{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 511,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0019569471624266144,
      "grad_norm": 0.66796875,
      "learning_rate": 9.980430528375734e-06,
      "loss": 1.4461,
      "step": 1
    },
    {
      "epoch": 0.003913894324853229,
      "grad_norm": 0.6796875,
      "learning_rate": 9.960861056751468e-06,
      "loss": 1.3661,
      "step": 2
    },
    {
      "epoch": 0.005870841487279843,
      "grad_norm": 0.59765625,
      "learning_rate": 9.941291585127202e-06,
      "loss": 1.4274,
      "step": 3
    },
    {
      "epoch": 0.007827788649706457,
      "grad_norm": 0.486328125,
      "learning_rate": 9.921722113502935e-06,
      "loss": 1.3709,
      "step": 4
    },
    {
      "epoch": 0.009784735812133072,
      "grad_norm": 0.4609375,
      "learning_rate": 9.902152641878669e-06,
      "loss": 1.2909,
      "step": 5
    },
    {
      "epoch": 0.011741682974559686,
      "grad_norm": 0.466796875,
      "learning_rate": 9.882583170254404e-06,
      "loss": 1.4032,
      "step": 6
    },
    {
      "epoch": 0.0136986301369863,
      "grad_norm": 0.443359375,
      "learning_rate": 9.863013698630138e-06,
      "loss": 1.3758,
      "step": 7
    },
    {
      "epoch": 0.015655577299412915,
      "grad_norm": 0.37109375,
      "learning_rate": 9.843444227005872e-06,
      "loss": 1.2739,
      "step": 8
    },
    {
      "epoch": 0.01761252446183953,
      "grad_norm": 0.3359375,
      "learning_rate": 9.823874755381605e-06,
      "loss": 1.1982,
      "step": 9
    },
    {
      "epoch": 0.019569471624266144,
      "grad_norm": 0.328125,
      "learning_rate": 9.804305283757339e-06,
      "loss": 1.2455,
      "step": 10
    },
    {
      "epoch": 0.021526418786692758,
      "grad_norm": 0.330078125,
      "learning_rate": 9.784735812133073e-06,
      "loss": 1.2607,
      "step": 11
    },
    {
      "epoch": 0.023483365949119372,
      "grad_norm": 0.392578125,
      "learning_rate": 9.765166340508806e-06,
      "loss": 1.246,
      "step": 12
    },
    {
      "epoch": 0.025440313111545987,
      "grad_norm": 0.30078125,
      "learning_rate": 9.74559686888454e-06,
      "loss": 1.3415,
      "step": 13
    },
    {
      "epoch": 0.0273972602739726,
      "grad_norm": 0.33203125,
      "learning_rate": 9.726027397260275e-06,
      "loss": 1.2426,
      "step": 14
    },
    {
      "epoch": 0.029354207436399216,
      "grad_norm": 0.28125,
      "learning_rate": 9.706457925636007e-06,
      "loss": 1.1957,
      "step": 15
    },
    {
      "epoch": 0.03131115459882583,
      "grad_norm": 0.240234375,
      "learning_rate": 9.686888454011743e-06,
      "loss": 1.196,
      "step": 16
    },
    {
      "epoch": 0.033268101761252444,
      "grad_norm": 0.3359375,
      "learning_rate": 9.667318982387476e-06,
      "loss": 1.2096,
      "step": 17
    },
    {
      "epoch": 0.03522504892367906,
      "grad_norm": 0.2353515625,
      "learning_rate": 9.64774951076321e-06,
      "loss": 1.2594,
      "step": 18
    },
    {
      "epoch": 0.03718199608610567,
      "grad_norm": 0.228515625,
      "learning_rate": 9.628180039138944e-06,
      "loss": 1.169,
      "step": 19
    },
    {
      "epoch": 0.03913894324853229,
      "grad_norm": 0.2060546875,
      "learning_rate": 9.608610567514677e-06,
      "loss": 1.1443,
      "step": 20
    },
    {
      "epoch": 0.0410958904109589,
      "grad_norm": 0.2421875,
      "learning_rate": 9.589041095890411e-06,
      "loss": 1.1179,
      "step": 21
    },
    {
      "epoch": 0.043052837573385516,
      "grad_norm": 0.212890625,
      "learning_rate": 9.569471624266146e-06,
      "loss": 1.1259,
      "step": 22
    },
    {
      "epoch": 0.04500978473581213,
      "grad_norm": 0.201171875,
      "learning_rate": 9.549902152641878e-06,
      "loss": 1.154,
      "step": 23
    },
    {
      "epoch": 0.046966731898238745,
      "grad_norm": 0.2158203125,
      "learning_rate": 9.530332681017614e-06,
      "loss": 1.1163,
      "step": 24
    },
    {
      "epoch": 0.04892367906066536,
      "grad_norm": 0.21484375,
      "learning_rate": 9.510763209393347e-06,
      "loss": 1.0888,
      "step": 25
    },
    {
      "epoch": 0.050880626223091974,
      "grad_norm": 0.2109375,
      "learning_rate": 9.49119373776908e-06,
      "loss": 1.0933,
      "step": 26
    },
    {
      "epoch": 0.05283757338551859,
      "grad_norm": 0.2119140625,
      "learning_rate": 9.471624266144814e-06,
      "loss": 1.1238,
      "step": 27
    },
    {
      "epoch": 0.0547945205479452,
      "grad_norm": 0.1845703125,
      "learning_rate": 9.452054794520548e-06,
      "loss": 1.1178,
      "step": 28
    },
    {
      "epoch": 0.05675146771037182,
      "grad_norm": 0.19921875,
      "learning_rate": 9.432485322896282e-06,
      "loss": 1.1092,
      "step": 29
    },
    {
      "epoch": 0.05870841487279843,
      "grad_norm": 0.20703125,
      "learning_rate": 9.412915851272017e-06,
      "loss": 1.1798,
      "step": 30
    },
    {
      "epoch": 0.060665362035225046,
      "grad_norm": 0.189453125,
      "learning_rate": 9.393346379647749e-06,
      "loss": 1.0471,
      "step": 31
    },
    {
      "epoch": 0.06262230919765166,
      "grad_norm": 0.177734375,
      "learning_rate": 9.373776908023484e-06,
      "loss": 1.1234,
      "step": 32
    },
    {
      "epoch": 0.06457925636007827,
      "grad_norm": 0.1591796875,
      "learning_rate": 9.354207436399218e-06,
      "loss": 0.9983,
      "step": 33
    },
    {
      "epoch": 0.06653620352250489,
      "grad_norm": 0.1474609375,
      "learning_rate": 9.334637964774952e-06,
      "loss": 1.0824,
      "step": 34
    },
    {
      "epoch": 0.0684931506849315,
      "grad_norm": 0.169921875,
      "learning_rate": 9.315068493150685e-06,
      "loss": 1.0707,
      "step": 35
    },
    {
      "epoch": 0.07045009784735812,
      "grad_norm": 0.1533203125,
      "learning_rate": 9.295499021526419e-06,
      "loss": 1.0503,
      "step": 36
    },
    {
      "epoch": 0.07240704500978473,
      "grad_norm": 0.142578125,
      "learning_rate": 9.275929549902153e-06,
      "loss": 1.0387,
      "step": 37
    },
    {
      "epoch": 0.07436399217221135,
      "grad_norm": 0.158203125,
      "learning_rate": 9.256360078277888e-06,
      "loss": 1.0149,
      "step": 38
    },
    {
      "epoch": 0.07632093933463796,
      "grad_norm": 0.173828125,
      "learning_rate": 9.23679060665362e-06,
      "loss": 1.0664,
      "step": 39
    },
    {
      "epoch": 0.07827788649706457,
      "grad_norm": 0.1630859375,
      "learning_rate": 9.217221135029355e-06,
      "loss": 0.9682,
      "step": 40
    },
    {
      "epoch": 0.08023483365949119,
      "grad_norm": 0.1923828125,
      "learning_rate": 9.197651663405089e-06,
      "loss": 1.1108,
      "step": 41
    },
    {
      "epoch": 0.0821917808219178,
      "grad_norm": 0.16015625,
      "learning_rate": 9.178082191780823e-06,
      "loss": 1.0633,
      "step": 42
    },
    {
      "epoch": 0.08414872798434442,
      "grad_norm": 0.2060546875,
      "learning_rate": 9.158512720156556e-06,
      "loss": 0.9949,
      "step": 43
    },
    {
      "epoch": 0.08610567514677103,
      "grad_norm": 0.1376953125,
      "learning_rate": 9.13894324853229e-06,
      "loss": 0.9986,
      "step": 44
    },
    {
      "epoch": 0.08806262230919765,
      "grad_norm": 0.1953125,
      "learning_rate": 9.119373776908024e-06,
      "loss": 0.9542,
      "step": 45
    },
    {
      "epoch": 0.09001956947162426,
      "grad_norm": 0.1591796875,
      "learning_rate": 9.099804305283759e-06,
      "loss": 0.9736,
      "step": 46
    },
    {
      "epoch": 0.09197651663405088,
      "grad_norm": 0.16015625,
      "learning_rate": 9.080234833659491e-06,
      "loss": 1.0057,
      "step": 47
    },
    {
      "epoch": 0.09393346379647749,
      "grad_norm": 0.279296875,
      "learning_rate": 9.060665362035226e-06,
      "loss": 0.9697,
      "step": 48
    },
    {
      "epoch": 0.0958904109589041,
      "grad_norm": 0.1533203125,
      "learning_rate": 9.04109589041096e-06,
      "loss": 0.9534,
      "step": 49
    },
    {
      "epoch": 0.09784735812133072,
      "grad_norm": 0.14453125,
      "learning_rate": 9.021526418786694e-06,
      "loss": 0.9397,
      "step": 50
    },
    {
      "epoch": 0.09980430528375733,
      "grad_norm": 0.1416015625,
      "learning_rate": 9.001956947162427e-06,
      "loss": 0.9926,
      "step": 51
    },
    {
      "epoch": 0.10176125244618395,
      "grad_norm": 0.1455078125,
      "learning_rate": 8.982387475538161e-06,
      "loss": 0.9833,
      "step": 52
    },
    {
      "epoch": 0.10371819960861056,
      "grad_norm": 0.146484375,
      "learning_rate": 8.962818003913895e-06,
      "loss": 1.0001,
      "step": 53
    },
    {
      "epoch": 0.10567514677103718,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.943248532289628e-06,
      "loss": 1.0054,
      "step": 54
    },
    {
      "epoch": 0.10763209393346379,
      "grad_norm": 0.1923828125,
      "learning_rate": 8.923679060665362e-06,
      "loss": 0.8902,
      "step": 55
    },
    {
      "epoch": 0.1095890410958904,
      "grad_norm": 0.15625,
      "learning_rate": 8.904109589041097e-06,
      "loss": 0.9753,
      "step": 56
    },
    {
      "epoch": 0.11154598825831702,
      "grad_norm": 0.1416015625,
      "learning_rate": 8.88454011741683e-06,
      "loss": 0.9192,
      "step": 57
    },
    {
      "epoch": 0.11350293542074363,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.864970645792564e-06,
      "loss": 0.9731,
      "step": 58
    },
    {
      "epoch": 0.11545988258317025,
      "grad_norm": 0.140625,
      "learning_rate": 8.845401174168298e-06,
      "loss": 0.993,
      "step": 59
    },
    {
      "epoch": 0.11741682974559686,
      "grad_norm": 0.1171875,
      "learning_rate": 8.825831702544032e-06,
      "loss": 1.013,
      "step": 60
    },
    {
      "epoch": 0.11937377690802348,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.806262230919765e-06,
      "loss": 0.9423,
      "step": 61
    },
    {
      "epoch": 0.12133072407045009,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.786692759295499e-06,
      "loss": 0.9794,
      "step": 62
    },
    {
      "epoch": 0.1232876712328767,
