    },
    {
      "epoch": 1.8826247689463955,
      "grad_norm": 2.203125,
      "learning_rate": 1.8788360455309519e-07,
      "loss": 1.7678,
      "step": 2037
    },
    {
      "epoch": 1.8835489833641406,
      "grad_norm": 2.109375,
      "learning_rate": 1.8494557485518427e-07,
      "loss": 1.6532,
      "step": 2038
    },
    {
      "epoch": 1.8844731977818854,
      "grad_norm": 2.25,
      "learning_rate": 1.8203048385299183e-07,
      "loss": 1.7097,
      "step": 2039
    },
    {
      "epoch": 1.8853974121996302,
      "grad_norm": 2.15625,
      "learning_rate": 1.791383383593548e-07,
      "loss": 1.728,
      "step": 2040
    },
    {
      "epoch": 1.8863216266173752,
      "grad_norm": 2.15625,
      "learning_rate": 1.7626914513348858e-07,
      "loss": 1.7252,
      "step": 2041
    },
    {
      "epoch": 1.8872458410351203,
      "grad_norm": 2.09375,
      "learning_rate": 1.7342291088096484e-07,
      "loss": 1.6997,
      "step": 2042
    },
    {
      "epoch": 1.888170055452865,
      "grad_norm": 2.203125,
      "learning_rate": 1.7059964225369707e-07,
      "loss": 1.7681,
      "step": 2043
    },
    {
      "epoch": 1.88909426987061,
      "grad_norm": 2.109375,
      "learning_rate": 1.677993458499272e-07,
      "loss": 1.5783,
      "step": 2044
    },
    {
      "epoch": 1.890018484288355,
      "grad_norm": 2.15625,
      "learning_rate": 1.6502202821420677e-07,
      "loss": 1.7424,
      "step": 2045
    },
    {
      "epoch": 1.8909426987060998,
      "grad_norm": 2.125,
      "learning_rate": 1.6226769583738923e-07,
      "loss": 1.6392,
      "step": 2046
    },
    {
      "epoch": 1.8918669131238448,
      "grad_norm": 2.234375,
      "learning_rate": 1.5953635515660427e-07,
      "loss": 1.6274,
      "step": 2047
    },
    {
      "epoch": 1.8927911275415896,
      "grad_norm": 2.203125,
      "learning_rate": 1.5682801255524793e-07,
      "loss": 1.6437,
      "step": 2048
    },
    {
      "epoch": 1.8937153419593344,
      "grad_norm": 2.140625,
      "learning_rate": 1.5414267436297036e-07,
      "loss": 1.7061,
      "step": 2049
    },
    {
      "epoch": 1.8946395563770795,
      "grad_norm": 2.125,
      "learning_rate": 1.5148034685565473e-07,
      "loss": 1.6462,
      "step": 2050
    },
    {
      "epoch": 1.8955637707948245,
      "grad_norm": 2.140625,
      "learning_rate": 1.4884103625541156e-07,
      "loss": 1.7009,
      "step": 2051
    },
    {
      "epoch": 1.8964879852125693,
      "grad_norm": 2.125,
      "learning_rate": 1.462247487305535e-07,
      "loss": 1.6567,
      "step": 2052
    },
    {
      "epoch": 1.8974121996303142,
      "grad_norm": 2.171875,
      "learning_rate": 1.436314903955871e-07,
      "loss": 1.7031,
      "step": 2053
    },
    {
      "epoch": 1.8983364140480592,
      "grad_norm": 2.25,
      "learning_rate": 1.4106126731119997e-07,
      "loss": 1.7494,
      "step": 2054
    },
    {
      "epoch": 1.899260628465804,
      "grad_norm": 2.21875,
      "learning_rate": 1.385140854842426e-07,
      "loss": 1.6983,
      "step": 2055
    },
    {
      "epoch": 1.900184842883549,
      "grad_norm": 2.171875,
      "learning_rate": 1.359899508677154e-07,
      "loss": 1.6875,
      "step": 2056
    },
    {
      "epoch": 1.9011090573012939,
      "grad_norm": 2.140625,
      "learning_rate": 1.33488869360755e-07,
      "loss": 1.6581,
      "step": 2057
    },
    {
      "epoch": 1.9020332717190387,
      "grad_norm": 2.25,
      "learning_rate": 1.3101084680862464e-07,
      "loss": 1.6706,
      "step": 2058
    },
    {
      "epoch": 1.9029574861367837,
      "grad_norm": 2.1875,
      "learning_rate": 1.2855588900269057e-07,
      "loss": 1.6729,
      "step": 2059
    },
    {
      "epoch": 1.9038817005545288,
      "grad_norm": 2.1875,
      "learning_rate": 1.2612400168041773e-07,
      "loss": 1.6942,
      "step": 2060
    },
    {
      "epoch": 1.9048059149722736,
      "grad_norm": 2.125,
      "learning_rate": 1.237151905253564e-07,
      "loss": 1.6305,
      "step": 2061
    },
    {
      "epoch": 1.9057301293900184,
      "grad_norm": 2.15625,
      "learning_rate": 1.2132946116711896e-07,
      "loss": 1.6687,
      "step": 2062
    },
    {
      "epoch": 1.9066543438077634,
      "grad_norm": 2.21875,
      "learning_rate": 1.1896681918137865e-07,
      "loss": 1.7085,
      "step": 2063
    },
    {
      "epoch": 1.9075785582255085,
      "grad_norm": 2.125,
      "learning_rate": 1.1662727008984964e-07,
      "loss": 1.6132,
      "step": 2064
    },
    {
      "epoch": 1.9085027726432533,
      "grad_norm": 2.15625,
      "learning_rate": 1.1431081936027599e-07,
      "loss": 1.6911,
      "step": 2065
    },
    {
      "epoch": 1.9094269870609981,
      "grad_norm": 2.140625,
      "learning_rate": 1.1201747240641825e-07,
      "loss": 1.725,
      "step": 2066
    },
    {
      "epoch": 1.910351201478743,
      "grad_norm": 2.125,
      "learning_rate": 1.0974723458804127e-07,
      "loss": 1.6656,
      "step": 2067
    },
    {
      "epoch": 1.911275415896488,
      "grad_norm": 2.171875,
      "learning_rate": 1.0750011121090309e-07,
      "loss": 1.6886,
      "step": 2068
    },
    {
      "epoch": 1.912199630314233,
      "grad_norm": 2.15625,
      "learning_rate": 1.0527610752673944e-07,
      "loss": 1.7052,
      "step": 2069
    },
    {
      "epoch": 1.9131238447319778,
      "grad_norm": 2.109375,
      "learning_rate": 1.030752287332537e-07,
      "loss": 1.6537,
      "step": 2070
    },
    {
      "epoch": 1.9140480591497226,
      "grad_norm": 2.171875,
      "learning_rate": 1.0089747997410582e-07,
      "loss": 1.6729,
      "step": 2071
    },
    {
      "epoch": 1.9149722735674677,
      "grad_norm": 2.09375,
      "learning_rate": 9.874286633889562e-08,
      "loss": 1.6629,
      "step": 2072
    },
    {
      "epoch": 1.9158964879852127,
      "grad_norm": 2.09375,
      "learning_rate": 9.661139286315735e-08,
      "loss": 1.6588,
      "step": 2073
    },
    {
      "epoch": 1.9168207024029575,
      "grad_norm": 2.125,
      "learning_rate": 9.45030645283418e-08,
      "loss": 1.635,
      "step": 2074
    },
    {
      "epoch": 1.9177449168207024,
      "grad_norm": 2.171875,
      "learning_rate": 9.24178862618097e-08,
      "loss": 1.6955,
      "step": 2075
    },
    {
      "epoch": 1.9186691312384472,
      "grad_norm": 2.109375,
      "learning_rate": 9.035586293681509e-08,
      "loss": 1.6826,
      "step": 2076
    },
    {
      "epoch": 1.9195933456561922,
      "grad_norm": 2.109375,
      "learning_rate": 8.83169993724986e-08,
      "loss": 1.6287,
      "step": 2077
    },
    {
      "epoch": 1.9205175600739373,
      "grad_norm": 2.171875,
      "learning_rate": 8.63013003338753e-08,
      "loss": 1.7415,
      "step": 2078
    },
    {
      "epoch": 1.921441774491682,
      "grad_norm": 2.125,
      "learning_rate": 8.43087705318213e-08,
      "loss": 1.6821,
      "step": 2079
    },