      "grad_norm": 0.11572265625,
      "learning_rate": 8.767123287671233e-06,
      "loss": 0.9225,
      "step": 63
    },
    {
      "epoch": 0.12524461839530332,
      "grad_norm": 0.12158203125,
      "learning_rate": 8.747553816046968e-06,
      "loss": 0.9053,
      "step": 64
    },
    {
      "epoch": 0.12720156555772993,
      "grad_norm": 0.11767578125,
      "learning_rate": 8.7279843444227e-06,
      "loss": 0.9814,
      "step": 65
    },
    {
      "epoch": 0.12915851272015655,
      "grad_norm": 0.12890625,
      "learning_rate": 8.708414872798435e-06,
      "loss": 0.9611,
      "step": 66
    },
    {
      "epoch": 0.13111545988258316,
      "grad_norm": 0.12353515625,
      "learning_rate": 8.688845401174169e-06,
      "loss": 0.9405,
      "step": 67
    },
    {
      "epoch": 0.13307240704500978,
      "grad_norm": 0.146484375,
      "learning_rate": 8.669275929549903e-06,
      "loss": 0.8448,
      "step": 68
    },
    {
      "epoch": 0.1350293542074364,
      "grad_norm": 0.1171875,
      "learning_rate": 8.649706457925636e-06,
      "loss": 0.959,
      "step": 69
    },
    {
      "epoch": 0.136986301369863,
      "grad_norm": 0.130859375,
      "learning_rate": 8.63013698630137e-06,
      "loss": 0.978,
      "step": 70
    },
    {
      "epoch": 0.13894324853228962,
      "grad_norm": 0.12255859375,
      "learning_rate": 8.610567514677104e-06,
      "loss": 0.9405,
      "step": 71
    },
    {
      "epoch": 0.14090019569471623,
      "grad_norm": 0.11865234375,
      "learning_rate": 8.590998043052839e-06,
      "loss": 0.9356,
      "step": 72
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 0.1435546875,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.903,
      "step": 73
    },
    {
      "epoch": 0.14481409001956946,
      "grad_norm": 0.1337890625,
      "learning_rate": 8.551859099804306e-06,
      "loss": 0.9432,
      "step": 74
    },
    {
      "epoch": 0.14677103718199608,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.53228962818004e-06,
      "loss": 0.9193,
      "step": 75
    },
    {
      "epoch": 0.1487279843444227,
      "grad_norm": 0.1337890625,
      "learning_rate": 8.512720156555774e-06,
      "loss": 0.9166,
      "step": 76
    },
    {
      "epoch": 0.1506849315068493,
      "grad_norm": 0.1279296875,
      "learning_rate": 8.493150684931507e-06,
      "loss": 0.9898,
      "step": 77
    },
    {
      "epoch": 0.15264187866927592,
      "grad_norm": 0.15234375,
      "learning_rate": 8.473581213307241e-06,
      "loss": 0.8916,
      "step": 78
    },
    {
      "epoch": 0.15459882583170254,
      "grad_norm": 0.1533203125,
      "learning_rate": 8.454011741682975e-06,
      "loss": 0.8989,
      "step": 79
    },
    {
      "epoch": 0.15655577299412915,
      "grad_norm": 0.1181640625,
      "learning_rate": 8.43444227005871e-06,
      "loss": 0.8994,
      "step": 80
    },
    {
      "epoch": 0.15851272015655576,
      "grad_norm": 0.119140625,
      "learning_rate": 8.414872798434442e-06,
      "loss": 0.9294,
      "step": 81
    },
    {
      "epoch": 0.16046966731898238,
      "grad_norm": 0.123046875,
      "learning_rate": 8.395303326810177e-06,
      "loss": 0.9239,
      "step": 82
    },
    {
      "epoch": 0.162426614481409,
      "grad_norm": 0.234375,
      "learning_rate": 8.37573385518591e-06,
      "loss": 0.9128,
      "step": 83
    },
    {
      "epoch": 0.1643835616438356,
      "grad_norm": 0.1396484375,
      "learning_rate": 8.356164383561644e-06,
      "loss": 0.8694,
      "step": 84
    },
    {
      "epoch": 0.16634050880626222,
      "grad_norm": 0.1279296875,
      "learning_rate": 8.336594911937378e-06,
      "loss": 0.8914,
      "step": 85
    },
    {
      "epoch": 0.16829745596868884,
      "grad_norm": 0.107421875,
      "learning_rate": 8.317025440313112e-06,
      "loss": 0.9794,
      "step": 86
    },
    {
      "epoch": 0.17025440313111545,
      "grad_norm": 0.119140625,
      "learning_rate": 8.297455968688845e-06,
      "loss": 0.8922,
      "step": 87
    },
    {
      "epoch": 0.17221135029354206,
      "grad_norm": 0.12451171875,
      "learning_rate": 8.27788649706458e-06,
      "loss": 0.9167,
      "step": 88
    },
    {
      "epoch": 0.17416829745596868,
      "grad_norm": 0.1279296875,
      "learning_rate": 8.258317025440313e-06,
      "loss": 0.8793,
      "step": 89
    },
    {
      "epoch": 0.1761252446183953,
      "grad_norm": 0.11669921875,
      "learning_rate": 8.238747553816048e-06,
      "loss": 0.9054,
      "step": 90
    },
    {
      "epoch": 0.1780821917808219,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.219178082191782e-06,
      "loss": 0.9021,
      "step": 91
    },
    {
      "epoch": 0.18003913894324852,
      "grad_norm": 0.12255859375,
      "learning_rate": 8.199608610567515e-06,
      "loss": 0.858,
      "step": 92
    },
    {
      "epoch": 0.18199608610567514,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.180039138943249e-06,
      "loss": 0.9246,
      "step": 93
    },
    {
      "epoch": 0.18395303326810175,
      "grad_norm": 0.11865234375,
      "learning_rate": 8.160469667318983e-06,
      "loss": 0.8801,
      "step": 94
    },
    {
      "epoch": 0.18590998043052837,
      "grad_norm": 0.162109375,
      "learning_rate": 8.140900195694716e-06,
      "loss": 0.8954,
      "step": 95
    },
    {
      "epoch": 0.18786692759295498,
      "grad_norm": 0.1171875,
      "learning_rate": 8.121330724070452e-06,
      "loss": 0.9247,
      "step": 96
    },
    {
      "epoch": 0.1898238747553816,
      "grad_norm": 0.1396484375,
      "learning_rate": 8.101761252446184e-06,
      "loss": 0.8993,
      "step": 97
    },
    {
      "epoch": 0.1917808219178082,
      "grad_norm": 0.1611328125,
      "learning_rate": 8.082191780821919e-06,
      "loss": 0.8513,
      "step": 98
    },
    {
      "epoch": 0.19373776908023482,
      "grad_norm": 0.1201171875,
      "learning_rate": 8.062622309197653e-06,
      "loss": 0.8881,
      "step": 99
    },
    {
      "epoch": 0.19569471624266144,
      "grad_norm": 0.125,
      "learning_rate": 8.043052837573386e-06,
      "loss": 0.8603,
      "step": 100
    },
    {
      "epoch": 0.19765166340508805,
      "grad_norm": 0.11572265625,
      "learning_rate": 8.02348336594912e-06,
      "loss": 0.9218,
      "step": 101
    },
    {
      "epoch": 0.19960861056751467,
      "grad_norm": 0.13671875,
      "learning_rate": 8.003913894324854e-06,
      "loss": 0.8472,
      "step": 102
    },
    {
      "epoch": 0.20156555772994128,
      "grad_norm": 0.1494140625,
      "learning_rate": 7.984344422700587e-06,
      "loss": 0.8988,
      "step": 103
    },
    {
      "epoch": 0.2035225048923679,
      "grad_norm": 0.1220703125,
      "learning_rate": 7.964774951076321e-06,
      "loss": 0.9051,
      "step": 104
    },
    {
      "epoch": 0.2054794520547945,
      "grad_norm": 0.11865234375,
      "learning_rate": 7.945205479452055e-06,
      "loss": 0.8947,
      "step": 105
    },
    {
      "epoch": 0.20743639921722112,
      "grad_norm": 0.12109375,
      "learning_rate": 7.92563600782779e-06,
      "loss": 0.8639,
      "step": 106
    },
    {
      "epoch": 0.20939334637964774,
      "grad_norm": 0.12109375,
      "learning_rate": 7.906066536203524e-06,
      "loss": 0.8954,
      "step": 107
    },
    {
      "epoch": 0.21135029354207435,
      "grad_norm": 0.125,
      "learning_rate": 7.886497064579257e-06,
      "loss": 0.8478,
      "step": 108
    },
    {
      "epoch": 0.21330724070450097,
      "grad_norm": 0.13671875,
      "learning_rate": 7.86692759295499e-06,
      "loss": 0.9215,
      "step": 109
    },
    {
      "epoch": 0.21526418786692758,
      "grad_norm": 0.12060546875,
      "learning_rate": 7.847358121330724e-06,
      "loss": 0.8932,
      "step": 110
    },
    {
      "epoch": 0.2172211350293542,
      "grad_norm": 0.140625,
      "learning_rate": 7.827788649706458e-06,
      "loss": 0.8828,
      "step": 111
    },
    {
      "epoch": 0.2191780821917808,
      "grad_norm": 0.1416015625,
      "learning_rate": 7.808219178082192e-06,
      "loss": 0.9027,
      "step": 112
    },
    {
      "epoch": 0.22113502935420742,
      "grad_norm": 0.1240234375,
      "learning_rate": 7.788649706457925e-06,
      "loss": 0.8689,
      "step": 113
    },
    {
      "epoch": 0.22309197651663404,
      "grad_norm": 0.1357421875,
      "learning_rate": 7.76908023483366e-06,
      "loss": 0.8872,
      "step": 114
    },
    {
      "epoch": 0.22504892367906065,
      "grad_norm": 0.1328125,
      "learning_rate": 7.749510763209393e-06,
      "loss": 0.7915,
      "step": 115
    },
    {
      "epoch": 0.22700587084148727,
      "grad_norm": 0.125,
      "learning_rate": 7.729941291585128e-06,
      "loss": 0.8166,
      "step": 116
    },
    {
      "epoch": 0.22896281800391388,
      "grad_norm": 0.1787109375,
      "learning_rate": 7.710371819960862e-06,
      "loss": 0.8751,
      "step": 117
    },
    {
      "epoch": 0.2309197651663405,
      "grad_norm": 0.1484375,
      "learning_rate": 7.690802348336595e-06,
      "loss": 0.9084,
      "step": 118
    },
    {
      "epoch": 0.2328767123287671,
      "grad_norm": 0.1435546875,
      "learning_rate": 7.671232876712329e-06,
      "loss": 0.8125,
      "step": 119
    },
    {
      "epoch": 0.23483365949119372,
      "grad_norm": 0.1416015625,
      "learning_rate": 7.651663405088063e-06,
      "loss": 0.7863,
      "step": 120
    },
    {
      "epoch": 0.23679060665362034,
      "grad_norm": 0.154296875,
      "learning_rate": 7.632093933463796e-06,
      "loss": 0.8586,
      "step": 121
    },
    {
      "epoch": 0.23874755381604695,
      "grad_norm": 0.1337890625,
      "learning_rate": 7.612524461839531e-06,
      "loss": 0.8224,
      "step": 122
    },
    {
      "epoch": 0.24070450097847357,
      "grad_norm": 0.1337890625,
      "learning_rate": 7.5929549902152645e-06,
      "loss": 0.8387,
      "step": 123
    },
    {
      "epoch": 0.24266144814090018,
      "grad_norm": 0.1181640625,
      "learning_rate": 7.573385518590999e-06,
      "loss": 0.9003,
      "step": 124
    },
    {
      "epoch": 0.2446183953033268,
      "grad_norm": 0.130859375,
      "learning_rate": 7.553816046966732e-06,
      "loss": 0.9102,
      "step": 125