    {
      "epoch": 1.922365988909427,
      "grad_norm": 2.234375,
      "learning_rate": 8.233941462306272e-08,
      "loss": 1.6868,
      "step": 2080
    },
    {
      "epoch": 1.923290203327172,
      "grad_norm": 2.171875,
      "learning_rate": 8.039323721017011e-08,
      "loss": 1.6659,
      "step": 2081
    },
    {
      "epoch": 1.924214417744917,
      "grad_norm": 2.171875,
      "learning_rate": 7.847024284153959e-08,
      "loss": 1.723,
      "step": 2082
    },
    {
      "epoch": 1.9251386321626618,
      "grad_norm": 2.125,
      "learning_rate": 7.657043601138835e-08,
      "loss": 1.6447,
      "step": 2083
    },
    {
      "epoch": 1.9260628465804066,
      "grad_norm": 2.171875,
      "learning_rate": 7.469382115974034e-08,
      "loss": 1.7519,
      "step": 2084
    },
    {
      "epoch": 1.9269870609981514,
      "grad_norm": 2.171875,
      "learning_rate": 7.284040267242054e-08,
      "loss": 1.6459,
      "step": 2085
    },
    {
      "epoch": 1.9279112754158965,
      "grad_norm": 2.109375,
      "learning_rate": 7.101018488104073e-08,
      "loss": 1.6554,
      "step": 2086
    },
    {
      "epoch": 1.9288354898336415,
      "grad_norm": 2.109375,
      "learning_rate": 6.920317206299043e-08,
      "loss": 1.6471,
      "step": 2087
    },
    {
      "epoch": 1.9297597042513863,
      "grad_norm": 2.109375,
      "learning_rate": 6.741936844142594e-08,
      "loss": 1.683,
      "step": 2088
    },
    {
      "epoch": 1.9306839186691311,
      "grad_norm": 2.265625,
      "learning_rate": 6.565877818526245e-08,
      "loss": 1.6961,
      "step": 2089
    },
    {
      "epoch": 1.9316081330868762,
      "grad_norm": 2.140625,
      "learning_rate": 6.392140540916303e-08,
      "loss": 1.7916,
      "step": 2090
    },
    {
      "epoch": 1.9325323475046212,
      "grad_norm": 2.1875,
      "learning_rate": 6.22072541735308e-08,
      "loss": 1.7443,
      "step": 2091
    },
    {
      "epoch": 1.933456561922366,
      "grad_norm": 2.171875,
      "learning_rate": 6.051632848449563e-08,
      "loss": 1.685,
      "step": 2092
    },
    {
      "epoch": 1.9343807763401109,
      "grad_norm": 2.21875,
      "learning_rate": 5.884863229390747e-08,
      "loss": 1.7679,
      "step": 2093
    },
    {
      "epoch": 1.9353049907578557,
      "grad_norm": 2.140625,
      "learning_rate": 5.720416949932861e-08,
      "loss": 1.6701,
      "step": 2094
    },
    {
      "epoch": 1.9362292051756007,
      "grad_norm": 2.15625,
      "learning_rate": 5.558294394402253e-08,
      "loss": 1.6348,
      "step": 2095
    },
    {
      "epoch": 1.9371534195933457,
      "grad_norm": 2.21875,
      "learning_rate": 5.3984959416942817e-08,
      "loss": 1.6908,
      "step": 2096
    },
    {
      "epoch": 1.9380776340110906,
      "grad_norm": 2.125,
      "learning_rate": 5.2410219652729854e-08,
      "loss": 1.6585,
      "step": 2097
    },
    {
      "epoch": 1.9390018484288354,
      "grad_norm": 2.125,
      "learning_rate": 5.085872833169414e-08,
      "loss": 1.6678,
      "step": 2098
    },
    {
      "epoch": 1.9399260628465804,
      "grad_norm": 2.125,
      "learning_rate": 4.933048907981741e-08,
      "loss": 1.5337,
      "step": 2099
    },
    {
      "epoch": 1.9408502772643255,
      "grad_norm": 2.140625,
      "learning_rate": 4.782550546873599e-08,
      "loss": 1.7536,
      "step": 2100
    },
    {
      "epoch": 1.9417744916820703,
      "grad_norm": 2.140625,