    },
    {
      "epoch": 0.2465753424657534,
      "grad_norm": 0.130859375,
      "learning_rate": 7.534246575342466e-06,
      "loss": 0.8381,
      "step": 126
    },
    {
      "epoch": 0.24853228962818003,
      "grad_norm": 0.11328125,
      "learning_rate": 7.5146771037182e-06,
      "loss": 0.8954,
      "step": 127
    },
    {
      "epoch": 0.25048923679060664,
      "grad_norm": 0.13671875,
      "learning_rate": 7.4951076320939344e-06,
      "loss": 0.8644,
      "step": 128
    },
    {
      "epoch": 0.25244618395303325,
      "grad_norm": 0.125,
      "learning_rate": 7.475538160469667e-06,
      "loss": 0.847,
      "step": 129
    },
    {
      "epoch": 0.25440313111545987,
      "grad_norm": 0.1298828125,
      "learning_rate": 7.455968688845402e-06,
      "loss": 0.8124,
      "step": 130
    },
    {
      "epoch": 0.2563600782778865,
      "grad_norm": 0.1396484375,
      "learning_rate": 7.436399217221135e-06,
      "loss": 0.8837,
      "step": 131
    },
    {
      "epoch": 0.2583170254403131,
      "grad_norm": 0.1259765625,
      "learning_rate": 7.41682974559687e-06,
      "loss": 0.849,
      "step": 132
    },
    {
      "epoch": 0.2602739726027397,
      "grad_norm": 0.123046875,
      "learning_rate": 7.397260273972603e-06,
      "loss": 0.8032,
      "step": 133
    },
    {
      "epoch": 0.2622309197651663,
      "grad_norm": 0.1279296875,
      "learning_rate": 7.377690802348337e-06,
      "loss": 0.908,
      "step": 134
    },
    {
      "epoch": 0.26418786692759294,
      "grad_norm": 0.1572265625,
      "learning_rate": 7.358121330724071e-06,
      "loss": 0.8655,
      "step": 135
    },
    {
      "epoch": 0.26614481409001955,
      "grad_norm": 0.140625,
      "learning_rate": 7.338551859099805e-06,
      "loss": 0.8951,
      "step": 136
    },
    {
      "epoch": 0.26810176125244617,
      "grad_norm": 0.1376953125,
      "learning_rate": 7.318982387475538e-06,
      "loss": 0.8982,
      "step": 137
    },
    {
      "epoch": 0.2700587084148728,
      "grad_norm": 0.126953125,
      "learning_rate": 7.299412915851273e-06,
      "loss": 0.9128,
      "step": 138
    },
    {
      "epoch": 0.2720156555772994,
      "grad_norm": 0.138671875,
      "learning_rate": 7.279843444227006e-06,
      "loss": 0.8041,
      "step": 139
    },
    {
      "epoch": 0.273972602739726,
      "grad_norm": 0.12109375,
      "learning_rate": 7.260273972602741e-06,
      "loss": 0.9096,
      "step": 140
    },
    {
      "epoch": 0.2759295499021526,
      "grad_norm": 0.140625,
      "learning_rate": 7.240704500978474e-06,
      "loss": 0.8573,
      "step": 141
    },
    {
      "epoch": 0.27788649706457924,
      "grad_norm": 0.1318359375,
      "learning_rate": 7.221135029354208e-06,
      "loss": 0.8514,
      "step": 142
    },
    {
      "epoch": 0.27984344422700586,
      "grad_norm": 0.130859375,
      "learning_rate": 7.201565557729942e-06,
      "loss": 0.7995,
      "step": 143
    },
    {
      "epoch": 0.28180039138943247,
      "grad_norm": 0.12353515625,
      "learning_rate": 7.181996086105676e-06,
      "loss": 0.8932,
      "step": 144
    },
    {
      "epoch": 0.2837573385518591,
      "grad_norm": 0.1845703125,
      "learning_rate": 7.162426614481409e-06,
      "loss": 0.9044,
      "step": 145
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.1923828125,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.9009,
      "step": 146
    },
    {
      "epoch": 0.2876712328767123,
      "grad_norm": 0.1318359375,
      "learning_rate": 7.123287671232877e-06,
      "loss": 0.8576,
      "step": 147
    },
    {
      "epoch": 0.2896281800391389,
      "grad_norm": 0.12451171875,
      "learning_rate": 7.103718199608612e-06,
      "loss": 0.8488,
      "step": 148
    },
    {
      "epoch": 0.29158512720156554,
      "grad_norm": 0.1376953125,
      "learning_rate": 7.0841487279843445e-06,
      "loss": 0.8388,
      "step": 149
    },
    {
      "epoch": 0.29354207436399216,
      "grad_norm": 0.1337890625,
      "learning_rate": 7.064579256360079e-06,
      "loss": 0.9016,
      "step": 150
    },
    {
      "epoch": 0.29549902152641877,
      "grad_norm": 0.126953125,
      "learning_rate": 7.045009784735813e-06,
      "loss": 0.8215,
      "step": 151
    },
    {
      "epoch": 0.2974559686888454,
      "grad_norm": 0.130859375,
      "learning_rate": 7.025440313111546e-06,
      "loss": 0.8762,
      "step": 152
    },
    {
      "epoch": 0.299412915851272,
      "grad_norm": 0.1279296875,
      "learning_rate": 7.00587084148728e-06,
      "loss": 0.8513,
      "step": 153
    },
    {
      "epoch": 0.3013698630136986,
      "grad_norm": 0.123046875,
      "learning_rate": 6.9863013698630145e-06,
      "loss": 0.8267,
      "step": 154
    },
    {
      "epoch": 0.30332681017612523,
      "grad_norm": 0.115234375,
      "learning_rate": 6.966731898238748e-06,
      "loss": 0.8657,
      "step": 155
    },
    {
      "epoch": 0.30528375733855184,
      "grad_norm": 0.13671875,
      "learning_rate": 6.947162426614482e-06,
      "loss": 0.8961,
      "step": 156
    },
    {
      "epoch": 0.30724070450097846,
      "grad_norm": 0.126953125,
      "learning_rate": 6.927592954990215e-06,
      "loss": 0.8809,
      "step": 157
    },
    {
      "epoch": 0.30919765166340507,
      "grad_norm": 0.1435546875,
      "learning_rate": 6.90802348336595e-06,
      "loss": 0.8914,
      "step": 158
    },
    {
      "epoch": 0.3111545988258317,
      "grad_norm": 0.12890625,
      "learning_rate": 6.8884540117416836e-06,
      "loss": 0.8824,
      "step": 159
    },
    {
      "epoch": 0.3131115459882583,
      "grad_norm": 0.1416015625,
      "learning_rate": 6.868884540117417e-06,
      "loss": 0.7989,
      "step": 160
    },
    {
      "epoch": 0.3150684931506849,
      "grad_norm": 0.12353515625,
      "learning_rate": 6.849315068493151e-06,
      "loss": 0.8252,
      "step": 161
    },
    {
      "epoch": 0.31702544031311153,
      "grad_norm": 0.138671875,
      "learning_rate": 6.829745596868885e-06,
      "loss": 0.9499,
      "step": 162
    },
    {
      "epoch": 0.31898238747553814,
      "grad_norm": 0.12451171875,
      "learning_rate": 6.810176125244618e-06,
      "loss": 0.8282,
      "step": 163
    },
    {
      "epoch": 0.32093933463796476,
      "grad_norm": 0.21484375,
      "learning_rate": 6.790606653620353e-06,
      "loss": 0.8443,
      "step": 164
    },
    {
      "epoch": 0.32289628180039137,
      "grad_norm": 0.125,
      "learning_rate": 6.771037181996086e-06,
      "loss": 0.862,
      "step": 165
    },
    {
      "epoch": 0.324853228962818,
      "grad_norm": 0.12890625,
      "learning_rate": 6.751467710371821e-06,
      "loss": 0.873,
      "step": 166
    },
    {
      "epoch": 0.3268101761252446,
      "grad_norm": 0.1689453125,
      "learning_rate": 6.731898238747554e-06,
      "loss": 0.7797,
      "step": 167
    },
    {
      "epoch": 0.3287671232876712,
      "grad_norm": 0.1318359375,
      "learning_rate": 6.712328767123288e-06,
      "loss": 0.8868,
      "step": 168
    },
    {
      "epoch": 0.33072407045009783,
      "grad_norm": 0.1337890625,
      "learning_rate": 6.692759295499022e-06,
      "loss": 0.8322,
      "step": 169
    },
    {
      "epoch": 0.33268101761252444,
      "grad_norm": 0.18359375,
      "learning_rate": 6.673189823874756e-06,
      "loss": 0.8408,
      "step": 170
    },
    {
      "epoch": 0.33463796477495106,
      "grad_norm": 0.1357421875,
      "learning_rate": 6.653620352250489e-06,
      "loss": 0.7677,
      "step": 171
    },
    {
      "epoch": 0.33659491193737767,
      "grad_norm": 0.150390625,
      "learning_rate": 6.634050880626224e-06,
      "loss": 0.9103,
      "step": 172
    },
    {
      "epoch": 0.3385518590998043,
      "grad_norm": 0.140625,
      "learning_rate": 6.614481409001957e-06,
      "loss": 0.8327,
      "step": 173
    },
    {
      "epoch": 0.3405088062622309,
      "grad_norm": 0.1328125,
      "learning_rate": 6.594911937377692e-06,
      "loss": 0.8531,
      "step": 174
    },
    {
      "epoch": 0.3424657534246575,
      "grad_norm": 0.138671875,
      "learning_rate": 6.5753424657534245e-06,
      "loss": 0.8673,
      "step": 175
    },
    {
      "epoch": 0.34442270058708413,
      "grad_norm": 0.134765625,
      "learning_rate": 6.555772994129159e-06,
      "loss": 0.828,
      "step": 176
    },
    {
      "epoch": 0.34637964774951074,
      "grad_norm": 0.140625,
      "learning_rate": 6.536203522504893e-06,
      "loss": 0.8719,
      "step": 177
    },
    {
      "epoch": 0.34833659491193736,
      "grad_norm": 0.1279296875,
      "learning_rate": 6.516634050880627e-06,
      "loss": 0.864,
      "step": 178
    },
    {
      "epoch": 0.350293542074364,
      "grad_norm": 0.12890625,
      "learning_rate": 6.49706457925636e-06,
      "loss": 0.8971,
      "step": 179
    },
    {
      "epoch": 0.3522504892367906,
      "grad_norm": 0.1220703125,
      "learning_rate": 6.4774951076320945e-06,
      "loss": 0.8787,
      "step": 180
    },
    {
      "epoch": 0.3542074363992172,
      "grad_norm": 0.1416015625,
      "learning_rate": 6.457925636007828e-06,
      "loss": 0.8259,
      "step": 181
    },
    {
      "epoch": 0.3561643835616438,
      "grad_norm": 0.212890625,
      "learning_rate": 6.438356164383563e-06,
      "loss": 0.8709,
      "step": 182
    },
    {
      "epoch": 0.35812133072407043,
      "grad_norm": 0.12109375,
      "learning_rate": 6.4187866927592954e-06,
      "loss": 0.8821,
      "step": 183
    },
    {
      "epoch": 0.36007827788649704,
      "grad_norm": 0.138671875,
      "learning_rate": 6.39921722113503e-06,
      "loss": 0.8212,
      "step": 184
    },
    {
      "epoch": 0.36203522504892366,
      "grad_norm": 1.5703125,
      "learning_rate": 6.379647749510764e-06,
      "loss": 0.874,
      "step": 185
    },
    {
      "epoch": 0.3639921722113503,
      "grad_norm": 0.23828125,
      "learning_rate": 6.360078277886498e-06,
      "loss": 0.8168,
      "step": 186
    },
    {
      "epoch": 0.3659491193737769,
      "grad_norm": 0.1455078125,
      "learning_rate": 6.340508806262231e-06,
      "loss": 0.8382,
      "step": 187
    },
    {
      "epoch": 0.3679060665362035,
      "grad_norm": 0.1669921875,
      "learning_rate": 6.320939334637965e-06,
      "loss": 0.8943,
      "step": 188
    },
    {
      "epoch": 0.3698630136986301,
      "grad_norm": 0.1298828125,
      "learning_rate": 6.301369863013699e-06,
      "loss": 0.8834,
      "step": 189
    },
    {
      "epoch": 0.37181996086105673,
      "grad_norm": 0.1484375,
      "learning_rate": 6.2818003913894335e-06,
      "loss": 0.8424,
      "step": 190
    },
    {
      "epoch": 0.37377690802348335,
      "grad_norm": 0.138671875,
      "learning_rate": 6.262230919765166e-06,
      "loss": 0.8496,
      "step": 191
    },
    {
      "epoch": 0.37573385518590996,
      "grad_norm": 0.1220703125,
      "learning_rate": 6.242661448140901e-06,
      "loss": 0.8941,
      "step": 192
    },
    {
      "epoch": 0.3776908023483366,
      "grad_norm": 0.1728515625,
      "learning_rate": 6.2230919765166345e-06,
      "loss": 0.8154,
      "step": 193
    },
    {
      "epoch": 0.3796477495107632,
      "grad_norm": 0.1484375,
      "learning_rate": 6.203522504892369e-06,
      "loss": 0.8654,
      "step": 194
    },
    {
      "epoch": 0.3816046966731898,
      "grad_norm": 0.2158203125,
      "learning_rate": 6.183953033268102e-06,
      "loss": 0.848,
      "step": 195
    },
    {
      "epoch": 0.3835616438356164,
      "grad_norm": 0.1416015625,
      "learning_rate": 6.164383561643836e-06,
      "loss": 0.8558,
      "step": 196
    },
    {
      "epoch": 0.38551859099804303,
      "grad_norm": 0.27734375,
      "learning_rate": 6.14481409001957e-06,
      "loss": 0.8601,
      "step": 197
    },
    {
      "epoch": 0.38747553816046965,
      "grad_norm": 0.1201171875,
      "learning_rate": 6.1252446183953044e-06,
      "loss": 0.8833,
      "step": 198
    },
    {
      "epoch": 0.38943248532289626,
      "grad_norm": 0.126953125,
      "learning_rate": 6.105675146771037e-06,
      "loss": 0.852,
      "step": 199
    },
    {
      "epoch": 0.3913894324853229,
      "grad_norm": 0.1298828125,
      "learning_rate": 6.086105675146772e-06,
      "loss": 0.8546,
      "step": 200
    },
    {
      "epoch": 0.3933463796477495,
      "grad_norm": 0.1494140625,
      "learning_rate": 6.066536203522505e-06,
      "loss": 0.9004,
      "step": 201
    },
    {
      "epoch": 0.3953033268101761,
      "grad_norm": 0.1318359375,
      "learning_rate": 6.046966731898239e-06,
      "loss": 0.8638,
      "step": 202
    },
    {
      "epoch": 0.3972602739726027,
      "grad_norm": 0.1328125,
      "learning_rate": 6.027397260273973e-06,
      "loss": 0.8235,
      "step": 203
    },
    {
      "epoch": 0.39921722113502933,
      "grad_norm": 0.14453125,
      "learning_rate": 6.007827788649707e-06,
      "loss": 0.7872,
      "step": 204
    },
    {
      "epoch": 0.40117416829745595,
      "grad_norm": 0.1484375,
      "learning_rate": 5.988258317025441e-06,
      "loss": 0.8676,
      "step": 205
    },
    {
      "epoch": 0.40313111545988256,
      "grad_norm": 0.171875,
      "learning_rate": 5.9686888454011745e-06,
      "loss": 0.8139,
      "step": 206
    },
    {
      "epoch": 0.4050880626223092,
      "grad_norm": 0.126953125,
      "learning_rate": 5.949119373776908e-06,
      "loss": 0.8064,
      "step": 207
    },
    {
      "epoch": 0.4070450097847358,
      "grad_norm": 0.1318359375,
      "learning_rate": 5.929549902152643e-06,
      "loss": 0.8975,
      "step": 208
    },
    {
      "epoch": 0.4090019569471624,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.909980430528376e-06,
      "loss": 0.8096,
      "step": 209
    },
    {
      "epoch": 0.410958904109589,
      "grad_norm": 0.1416015625,
      "learning_rate": 5.89041095890411e-06,
      "loss": 0.7626,
      "step": 210
    },
    {
      "epoch": 0.41291585127201563,
      "grad_norm": 0.138671875,
      "learning_rate": 5.870841487279844e-06,
      "loss": 0.8408,
      "step": 211
    },
    {
      "epoch": 0.41487279843444225,
      "grad_norm": 0.12890625,
      "learning_rate": 5.851272015655578e-06,
      "loss": 0.8423,
      "step": 212
    },
    {
      "epoch": 0.41682974559686886,
      "grad_norm": 0.1513671875,
      "learning_rate": 5.831702544031311e-06,
      "loss": 0.8388,
      "step": 213
    },
    {
      "epoch": 0.4187866927592955,
      "grad_norm": 0.140625,
      "learning_rate": 5.812133072407045e-06,
      "loss": 0.8358,
      "step": 214
    },
    {
      "epoch": 0.4207436399217221,
      "grad_norm": 0.138671875,
      "learning_rate": 5.792563600782779e-06,
      "loss": 0.8598,
      "step": 215
    },
    {
      "epoch": 0.4227005870841487,
      "grad_norm": 0.16796875,
      "learning_rate": 5.7729941291585136e-06,
      "loss": 0.8662,
      "step": 216
    },
    {
      "epoch": 0.4246575342465753,
      "grad_norm": 0.1337890625,
      "learning_rate": 5.753424657534246e-06,
      "loss": 0.8337,
      "step": 217
    },
    {
      "epoch": 0.42661448140900193,
      "grad_norm": 0.140625,
      "learning_rate": 5.733855185909981e-06,
      "loss": 0.7651,
      "step": 218
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 0.1416015625,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.8396,
      "step": 219
    },
    {
      "epoch": 0.43052837573385516,
      "grad_norm": 0.140625,
      "learning_rate": 5.694716242661449e-06,
      "loss": 0.8224,
      "step": 220
    },
    {
      "epoch": 0.4324853228962818,
      "grad_norm": 0.1416015625,
      "learning_rate": 5.675146771037182e-06,
      "loss": 0.8409,
      "step": 221
    },
    {
      "epoch": 0.4344422700587084,
      "grad_norm": 0.1376953125,
      "learning_rate": 5.655577299412916e-06,
      "loss": 0.8605,
      "step": 222
    },
    {
      "epoch": 0.436399217221135,
      "grad_norm": 0.1318359375,
      "learning_rate": 5.63600782778865e-06,
      "loss": 0.908,
      "step": 223
    },
    {
      "epoch": 0.4383561643835616,
      "grad_norm": 0.1474609375,
      "learning_rate": 5.6164383561643845e-06,
      "loss": 0.865,
      "step": 224
    },
    {
      "epoch": 0.44031311154598823,
      "grad_norm": 0.1337890625,
      "learning_rate": 5.596868884540117e-06,
      "loss": 0.7979,
      "step": 225
    },
    {
      "epoch": 0.44227005870841485,
      "grad_norm": 0.2021484375,
      "learning_rate": 5.577299412915852e-06,
      "loss": 0.7827,
      "step": 226
    },
    {
      "epoch": 0.44422700587084146,
      "grad_norm": 0.16015625,
      "learning_rate": 5.557729941291585e-06,
      "loss": 0.8459,
      "step": 227
    },
    {
      "epoch": 0.4461839530332681,
      "grad_norm": 0.1962890625,
      "learning_rate": 5.53816046966732e-06,
      "loss": 0.71,
      "step": 228
    },
    {
      "epoch": 0.4481409001956947,
      "grad_norm": 0.142578125,
      "learning_rate": 5.518590998043053e-06,
      "loss": 0.8219,
      "step": 229
    },
    {
      "epoch": 0.4500978473581213,
      "grad_norm": 0.1279296875,
      "learning_rate": 5.499021526418787e-06,
      "loss": 0.8574,
      "step": 230
    },
    {
      "epoch": 0.4520547945205479,
      "grad_norm": 0.17578125,
      "learning_rate": 5.479452054794521e-06,
      "loss": 0.8048,
      "step": 231
    },
    {
      "epoch": 0.45401174168297453,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.459882583170255e-06,
      "loss": 0.8098,
      "step": 232
    },
    {
      "epoch": 0.45596868884540115,
      "grad_norm": 0.1484375,
      "learning_rate": 5.440313111545988e-06,
      "loss": 0.7238,
      "step": 233
    },
    {
      "epoch": 0.45792563600782776,
      "grad_norm": 0.1328125,
      "learning_rate": 5.420743639921723e-06,
      "loss": 0.815,
      "step": 234
    },
    {
      "epoch": 0.4598825831702544,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.401174168297456e-06,
      "loss": 0.8157,
      "step": 235
    },
    {
      "epoch": 0.461839530332681,
      "grad_norm": 0.1552734375,
      "learning_rate": 5.381604696673191e-06,
      "loss": 0.8267,
      "step": 236
    },
    {
      "epoch": 0.4637964774951076,
      "grad_norm": 0.173828125,
      "learning_rate": 5.362035225048924e-06,
      "loss": 0.8532,
      "step": 237
    },
    {
      "epoch": 0.4657534246575342,
      "grad_norm": 0.1435546875,
      "learning_rate": 5.342465753424658e-06,
      "loss": 0.8053,
      "step": 238
    },
    {
      "epoch": 0.46771037181996084,
      "grad_norm": 0.1328125,
      "learning_rate": 5.322896281800392e-06,
      "loss": 0.7989,
      "step": 239
    },
    {
      "epoch": 0.46966731898238745,
      "grad_norm": 0.1357421875,
      "learning_rate": 5.303326810176126e-06,
      "loss": 0.8279,
      "step": 240
    },
    {
      "epoch": 0.47162426614481406,
      "grad_norm": 0.142578125,
      "learning_rate": 5.283757338551859e-06,
      "loss": 0.8052,
      "step": 241
    },
    {
      "epoch": 0.4735812133072407,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.2641878669275936e-06,
      "loss": 0.8135,
      "step": 242
    },
    {
      "epoch": 0.4755381604696673,
      "grad_norm": 0.1591796875,
      "learning_rate": 5.244618395303327e-06,
      "loss": 0.837,
      "step": 243
    },
    {
      "epoch": 0.4774951076320939,
      "grad_norm": 0.1982421875,
      "learning_rate": 5.225048923679062e-06,
      "loss": 0.8033,
      "step": 244
    },
    {
      "epoch": 0.4794520547945205,
      "grad_norm": 0.140625,
      "learning_rate": 5.2054794520547945e-06,
      "loss": 0.8504,
      "step": 245
    },
    {
      "epoch": 0.48140900195694714,
      "grad_norm": 0.205078125,
      "learning_rate": 5.185909980430529e-06,
      "loss": 0.8732,
      "step": 246
    },
    {
      "epoch": 0.48336594911937375,
      "grad_norm": 0.1337890625,
      "learning_rate": 5.166340508806263e-06,
      "loss": 0.8369,
      "step": 247
    },
    {
      "epoch": 0.48532289628180036,
      "grad_norm": 0.1513671875,
      "learning_rate": 5.146771037181997e-06,
      "loss": 0.815,
      "step": 248
    },
    {
      "epoch": 0.487279843444227,
      "grad_norm": 0.138671875,
      "learning_rate": 5.12720156555773e-06,
      "loss": 0.8367,
      "step": 249
    },
    {
      "epoch": 0.4892367906066536,
      "grad_norm": 0.259765625,
      "learning_rate": 5.1076320939334645e-06,
      "loss": 0.8433,
      "step": 250
    },
    {
      "epoch": 0.4911937377690802,