      "learning_rate": 4.6343781015738555e-08,
      "loss": 1.6557,
      "step": 2101
    },
    {
      "epoch": 1.942698706099815,
      "grad_norm": 2.171875,
      "learning_rate": 4.488531918375061e-08,
      "loss": 1.718,
      "step": 2102
    },
    {
      "epoch": 1.94362292051756,
      "grad_norm": 2.125,
      "learning_rate": 4.3450123381334476e-08,
      "loss": 1.6587,
      "step": 2103
    },
    {
      "epoch": 1.944547134935305,
      "grad_norm": 2.1875,
      "learning_rate": 4.203819696267486e-08,
      "loss": 1.647,
      "step": 2104
    },
    {
      "epoch": 1.94547134935305,
      "grad_norm": 2.109375,
      "learning_rate": 4.064954322757775e-08,
      "loss": 1.6272,
      "step": 2105
    },
    {
      "epoch": 1.9463955637707948,
      "grad_norm": 2.328125,
      "learning_rate": 3.9284165421454854e-08,
      "loss": 1.6002,
      "step": 2106
    },
    {
      "epoch": 1.9473197781885396,
      "grad_norm": 2.171875,
      "learning_rate": 3.7942066735321416e-08,
      "loss": 1.7355,
      "step": 2107
    },
    {
      "epoch": 1.9482439926062847,
      "grad_norm": 2.15625,
      "learning_rate": 3.6623250305786177e-08,
      "loss": 1.7497,
      "step": 2108
    },
    {
      "epoch": 1.9491682070240297,
      "grad_norm": 2.140625,
      "learning_rate": 3.5327719215046965e-08,
      "loss": 1.6461,
      "step": 2109
    },
    {
      "epoch": 1.9500924214417745,
      "grad_norm": 2.15625,
      "learning_rate": 3.405547649087959e-08,
      "loss": 1.6912,
      "step": 2110
    },
    {
      "epoch": 1.9510166358595193,
      "grad_norm": 2.140625,
      "learning_rate": 3.2806525106634514e-08,
      "loss": 1.6582,
      "step": 2111
    },
    {
      "epoch": 1.9519408502772642,
      "grad_norm": 2.234375,
      "learning_rate": 3.158086798122684e-08,
      "loss": 1.6888,
      "step": 2112
    },
    {
      "epoch": 1.9528650646950092,
      "grad_norm": 2.125,
      "learning_rate": 3.0378507979131886e-08,
      "loss": 1.6608,
      "step": 2113
    },
    {
      "epoch": 1.9537892791127542,
      "grad_norm": 2.15625,
      "learning_rate": 2.919944791037632e-08,
      "loss": 1.686,
      "step": 2114
    },
    {
      "epoch": 1.954713493530499,
      "grad_norm": 2.203125,
      "learning_rate": 2.8043690530532575e-08,
      "loss": 1.6597,
      "step": 2115
    },
    {
      "epoch": 1.9556377079482439,
      "grad_norm": 2.140625,
      "learning_rate": 2.6911238540715534e-08,
      "loss": 1.6216,
      "step": 2116
    },
    {
      "epoch": 1.956561922365989,
      "grad_norm": 2.171875,
      "learning_rate": 2.5802094587569215e-08,
      "loss": 1.7193,
      "step": 2117
    },
    {
      "epoch": 1.957486136783734,
      "grad_norm": 2.109375,
      "learning_rate": 2.471626126326565e-08,
      "loss": 1.7053,
      "step": 2118
    },
    {
      "epoch": 1.9584103512014788,
      "grad_norm": 2.15625,
      "learning_rate": 2.3653741105499338e-08,
      "loss": 1.6858,
      "step": 2119
    },
    {
      "epoch": 1.9593345656192236,
      "grad_norm": 2.140625,
      "learning_rate": 2.2614536597479474e-08,
      "loss": 1.6537,
      "step": 2120
    },
    {
      "epoch": 1.9602587800369686,
      "grad_norm": 2.109375,
      "learning_rate": 2.1598650167923284e-08,
      "loss": 1.6271,
      "step": 2121
    },
    {
      "epoch": 1.9611829944547134,
      "grad_norm": 2.140625,
      "learning_rate": 2.0606084191050478e-08,
      "loss": 1.7058,
      "step": 2122