      "grad_norm": 0.15625,
      "learning_rate": 5.088062622309198e-06,
      "loss": 0.8641,
      "step": 251
    },
    {
      "epoch": 0.4931506849315068,
      "grad_norm": 0.1357421875,
      "learning_rate": 5.068493150684932e-06,
      "loss": 0.8243,
      "step": 252
    },
    {
      "epoch": 0.49510763209393344,
      "grad_norm": 0.134765625,
      "learning_rate": 5.0489236790606654e-06,
      "loss": 0.8296,
      "step": 253
    },
    {
      "epoch": 0.49706457925636005,
      "grad_norm": 0.1533203125,
      "learning_rate": 5.0293542074364e-06,
      "loss": 0.8127,
      "step": 254
    },
    {
      "epoch": 0.49902152641878667,
      "grad_norm": 0.1640625,
      "learning_rate": 5.009784735812134e-06,
      "loss": 0.8214,
      "step": 255
    },
    {
      "epoch": 0.5009784735812133,
      "grad_norm": 0.1376953125,
      "learning_rate": 4.990215264187867e-06,
      "loss": 0.7784,
      "step": 256
    },
    {
      "epoch": 0.50293542074364,
      "grad_norm": 0.123046875,
      "learning_rate": 4.970645792563601e-06,
      "loss": 0.8956,
      "step": 257
    },
    {
      "epoch": 0.5048923679060665,
      "grad_norm": 0.142578125,
      "learning_rate": 4.9510763209393345e-06,
      "loss": 0.7886,
      "step": 258
    },
    {
      "epoch": 0.5068493150684932,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.931506849315069e-06,
      "loss": 0.8058,
      "step": 259
    },
    {
      "epoch": 0.5088062622309197,
      "grad_norm": 0.14453125,
      "learning_rate": 4.911937377690803e-06,
      "loss": 0.8415,
      "step": 260
    },
    {
      "epoch": 0.5107632093933464,
      "grad_norm": 0.15234375,
      "learning_rate": 4.892367906066536e-06,
      "loss": 0.8112,
      "step": 261
    },
    {
      "epoch": 0.512720156555773,
      "grad_norm": 0.1357421875,
      "learning_rate": 4.87279843444227e-06,
      "loss": 0.8293,
      "step": 262
    },
    {
      "epoch": 0.5146771037181996,
      "grad_norm": 0.150390625,
      "learning_rate": 4.853228962818004e-06,
      "loss": 0.8278,
      "step": 263
    },
    {
      "epoch": 0.5166340508806262,
      "grad_norm": 0.1669921875,
      "learning_rate": 4.833659491193738e-06,
      "loss": 0.8316,
      "step": 264
    },
    {
      "epoch": 0.5185909980430529,
      "grad_norm": 0.1396484375,
      "learning_rate": 4.814090019569472e-06,
      "loss": 0.8299,
      "step": 265
    },
    {
      "epoch": 0.5205479452054794,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.7945205479452054e-06,
      "loss": 0.8322,
      "step": 266
    },
    {
      "epoch": 0.5225048923679061,
      "grad_norm": 0.1435546875,
      "learning_rate": 4.774951076320939e-06,
      "loss": 0.7967,
      "step": 267
    },
    {
      "epoch": 0.5244618395303327,
      "grad_norm": 0.1455078125,
      "learning_rate": 4.755381604696674e-06,
      "loss": 0.8079,
      "step": 268
    },
    {
      "epoch": 0.5264187866927593,
      "grad_norm": 0.154296875,
      "learning_rate": 4.735812133072407e-06,
      "loss": 0.8257,
      "step": 269
    },
    {
      "epoch": 0.5283757338551859,
      "grad_norm": 0.1513671875,
      "learning_rate": 4.716242661448141e-06,
      "loss": 0.8509,
      "step": 270
    },
    {
      "epoch": 0.5303326810176126,
      "grad_norm": 0.150390625,
      "learning_rate": 4.6966731898238745e-06,
      "loss": 0.8002,
      "step": 271
    },
    {
      "epoch": 0.5322896281800391,
      "grad_norm": 0.1337890625,
      "learning_rate": 4.677103718199609e-06,
      "loss": 0.8707,
      "step": 272
    },
    {
      "epoch": 0.5342465753424658,
      "grad_norm": 0.130859375,
      "learning_rate": 4.657534246575343e-06,
      "loss": 0.8337,
      "step": 273
    },
    {
      "epoch": 0.5362035225048923,
      "grad_norm": 0.150390625,
      "learning_rate": 4.637964774951076e-06,
      "loss": 0.7674,
      "step": 274
    },
    {
      "epoch": 0.538160469667319,
      "grad_norm": 0.13671875,
      "learning_rate": 4.61839530332681e-06,
      "loss": 0.8229,
      "step": 275
    },
    {
      "epoch": 0.5401174168297456,
      "grad_norm": 0.1494140625,
      "learning_rate": 4.5988258317025445e-06,
      "loss": 0.8068,
      "step": 276
    },
    {
      "epoch": 0.5420743639921722,
      "grad_norm": 0.13671875,
      "learning_rate": 4.579256360078278e-06,
      "loss": 0.817,
      "step": 277
    },
    {
      "epoch": 0.5440313111545988,
      "grad_norm": 0.1220703125,
      "learning_rate": 4.559686888454012e-06,
      "loss": 0.9284,
      "step": 278
    },
    {
      "epoch": 0.5459882583170255,
      "grad_norm": 0.25390625,
      "learning_rate": 4.5401174168297455e-06,
      "loss": 0.8302,
      "step": 279
    },
    {
      "epoch": 0.547945205479452,
      "grad_norm": 0.1533203125,
      "learning_rate": 4.52054794520548e-06,
      "loss": 0.7787,
      "step": 280
    },
    {
      "epoch": 0.5499021526418787,
      "grad_norm": 0.1376953125,
      "learning_rate": 4.500978473581214e-06,
      "loss": 0.8362,
      "step": 281
    },
    {
      "epoch": 0.5518590998043053,
      "grad_norm": 0.2255859375,
      "learning_rate": 4.481409001956947e-06,
      "loss": 0.8446,
      "step": 282
    },
    {
      "epoch": 0.5538160469667319,
      "grad_norm": 0.1484375,
      "learning_rate": 4.461839530332681e-06,
      "loss": 0.7899,
      "step": 283
    },
    {
      "epoch": 0.5557729941291585,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.442270058708415e-06,
      "loss": 0.8018,
      "step": 284
    },
    {
      "epoch": 0.5577299412915852,
      "grad_norm": 0.150390625,
      "learning_rate": 4.422700587084149e-06,
      "loss": 0.8233,
      "step": 285
    },
    {
      "epoch": 0.5596868884540117,
      "grad_norm": 0.12890625,
      "learning_rate": 4.403131115459883e-06,
      "loss": 0.8734,
      "step": 286
    },
    {
      "epoch": 0.5616438356164384,
      "grad_norm": 0.1318359375,
      "learning_rate": 4.383561643835616e-06,
      "loss": 0.837,
      "step": 287
    },
    {
      "epoch": 0.5636007827788649,
      "grad_norm": 0.1416015625,
      "learning_rate": 4.36399217221135e-06,
      "loss": 0.8116,
      "step": 288
    },
    {
      "epoch": 0.5655577299412916,
      "grad_norm": 0.1455078125,
      "learning_rate": 4.3444227005870845e-06,
      "loss": 0.8029,
      "step": 289
    },
    {
      "epoch": 0.5675146771037182,
      "grad_norm": 0.345703125,
      "learning_rate": 4.324853228962818e-06,
      "loss": 0.7608,
      "step": 290
    },
    {
      "epoch": 0.5694716242661448,
      "grad_norm": 0.1416015625,
      "learning_rate": 4.305283757338552e-06,
      "loss": 0.7919,
      "step": 291
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.1533203125,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.7834,
      "step": 292
    },
    {
      "epoch": 0.5733855185909981,
      "grad_norm": 0.1337890625,
      "learning_rate": 4.26614481409002e-06,
      "loss": 0.8632,
      "step": 293
    },
    {
      "epoch": 0.5753424657534246,
      "grad_norm": 0.142578125,
      "learning_rate": 4.246575342465754e-06,
      "loss": 0.8458,
      "step": 294
    },
    {
      "epoch": 0.5772994129158513,
      "grad_norm": 0.138671875,
      "learning_rate": 4.227005870841487e-06,
      "loss": 0.8095,
      "step": 295
    },
    {
      "epoch": 0.5792563600782779,
      "grad_norm": 0.1357421875,
      "learning_rate": 4.207436399217221e-06,
      "loss": 0.7865,
      "step": 296
    },
    {
      "epoch": 0.5812133072407045,
      "grad_norm": 0.14453125,
      "learning_rate": 4.187866927592955e-06,
      "loss": 0.7765,
      "step": 297
    },
    {
      "epoch": 0.5831702544031311,
      "grad_norm": 0.15234375,
      "learning_rate": 4.168297455968689e-06,
      "loss": 0.7958,
      "step": 298
    },
    {
      "epoch": 0.5851272015655578,
      "grad_norm": 0.140625,
      "learning_rate": 4.148727984344423e-06,
      "loss": 0.8319,
      "step": 299
    },
    {
      "epoch": 0.5870841487279843,
      "grad_norm": 0.138671875,
      "learning_rate": 4.129158512720156e-06,
      "loss": 0.7867,
      "step": 300
    },
    {
      "epoch": 0.589041095890411,
      "grad_norm": 0.14453125,
      "learning_rate": 4.109589041095891e-06,
      "loss": 0.89,
      "step": 301
    },
    {
      "epoch": 0.5909980430528375,
      "grad_norm": 0.15234375,
      "learning_rate": 4.0900195694716245e-06,
      "loss": 0.7982,
      "step": 302
    },
    {
      "epoch": 0.5929549902152642,
      "grad_norm": 0.138671875,
      "learning_rate": 4.070450097847358e-06,
      "loss": 0.7748,
      "step": 303
    },
    {
      "epoch": 0.5949119373776908,
      "grad_norm": 0.1484375,
      "learning_rate": 4.050880626223092e-06,
      "loss": 0.7973,
      "step": 304
    },
    {
      "epoch": 0.5968688845401174,
      "grad_norm": 0.193359375,
      "learning_rate": 4.031311154598826e-06,
      "loss": 0.7829,
      "step": 305
    },
    {
      "epoch": 0.598825831702544,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.01174168297456e-06,
      "loss": 0.8327,
      "step": 306
    },
    {
      "epoch": 0.6007827788649707,
      "grad_norm": 0.1630859375,
      "learning_rate": 3.992172211350294e-06,
      "loss": 0.8311,
      "step": 307
    },
    {
      "epoch": 0.6027397260273972,
      "grad_norm": 0.140625,
      "learning_rate": 3.972602739726027e-06,
      "loss": 0.8463,
      "step": 308
    },
    {
      "epoch": 0.6046966731898239,
      "grad_norm": 0.15625,
      "learning_rate": 3.953033268101762e-06,
      "loss": 0.8147,
      "step": 309
    },
    {
      "epoch": 0.6066536203522505,
      "grad_norm": 0.134765625,
      "learning_rate": 3.933463796477495e-06,
      "loss": 0.809,
      "step": 310
    },
    {
      "epoch": 0.6086105675146771,
      "grad_norm": 0.21484375,
      "learning_rate": 3.913894324853229e-06,
      "loss": 0.7249,
      "step": 311
    },
    {
      "epoch": 0.6105675146771037,
      "grad_norm": 0.1640625,
      "learning_rate": 3.894324853228963e-06,
      "loss": 0.7849,
      "step": 312
    },
    {
      "epoch": 0.6125244618395304,
      "grad_norm": 0.1474609375,
      "learning_rate": 3.874755381604696e-06,
      "loss": 0.8829,
      "step": 313
    },
    {