    },
    {
      "epoch": 1.9621072088724585,
      "grad_norm": 2.125,
      "learning_rate": 1.963684098658103e-08,
      "loss": 1.7061,
      "step": 2123
    },
    {
      "epoch": 1.9630314232902033,
      "grad_norm": 2.21875,
      "learning_rate": 1.8690922819727398e-08,
      "loss": 1.6824,
      "step": 2124
    },
    {
      "epoch": 1.9639556377079481,
      "grad_norm": 2.09375,
      "learning_rate": 1.7768331901187873e-08,
      "loss": 1.5908,
      "step": 2125
    },
    {
      "epoch": 1.9648798521256932,
      "grad_norm": 2.171875,
      "learning_rate": 1.686907038714325e-08,
      "loss": 1.6579,
      "step": 2126
    },
    {
      "epoch": 1.9658040665434382,
      "grad_norm": 2.09375,
      "learning_rate": 1.599314037925015e-08,
      "loss": 1.6318,
      "step": 2127
    },
    {
      "epoch": 1.966728280961183,
      "grad_norm": 2.109375,
      "learning_rate": 1.514054392463993e-08,
      "loss": 1.6805,
      "step": 2128
    },
    {
      "epoch": 1.9676524953789278,
      "grad_norm": 2.140625,
      "learning_rate": 1.4311283015910893e-08,
      "loss": 1.7008,
      "step": 2129
    },
    {
      "epoch": 1.9685767097966729,
      "grad_norm": 2.171875,
      "learning_rate": 1.3505359591120537e-08,
      "loss": 1.6749,
      "step": 2130
    },
    {
      "epoch": 1.9695009242144177,
      "grad_norm": 2.171875,
      "learning_rate": 1.2722775533787758e-08,
      "loss": 1.692,
      "step": 2131
    },
    {
      "epoch": 1.9704251386321627,
      "grad_norm": 2.171875,
      "learning_rate": 1.1963532672885081e-08,
      "loss": 1.6907,
      "step": 2132
    },
    {
      "epoch": 1.9713493530499075,
      "grad_norm": 2.15625,
      "learning_rate": 1.122763278283312e-08,
      "loss": 1.7365,
      "step": 2133
    },
    {
      "epoch": 1.9722735674676524,
      "grad_norm": 2.15625,
      "learning_rate": 1.0515077583498346e-08,
      "loss": 1.7309,
      "step": 2134
    },
    {
      "epoch": 1.9731977818853974,
      "grad_norm": 2.171875,
      "learning_rate": 9.825868740188649e-09,
      "loss": 1.6991,
      "step": 2135
    },
    {
      "epoch": 1.9741219963031424,
      "grad_norm": 2.109375,
      "learning_rate": 9.160007863650011e-09,
      "loss": 1.6802,
      "step": 2136
    },
    {
      "epoch": 1.9750462107208873,
      "grad_norm": 2.1875,
      "learning_rate": 8.517496510059842e-09,
      "loss": 1.6736,
      "step": 2137
    },
    {
      "epoch": 1.975970425138632,
      "grad_norm": 2.109375,
      "learning_rate": 7.898336181028088e-09,
      "loss": 1.6542,
      "step": 2138
    },
    {
      "epoch": 1.9768946395563771,
      "grad_norm": 2.203125,
      "learning_rate": 7.302528323589464e-09,
      "loss": 1.7204,
      "step": 2139
    },
    {
      "epoch": 1.9778188539741222,
      "grad_norm": 2.15625,
      "learning_rate": 6.730074330203451e-09,
      "loss": 1.6704,
      "step": 2140
    },
    {
      "epoch": 1.978743068391867,
      "grad_norm": 2.125,
      "learning_rate": 6.180975538746525e-09,
      "loss": 1.7195,
      "step": 2141
    },
    {
      "epoch": 1.9796672828096118,
      "grad_norm": 2.140625,
      "learning_rate": 5.655233232515489e-09,
      "loss": 1.6639,
      "step": 2142
    },
    {
      "epoch": 1.9805914972273566,
      "grad_norm": 2.109375,
      "learning_rate": 5.152848640218588e-09,
      "loss": 1.6826,
      "step": 2143
    },
    {
      "epoch": 1.9815157116451017,
      "grad_norm": 2.15625,
      "learning_rate": 4.673822935973293e-09,
      "loss": 1.6826,
      "step": 2144
    },
    {
      "epoch": 1.9824399260628467,
      "grad_norm": 2.109375,
      "learning_rate": 4.2181572393107385e-09,
      "loss": 1.5905,
      "step": 2145
    },
    {
      "epoch": 1.9833641404805915,
      "grad_norm": 2.359375,
      "learning_rate": 3.785852615161289e-09,
      "loss": 1.7413,
      "step": 2146
    },
    {
      "epoch": 1.9842883548983363,
      "grad_norm": 2.09375,
      "learning_rate": 3.376910073862316e-09,
      "loss": 1.6407,
      "step": 2147
    },
    {
      "epoch": 1.9852125693160814,
      "grad_norm": 2.15625,
      "learning_rate": 2.9913305711504193e-09,
      "loss": 1.6535,
      "step": 2148
    },
    {
      "epoch": 1.9861367837338264,
      "grad_norm": 2.140625,
      "learning_rate": 2.629115008160321e-09,
      "loss": 1.6787,
      "step": 2149
    },
    {
      "epoch": 1.9870609981515712,
      "grad_norm": 2.15625,
      "learning_rate": 2.2902642314248657e-09,
      "loss": 1.6507,
      "step": 2150
    },
    {
      "epoch": 1.987985212569316,
      "grad_norm": 2.140625,
      "learning_rate": 1.974779032868357e-09,
      "loss": 1.6353,
      "step": 2151
    },
    {
      "epoch": 1.9889094269870609,
      "grad_norm": 2.125,
      "learning_rate": 1.6826601498098894e-09,
      "loss": 1.6357,
      "step": 2152
    },
    {
      "epoch": 1.989833641404806,
      "grad_norm": 2.171875,
      "learning_rate": 1.4139082649577974e-09,
      "loss": 1.7587,
      "step": 2153
    },
    {
      "epoch": 1.990757855822551,
      "grad_norm": 2.140625,
      "learning_rate": 1.168524006410765e-09,
      "loss": 1.6195,
      "step": 2154
    },
    {
      "epoch": 1.9916820702402958,
      "grad_norm": 2.140625,
      "learning_rate": 9.46507947655606e-10,
      "loss": 1.6284,
      "step": 2155
    },
    {
      "epoch": 1.9926062846580406,
      "grad_norm": 2.15625,
      "learning_rate": 7.478606075617123e-10,
      "loss": 1.7714,
      "step": 2156
    },
    {
      "epoch": 1.9935304990757856,
      "grad_norm": 2.171875,
      "learning_rate": 5.725824503888255e-10,
      "loss": 1.695,
      "step": 2157
    },
    {
      "epoch": 1.9944547134935307,
      "grad_norm": 2.265625,
      "learning_rate": 4.206738857781556e-10,
      "loss": 1.7907,
      "step": 2158
    },
    {
      "epoch": 1.9953789279112755,
      "grad_norm": 2.203125,
      "learning_rate": 2.9213526875349063e-10,
      "loss": 1.6692,
      "step": 2159
    },
    {
      "epoch": 1.9963031423290203,
      "grad_norm": 2.140625,
      "learning_rate": 1.869668997211971e-10,
      "loss": 1.6791,
      "step": 2160
    },
    {
      "epoch": 1.997227356746765,
      "grad_norm": 2.125,
      "learning_rate": 1.0516902447132993e-10,
      "loss": 1.6287,
      "step": 2161
    },
    {
      "epoch": 1.9981515711645101,
      "grad_norm": 2.21875,
      "learning_rate": 4.674183416986111e-11,
      "loss": 1.6804,
      "step": 2162
    },
    {
      "epoch": 1.9990757855822552,
      "grad_norm": 2.28125,
      "learning_rate": 1.1685465369781768e-11,
      "loss": 1.7482,
      "step": 2163
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.453125,
      "learning_rate": 0.0,
      "loss": 1.739,
      "step": 2164
    }
  ],
  "logging_steps": 1,
  "max_steps": 2164,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.810202513587569e+17,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}