      "epoch": 0.6144814090019569,
      "grad_norm": 0.14453125,
      "learning_rate": 3.855185909980431e-06,
      "loss": 0.816,
      "step": 314
    },
    {
      "epoch": 0.6164383561643836,
      "grad_norm": 0.1298828125,
      "learning_rate": 3.8356164383561645e-06,
      "loss": 0.8552,
      "step": 315
    },
    {
      "epoch": 0.6183953033268101,
      "grad_norm": 0.1357421875,
      "learning_rate": 3.816046966731898e-06,
      "loss": 0.8737,
      "step": 316
    },
    {
      "epoch": 0.6203522504892368,
      "grad_norm": 0.1337890625,
      "learning_rate": 3.7964774951076322e-06,
      "loss": 0.8471,
      "step": 317
    },
    {
      "epoch": 0.6223091976516634,
      "grad_norm": 0.173828125,
      "learning_rate": 3.776908023483366e-06,
      "loss": 0.7774,
      "step": 318
    },
    {
      "epoch": 0.62426614481409,
      "grad_norm": 0.138671875,
      "learning_rate": 3.7573385518591e-06,
      "loss": 0.8058,
      "step": 319
    },
    {
      "epoch": 0.6262230919765166,
      "grad_norm": 0.1689453125,
      "learning_rate": 3.7377690802348336e-06,
      "loss": 0.7918,
      "step": 320
    },
    {
      "epoch": 0.6281800391389433,
      "grad_norm": 0.1591796875,
      "learning_rate": 3.7181996086105677e-06,
      "loss": 0.824,
      "step": 321
    },
    {
      "epoch": 0.6301369863013698,
      "grad_norm": 0.1953125,
      "learning_rate": 3.6986301369863014e-06,
      "loss": 0.8411,
      "step": 322
    },
    {
      "epoch": 0.6320939334637965,
      "grad_norm": 0.1484375,
      "learning_rate": 3.6790606653620354e-06,
      "loss": 0.7759,
      "step": 323
    },
    {
      "epoch": 0.6340508806262231,
      "grad_norm": 0.138671875,
      "learning_rate": 3.659491193737769e-06,
      "loss": 0.8138,
      "step": 324
    },
    {
      "epoch": 0.6360078277886497,
      "grad_norm": 0.17578125,
      "learning_rate": 3.639921722113503e-06,
      "loss": 0.7988,
      "step": 325
    },
    {
      "epoch": 0.6379647749510763,
      "grad_norm": 0.171875,
      "learning_rate": 3.620352250489237e-06,
      "loss": 0.742,
      "step": 326
    },
    {
      "epoch": 0.639921722113503,
      "grad_norm": 0.1396484375,
      "learning_rate": 3.600782778864971e-06,
      "loss": 0.843,
      "step": 327
    },
    {
      "epoch": 0.6418786692759295,
      "grad_norm": 0.1435546875,
      "learning_rate": 3.5812133072407045e-06,
      "loss": 0.7595,
      "step": 328
    },
    {
      "epoch": 0.6438356164383562,
      "grad_norm": 0.142578125,
      "learning_rate": 3.5616438356164386e-06,
      "loss": 0.8492,
      "step": 329
    },
    {
      "epoch": 0.6457925636007827,
      "grad_norm": 0.1376953125,
      "learning_rate": 3.5420743639921723e-06,
      "loss": 0.7913,
      "step": 330
    },
    {
      "epoch": 0.6477495107632094,
      "grad_norm": 0.197265625,
      "learning_rate": 3.5225048923679063e-06,
      "loss": 0.7671,
      "step": 331
    },
    {
      "epoch": 0.649706457925636,
      "grad_norm": 0.1474609375,
      "learning_rate": 3.50293542074364e-06,
      "loss": 0.7292,
      "step": 332
    },
    {
      "epoch": 0.6516634050880626,
      "grad_norm": 0.158203125,
      "learning_rate": 3.483365949119374e-06,
      "loss": 0.8076,
      "step": 333
    },
    {
      "epoch": 0.6536203522504892,
      "grad_norm": 0.1669921875,
      "learning_rate": 3.4637964774951077e-06,
      "loss": 0.7857,
      "step": 334
    },
    {
      "epoch": 0.6555772994129159,
      "grad_norm": 0.1376953125,
      "learning_rate": 3.4442270058708418e-06,
      "loss": 0.7836,
      "step": 335
    },
    {
      "epoch": 0.6575342465753424,
      "grad_norm": 0.146484375,
      "learning_rate": 3.4246575342465754e-06,
      "loss": 0.8336,
      "step": 336
    },
    {
      "epoch": 0.6594911937377691,
      "grad_norm": 0.146484375,
      "learning_rate": 3.405088062622309e-06,
      "loss": 0.8058,
      "step": 337
    },
    {
      "epoch": 0.6614481409001957,
      "grad_norm": 0.1240234375,
      "learning_rate": 3.385518590998043e-06,
      "loss": 0.8918,
      "step": 338
    },
    {
      "epoch": 0.6634050880626223,
      "grad_norm": 0.16015625,
      "learning_rate": 3.365949119373777e-06,
      "loss": 0.7775,
      "step": 339
    },
    {
      "epoch": 0.6653620352250489,
      "grad_norm": 0.1669921875,
      "learning_rate": 3.346379647749511e-06,
      "loss": 0.7522,
      "step": 340
    },
    {
      "epoch": 0.6673189823874756,
      "grad_norm": 0.1650390625,
      "learning_rate": 3.3268101761252445e-06,
      "loss": 0.7652,
      "step": 341
    },
    {
      "epoch": 0.6692759295499021,
      "grad_norm": 0.1376953125,
      "learning_rate": 3.3072407045009786e-06,
      "loss": 0.8224,
      "step": 342
    },
    {
      "epoch": 0.6712328767123288,
      "grad_norm": 0.18359375,
      "learning_rate": 3.2876712328767123e-06,
      "loss": 0.7782,
      "step": 343
    },
    {
      "epoch": 0.6731898238747553,
      "grad_norm": 0.1259765625,
      "learning_rate": 3.2681017612524463e-06,
      "loss": 0.8399,
      "step": 344
    },
    {
      "epoch": 0.675146771037182,
      "grad_norm": 0.181640625,
      "learning_rate": 3.24853228962818e-06,
      "loss": 0.8376,
      "step": 345
    },
    {
      "epoch": 0.6771037181996086,
      "grad_norm": 0.1689453125,
      "learning_rate": 3.228962818003914e-06,
      "loss": 0.8122,
      "step": 346
    },
    {
      "epoch": 0.6790606653620352,
      "grad_norm": 0.1689453125,
      "learning_rate": 3.2093933463796477e-06,
      "loss": 0.8152,
      "step": 347
    },
    {
      "epoch": 0.6810176125244618,
      "grad_norm": 0.15234375,
      "learning_rate": 3.189823874755382e-06,
      "loss": 0.822,
      "step": 348
    },
    {
      "epoch": 0.6829745596868885,
      "grad_norm": 0.1474609375,
      "learning_rate": 3.1702544031311154e-06,
      "loss": 0.773,
      "step": 349
    },
    {
      "epoch": 0.684931506849315,
      "grad_norm": 0.197265625,
      "learning_rate": 3.1506849315068495e-06,
      "loss": 0.6956,
      "step": 350
    },
    {
      "epoch": 0.6868884540117417,
      "grad_norm": 0.14453125,
      "learning_rate": 3.131115459882583e-06,
      "loss": 0.8239,
      "step": 351
    },
    {
      "epoch": 0.6888454011741683,
      "grad_norm": 0.1474609375,
      "learning_rate": 3.1115459882583172e-06,
      "loss": 0.8126,
      "step": 352
    },
    {
      "epoch": 0.6908023483365949,
      "grad_norm": 0.1455078125,
      "learning_rate": 3.091976516634051e-06,
      "loss": 0.8178,
      "step": 353
    },
    {
      "epoch": 0.6927592954990215,
      "grad_norm": 0.1435546875,
      "learning_rate": 3.072407045009785e-06,
      "loss": 0.902,
      "step": 354
    },
    {
      "epoch": 0.6947162426614482,
      "grad_norm": 0.15625,
      "learning_rate": 3.0528375733855186e-06,
      "loss": 0.7964,
      "step": 355
    },
    {
      "epoch": 0.6966731898238747,
      "grad_norm": 0.1494140625,
      "learning_rate": 3.0332681017612527e-06,
      "loss": 0.7898,
      "step": 356
    },
    {
      "epoch": 0.6986301369863014,
      "grad_norm": 0.15234375,
      "learning_rate": 3.0136986301369864e-06,
      "loss": 0.7629,
      "step": 357
    },
    {
      "epoch": 0.700587084148728,
      "grad_norm": 0.142578125,
      "learning_rate": 2.9941291585127204e-06,
      "loss": 0.8343,
      "step": 358
    },
    {
      "epoch": 0.7025440313111546,
      "grad_norm": 0.12890625,
      "learning_rate": 2.974559686888454e-06,
      "loss": 0.872,
      "step": 359
    },
    {
      "epoch": 0.7045009784735812,
      "grad_norm": 0.138671875,
      "learning_rate": 2.954990215264188e-06,
      "loss": 0.8317,
      "step": 360
    },
    {
      "epoch": 0.7064579256360078,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.935420743639922e-06,
      "loss": 0.8259,
      "step": 361
    },
    {
      "epoch": 0.7084148727984344,
      "grad_norm": 0.158203125,
      "learning_rate": 2.9158512720156555e-06,
      "loss": 0.7376,
      "step": 362
    },
    {
      "epoch": 0.7103718199608611,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.8962818003913895e-06,
      "loss": 0.7962,
      "step": 363
    },
    {
      "epoch": 0.7123287671232876,
      "grad_norm": 0.138671875,
      "learning_rate": 2.876712328767123e-06,
      "loss": 0.8058,
      "step": 364
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.15625,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.8273,
      "step": 365
    },
    {
      "epoch": 0.7162426614481409,
      "grad_norm": 0.142578125,
      "learning_rate": 2.837573385518591e-06,
      "loss": 0.7809,
      "step": 366
    },
    {
      "epoch": 0.7181996086105675,
      "grad_norm": 0.138671875,
      "learning_rate": 2.818003913894325e-06,
      "loss": 0.7872,
      "step": 367
    },
    {
      "epoch": 0.7201565557729941,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.7984344422700586e-06,
      "loss": 0.8785,
      "step": 368
    },
    {
      "epoch": 0.7221135029354208,
      "grad_norm": 0.1484375,
      "learning_rate": 2.7788649706457927e-06,
      "loss": 0.8299,
      "step": 369
    },
    {
      "epoch": 0.7240704500978473,
      "grad_norm": 0.146484375,
      "learning_rate": 2.7592954990215264e-06,
      "loss": 0.8026,
      "step": 370
    },
    {
      "epoch": 0.726027397260274,
      "grad_norm": 0.154296875,
      "learning_rate": 2.7397260273972604e-06,
      "loss": 0.7787,
      "step": 371
    },
    {
      "epoch": 0.7279843444227005,
      "grad_norm": 0.173828125,
      "learning_rate": 2.720156555772994e-06,
      "loss": 0.7265,
      "step": 372
    },
    {
      "epoch": 0.7299412915851272,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.700587084148728e-06,
      "loss": 0.7992,
      "step": 373
    },
    {
      "epoch": 0.7318982387475538,
      "grad_norm": 0.1875,
      "learning_rate": 2.681017612524462e-06,
      "loss": 0.7495,
      "step": 374
    },
    {
      "epoch": 0.7338551859099804,
      "grad_norm": 0.150390625,
      "learning_rate": 2.661448140900196e-06,
      "loss": 0.8236,
      "step": 375
    },
    {
      "epoch": 0.735812133072407,
      "grad_norm": 0.13671875,
      "learning_rate": 2.6418786692759295e-06,
      "loss": 0.8181,
      "step": 376
    },
    {
      "epoch": 0.7377690802348337,
      "grad_norm": 0.13671875,
      "learning_rate": 2.6223091976516636e-06,
      "loss": 0.7977,
      "step": 377
    },
    {
      "epoch": 0.7397260273972602,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.6027397260273973e-06,
      "loss": 0.7887,
      "step": 378
    },
    {
      "epoch": 0.7416829745596869,
      "grad_norm": 0.1357421875,
      "learning_rate": 2.5831702544031313e-06,
      "loss": 0.8723,
      "step": 379
    },
    {
      "epoch": 0.7436399217221135,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.563600782778865e-06,
      "loss": 0.8492,
      "step": 380
    },
    {
      "epoch": 0.7455968688845401,
      "grad_norm": 0.1435546875,
      "learning_rate": 2.544031311154599e-06,
      "loss": 0.7904,
      "step": 381
    },
    {
      "epoch": 0.7475538160469667,
      "grad_norm": 0.150390625,
      "learning_rate": 2.5244618395303327e-06,
      "loss": 0.7835,
      "step": 382
    },
    {
      "epoch": 0.7495107632093934,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.504892367906067e-06,
      "loss": 0.86,
      "step": 383
    },
    {
      "epoch": 0.7514677103718199,
      "grad_norm": 0.142578125,
      "learning_rate": 2.4853228962818004e-06,
      "loss": 0.8163,
      "step": 384
    },
    {
      "epoch": 0.7534246575342466,
      "grad_norm": 0.1416015625,
      "learning_rate": 2.4657534246575345e-06,
      "loss": 0.7965,
      "step": 385
    },
    {
      "epoch": 0.7553816046966731,
      "grad_norm": 0.13671875,
      "learning_rate": 2.446183953033268e-06,
      "loss": 0.8441,
      "step": 386
    },
    {
      "epoch": 0.7573385518590998,
      "grad_norm": 0.13671875,
      "learning_rate": 2.426614481409002e-06,
      "loss": 0.784,
      "step": 387
    },
    {
      "epoch": 0.7592954990215264,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.407045009784736e-06,
      "loss": 0.8536,
      "step": 388
    },
    {
      "epoch": 0.761252446183953,
      "grad_norm": 0.1328125,
      "learning_rate": 2.3874755381604695e-06,
      "loss": 0.9018,
      "step": 389
    },
    {
      "epoch": 0.7632093933463796,
      "grad_norm": 0.1298828125,
      "learning_rate": 2.3679060665362036e-06,
      "loss": 0.8552,
      "step": 390
    },
    {
      "epoch": 0.7651663405088063,
      "grad_norm": 0.1689453125,
      "learning_rate": 2.3483365949119373e-06,
      "loss": 0.7223,
      "step": 391
    },
    {
      "epoch": 0.7671232876712328,
      "grad_norm": 0.1494140625,
      "learning_rate": 2.3287671232876713e-06,
      "loss": 0.8057,
      "step": 392
    },
    {
      "epoch": 0.7690802348336595,
      "grad_norm": 0.16796875,
      "learning_rate": 2.309197651663405e-06,
      "loss": 0.8436,
      "step": 393
    },
    {
      "epoch": 0.7710371819960861,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.289628180039139e-06,
      "loss": 0.7871,
      "step": 394
    },
    {
      "epoch": 0.7729941291585127,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.2700587084148727e-06,
      "loss": 0.8207,
      "step": 395
    },
    {
      "epoch": 0.7749510763209393,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.250489236790607e-06,
      "loss": 0.8627,
      "step": 396
    },
    {
      "epoch": 0.776908023483366,
      "grad_norm": 0.1337890625,
      "learning_rate": 2.2309197651663405e-06,
      "loss": 0.8278,
      "step": 397
    },
    {
      "epoch": 0.7788649706457925,
      "grad_norm": 0.1357421875,
      "learning_rate": 2.2113502935420745e-06,
      "loss": 0.8097,
      "step": 398
    },
    {
      "epoch": 0.7808219178082192,
      "grad_norm": 0.1474609375,
      "learning_rate": 2.191780821917808e-06,
      "loss": 0.7998,
      "step": 399
    },
    {
      "epoch": 0.7827788649706457,
      "grad_norm": 0.146484375,
      "learning_rate": 2.1722113502935423e-06,
      "loss": 0.8661,
      "step": 400
    },
    {
      "epoch": 0.7847358121330724,
      "grad_norm": 0.158203125,
      "learning_rate": 2.152641878669276e-06,
      "loss": 0.8593,
      "step": 401
    },
    {
      "epoch": 0.786692759295499,
      "grad_norm": 0.1513671875,
      "learning_rate": 2.13307240704501e-06,
      "loss": 0.7968,
      "step": 402
    },
    {
      "epoch": 0.7886497064579256,
      "grad_norm": 0.130859375,
      "learning_rate": 2.1135029354207436e-06,
      "loss": 0.8502,
      "step": 403
    },
    {
      "epoch": 0.7906066536203522,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.0939334637964777e-06,
      "loss": 0.8185,
      "step": 404
    },
    {
      "epoch": 0.7925636007827789,
      "grad_norm": 0.181640625,
      "learning_rate": 2.0743639921722114e-06,
      "loss": 0.768,
      "step": 405
    },
    {
      "epoch": 0.7945205479452054,
      "grad_norm": 0.1484375,
      "learning_rate": 2.0547945205479454e-06,
      "loss": 0.857,
      "step": 406
    },
    {
      "epoch": 0.7964774951076321,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.035225048923679e-06,
      "loss": 0.7867,
      "step": 407
    },
    {
      "epoch": 0.7984344422700587,
      "grad_norm": 0.146484375,
      "learning_rate": 2.015655577299413e-06,
      "loss": 0.8031,
      "step": 408
    },
    {
      "epoch": 0.8003913894324853,
      "grad_norm": 0.15625,
      "learning_rate": 1.996086105675147e-06,
      "loss": 0.8207,
      "step": 409
    },
    {
      "epoch": 0.8023483365949119,
      "grad_norm": 0.134765625,
      "learning_rate": 1.976516634050881e-06,
      "loss": 0.8736,
      "step": 410
    },
    {
      "epoch": 0.8043052837573386,
      "grad_norm": 0.1455078125,
      "learning_rate": 1.9569471624266145e-06,
      "loss": 0.7846,
      "step": 411
    },
    {
      "epoch": 0.8062622309197651,
      "grad_norm": 0.1455078125,
      "learning_rate": 1.937377690802348e-06,
      "loss": 0.7179,
      "step": 412
    },
    {
      "epoch": 0.8082191780821918,
      "grad_norm": 0.146484375,
      "learning_rate": 1.9178082191780823e-06,
      "loss": 0.8121,
      "step": 413
    },
    {
      "epoch": 0.8101761252446184,
      "grad_norm": 0.1416015625,
      "learning_rate": 1.8982387475538161e-06,
      "loss": 0.8256,
      "step": 414
    },
    {
      "epoch": 0.812133072407045,
      "grad_norm": 0.15234375,
      "learning_rate": 1.87866927592955e-06,
      "loss": 0.8072,
      "step": 415
    },
    {
      "epoch": 0.8140900195694716,
      "grad_norm": 0.1455078125,
      "learning_rate": 1.8590998043052839e-06,
      "loss": 0.7735,
      "step": 416
    },
    {
      "epoch": 0.8160469667318982,
      "grad_norm": 0.171875,
      "learning_rate": 1.8395303326810177e-06,
      "loss": 0.8447,
      "step": 417
    },
    {
      "epoch": 0.8180039138943248,
      "grad_norm": 0.30859375,
      "learning_rate": 1.8199608610567516e-06,
      "loss": 0.724,
      "step": 418
    },
    {
      "epoch": 0.8199608610567515,
      "grad_norm": 0.158203125,
      "learning_rate": 1.8003913894324854e-06,
      "loss": 0.8001,
      "step": 419
    },
    {
      "epoch": 0.821917808219178,
      "grad_norm": 0.150390625,
      "learning_rate": 1.7808219178082193e-06,
      "loss": 0.7854,
      "step": 420
    },
    {
      "epoch": 0.8238747553816047,
      "grad_norm": 0.1630859375,
      "learning_rate": 1.7612524461839532e-06,
      "loss": 0.8543,
      "step": 421
    },
    {
      "epoch": 0.8258317025440313,
      "grad_norm": 0.142578125,
      "learning_rate": 1.741682974559687e-06,
      "loss": 0.8381,
      "step": 422
    },
    {
      "epoch": 0.8277886497064579,
      "grad_norm": 0.150390625,
      "learning_rate": 1.7221135029354209e-06,
      "loss": 0.8048,
      "step": 423
    },
    {
      "epoch": 0.8297455968688845,
      "grad_norm": 0.142578125,
      "learning_rate": 1.7025440313111545e-06,
      "loss": 0.8001,
      "step": 424
    },
    {
      "epoch": 0.8317025440313112,
      "grad_norm": 0.1376953125,
      "learning_rate": 1.6829745596868884e-06,
      "loss": 0.8394,
      "step": 425
    },
    {
      "epoch": 0.8336594911937377,
      "grad_norm": 0.1513671875,
      "learning_rate": 1.6634050880626223e-06,
      "loss": 0.8285,
      "step": 426
    },
    {
      "epoch": 0.8356164383561644,
      "grad_norm": 0.1455078125,
      "learning_rate": 1.6438356164383561e-06,
      "loss": 0.7556,
      "step": 427
    },
    {
      "epoch": 0.837573385518591,
      "grad_norm": 0.18359375,
      "learning_rate": 1.62426614481409e-06,
      "loss": 0.8053,
      "step": 428
    },
    {
      "epoch": 0.8395303326810176,
      "grad_norm": 0.150390625,
      "learning_rate": 1.6046966731898239e-06,
      "loss": 0.8353,
      "step": 429
    },
    {
      "epoch": 0.8414872798434442,
      "grad_norm": 0.1376953125,
      "learning_rate": 1.5851272015655577e-06,
      "loss": 0.8326,
      "step": 430
    },
    {
      "epoch": 0.8434442270058709,
      "grad_norm": 0.1376953125,
      "learning_rate": 1.5655577299412916e-06,
      "loss": 0.7238,
      "step": 431
    },
    {
      "epoch": 0.8454011741682974,
      "grad_norm": 0.1416015625,
      "learning_rate": 1.5459882583170254e-06,
      "loss": 0.8071,
      "step": 432
    },
    {
      "epoch": 0.8473581213307241,
      "grad_norm": 0.1376953125,
      "learning_rate": 1.5264187866927593e-06,
      "loss": 0.7282,
      "step": 433
    },
    {
      "epoch": 0.8493150684931506,
      "grad_norm": 0.150390625,
      "learning_rate": 1.5068493150684932e-06,
      "loss": 0.7484,
      "step": 434
    },
    {
      "epoch": 0.8512720156555773,
      "grad_norm": 0.1416015625,
      "learning_rate": 1.487279843444227e-06,
      "loss": 0.8348,
      "step": 435
    },
    {
      "epoch": 0.8532289628180039,
      "grad_norm": 0.15625,
      "learning_rate": 1.467710371819961e-06,
      "loss": 0.8239,
      "step": 436
    },
    {
      "epoch": 0.8551859099804305,
      "grad_norm": 0.1455078125,
      "learning_rate": 1.4481409001956948e-06,
      "loss": 0.7993,
      "step": 437
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.1474609375,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.8169,
      "step": 438
    },
    {
      "epoch": 0.8590998043052838,
      "grad_norm": 0.15625,
      "learning_rate": 1.4090019569471625e-06,
      "loss": 0.7944,
      "step": 439
    },
    {
      "epoch": 0.8610567514677103,
      "grad_norm": 0.150390625,
      "learning_rate": 1.3894324853228964e-06,
      "loss": 0.8219,
      "step": 440
    },
    {
      "epoch": 0.863013698630137,
      "grad_norm": 0.1611328125,
      "learning_rate": 1.3698630136986302e-06,
      "loss": 0.8306,
      "step": 441
    },
    {
      "epoch": 0.8649706457925636,
      "grad_norm": 0.14453125,
      "learning_rate": 1.350293542074364e-06,
      "loss": 0.8336,
      "step": 442
    },
    {
      "epoch": 0.8669275929549902,
      "grad_norm": 0.16015625,
      "learning_rate": 1.330724070450098e-06,
      "loss": 0.8226,
      "step": 443
    },
    {
      "epoch": 0.8688845401174168,
      "grad_norm": 0.146484375,
      "learning_rate": 1.3111545988258318e-06,
      "loss": 0.8407,
      "step": 444
    },
    {
      "epoch": 0.8708414872798435,
      "grad_norm": 0.138671875,
      "learning_rate": 1.2915851272015657e-06,
      "loss": 0.8366,
      "step": 445
    },
    {
      "epoch": 0.87279843444227,
      "grad_norm": 0.1865234375,
      "learning_rate": 1.2720156555772995e-06,
      "loss": 0.6976,
      "step": 446
    },
    {
      "epoch": 0.8747553816046967,
      "grad_norm": 0.130859375,
      "learning_rate": 1.2524461839530334e-06,
      "loss": 0.8289,
      "step": 447
    },
    {
      "epoch": 0.8767123287671232,
      "grad_norm": 0.13671875,
      "learning_rate": 1.2328767123287673e-06,
      "loss": 0.8101,
      "step": 448
    },
    {
      "epoch": 0.8786692759295499,
      "grad_norm": 0.138671875,
      "learning_rate": 1.213307240704501e-06,
      "loss": 0.8111,
      "step": 449
    },
    {
      "epoch": 0.8806262230919765,
      "grad_norm": 0.146484375,
      "learning_rate": 1.1937377690802348e-06,
      "loss": 0.8277,
      "step": 450
    },
    {
      "epoch": 0.8825831702544031,
      "grad_norm": 0.1337890625,
      "learning_rate": 1.1741682974559686e-06,
      "loss": 0.8005,
      "step": 451
    },
    {
      "epoch": 0.8845401174168297,
      "grad_norm": 0.1376953125,
      "learning_rate": 1.1545988258317025e-06,
      "loss": 0.8484,
      "step": 452
    },
    {
      "epoch": 0.8864970645792564,
      "grad_norm": 0.1455078125,
      "learning_rate": 1.1350293542074364e-06,
      "loss": 0.818,
      "step": 453
    },
    {
      "epoch": 0.8884540117416829,
      "grad_norm": 0.1796875,
      "learning_rate": 1.1154598825831702e-06,
      "loss": 0.8746,
      "step": 454
    },
    {
      "epoch": 0.8904109589041096,
      "grad_norm": 0.1416015625,
      "learning_rate": 1.095890410958904e-06,
      "loss": 0.8444,
      "step": 455
    },
    {
      "epoch": 0.8923679060665362,
      "grad_norm": 0.142578125,
      "learning_rate": 1.076320939334638e-06,
      "loss": 0.7859,
      "step": 456
    },
    {
      "epoch": 0.8943248532289628,
      "grad_norm": 0.1552734375,
      "learning_rate": 1.0567514677103718e-06,
      "loss": 0.7849,
      "step": 457
    },
    {
      "epoch": 0.8962818003913894,
      "grad_norm": 0.2158203125,
      "learning_rate": 1.0371819960861057e-06,
      "loss": 0.7859,
      "step": 458
    },
    {
      "epoch": 0.898238747553816,
      "grad_norm": 0.146484375,
      "learning_rate": 1.0176125244618395e-06,
      "loss": 0.8224,
      "step": 459
    },
    {
      "epoch": 0.9001956947162426,
      "grad_norm": 0.1611328125,
      "learning_rate": 9.980430528375734e-07,
      "loss": 0.7779,
      "step": 460
    },
    {
      "epoch": 0.9021526418786693,
      "grad_norm": 0.1396484375,
      "learning_rate": 9.784735812133073e-07,
      "loss": 0.8562,
      "step": 461
    },
    {
      "epoch": 0.9041095890410958,
      "grad_norm": 0.1318359375,
      "learning_rate": 9.589041095890411e-07,
      "loss": 0.829,
      "step": 462
    },
    {
      "epoch": 0.9060665362035225,
      "grad_norm": 0.189453125,
      "learning_rate": 9.39334637964775e-07,
      "loss": 0.7345,
      "step": 463
    },
    {
      "epoch": 0.9080234833659491,
      "grad_norm": 0.1533203125,
      "learning_rate": 9.197651663405089e-07,
      "loss": 0.8129,
      "step": 464
    },
    {
      "epoch": 0.9099804305283757,
      "grad_norm": 0.1611328125,
      "learning_rate": 9.001956947162427e-07,
      "loss": 0.7889,
      "step": 465
    },
    {
      "epoch": 0.9119373776908023,
      "grad_norm": 0.1396484375,
      "learning_rate": 8.806262230919766e-07,
      "loss": 0.8081,
      "step": 466
    },
    {
      "epoch": 0.913894324853229,
      "grad_norm": 0.1552734375,
      "learning_rate": 8.610567514677104e-07,
      "loss": 0.8084,
      "step": 467
    },
    {
      "epoch": 0.9158512720156555,
      "grad_norm": 0.1533203125,
      "learning_rate": 8.414872798434442e-07,
      "loss": 0.7586,
      "step": 468
    },
    {
      "epoch": 0.9178082191780822,
      "grad_norm": 0.15234375,
      "learning_rate": 8.219178082191781e-07,
      "loss": 0.8032,
      "step": 469
    },
    {
      "epoch": 0.9197651663405088,
      "grad_norm": 0.1474609375,
      "learning_rate": 8.023483365949119e-07,
      "loss": 0.8043,
      "step": 470
    },
    {
      "epoch": 0.9217221135029354,
      "grad_norm": 0.1474609375,
      "learning_rate": 7.827788649706458e-07,
      "loss": 0.8214,
      "step": 471
    },
    {
      "epoch": 0.923679060665362,
      "grad_norm": 0.1337890625,
      "learning_rate": 7.632093933463797e-07,
      "loss": 0.8134,
      "step": 472
    },
    {
      "epoch": 0.9256360078277887,
      "grad_norm": 0.14453125,
      "learning_rate": 7.436399217221135e-07,
      "loss": 0.829,
      "step": 473
    },
    {
      "epoch": 0.9275929549902152,
      "grad_norm": 0.15625,
      "learning_rate": 7.240704500978474e-07,
      "loss": 0.8308,
      "step": 474
    },
    {
      "epoch": 0.9295499021526419,
      "grad_norm": 0.1484375,
      "learning_rate": 7.045009784735812e-07,
      "loss": 0.8163,
      "step": 475
    },
    {
      "epoch": 0.9315068493150684,
      "grad_norm": 0.2080078125,
      "learning_rate": 6.849315068493151e-07,
      "loss": 0.7667,
      "step": 476
    },
    {
      "epoch": 0.9334637964774951,
      "grad_norm": 0.1337890625,
      "learning_rate": 6.65362035225049e-07,
      "loss": 0.8218,
      "step": 477
    },
    {
      "epoch": 0.9354207436399217,
      "grad_norm": 0.14453125,
      "learning_rate": 6.457925636007828e-07,
      "loss": 0.7775,
      "step": 478
    },
    {
      "epoch": 0.9373776908023483,
      "grad_norm": 0.1416015625,
      "learning_rate": 6.262230919765167e-07,
      "loss": 0.8613,
      "step": 479
    },
    {
      "epoch": 0.9393346379647749,
      "grad_norm": 0.13671875,
      "learning_rate": 6.066536203522505e-07,
      "loss": 0.8279,
      "step": 480
    },
    {
      "epoch": 0.9412915851272016,
      "grad_norm": 0.13671875,
      "learning_rate": 5.870841487279843e-07,
      "loss": 0.8247,
      "step": 481
    },
    {
      "epoch": 0.9432485322896281,
      "grad_norm": 0.142578125,
      "learning_rate": 5.675146771037182e-07,
      "loss": 0.8399,
      "step": 482
    },
    {
      "epoch": 0.9452054794520548,
      "grad_norm": 0.142578125,
      "learning_rate": 5.47945205479452e-07,
      "loss": 0.7841,
      "step": 483
    },
    {
      "epoch": 0.9471624266144814,
      "grad_norm": 0.1494140625,
      "learning_rate": 5.283757338551859e-07,
      "loss": 0.7923,
      "step": 484
    },
    {
      "epoch": 0.949119373776908,
      "grad_norm": 0.14453125,
      "learning_rate": 5.088062622309198e-07,
      "loss": 0.8464,
      "step": 485
    },
    {
      "epoch": 0.9510763209393346,
      "grad_norm": 0.138671875,
      "learning_rate": 4.892367906066536e-07,
      "loss": 0.801,
      "step": 486
    },
    {
      "epoch": 0.9530332681017613,
      "grad_norm": 0.2470703125,
      "learning_rate": 4.696673189823875e-07,
      "loss": 0.7939,
      "step": 487
    },
    {
      "epoch": 0.9549902152641878,
      "grad_norm": 0.1376953125,
      "learning_rate": 4.5009784735812136e-07,
      "loss": 0.8744,
      "step": 488
    },
    {
      "epoch": 0.9569471624266145,
      "grad_norm": 0.15234375,
      "learning_rate": 4.305283757338552e-07,
      "loss": 0.8419,
      "step": 489
    },
    {
      "epoch": 0.958904109589041,
      "grad_norm": 0.208984375,
      "learning_rate": 4.1095890410958903e-07,
      "loss": 0.8459,
      "step": 490
    },
    {
      "epoch": 0.9608610567514677,
      "grad_norm": 0.1416015625,
      "learning_rate": 3.913894324853229e-07,
      "loss": 0.8289,
      "step": 491
    },
    {
      "epoch": 0.9628180039138943,
      "grad_norm": 0.140625,
      "learning_rate": 3.7181996086105676e-07,
      "loss": 0.8121,
      "step": 492
    },
    {
      "epoch": 0.9647749510763209,
      "grad_norm": 0.1494140625,
      "learning_rate": 3.522504892367906e-07,
      "loss": 0.8218,
      "step": 493
    },
    {
      "epoch": 0.9667318982387475,
      "grad_norm": 0.1875,
      "learning_rate": 3.326810176125245e-07,
      "loss": 0.7736,
      "step": 494
    },
    {
      "epoch": 0.9686888454011742,
      "grad_norm": 0.142578125,
      "learning_rate": 3.1311154598825835e-07,
      "loss": 0.8321,
      "step": 495
    },
    {
      "epoch": 0.9706457925636007,
      "grad_norm": 0.154296875,
      "learning_rate": 2.9354207436399216e-07,
      "loss": 0.8535,
      "step": 496
    },
    {
      "epoch": 0.9726027397260274,
      "grad_norm": 0.13671875,
      "learning_rate": 2.73972602739726e-07,
      "loss": 0.8368,
      "step": 497
    },
    {
      "epoch": 0.974559686888454,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.544031311154599e-07,
      "loss": 0.8223,
      "step": 498
    },
    {
      "epoch": 0.9765166340508806,
      "grad_norm": 0.150390625,
      "learning_rate": 2.3483365949119375e-07,
      "loss": 0.8236,
      "step": 499
    },
    {
      "epoch": 0.9784735812133072,
      "grad_norm": 0.1640625,
      "learning_rate": 2.152641878669276e-07,
      "loss": 0.8086,
      "step": 500
    },
    {
      "epoch": 0.9804305283757339,
      "grad_norm": 0.1611328125,
      "learning_rate": 1.9569471624266145e-07,
      "loss": 0.7586,
      "step": 501
    },
    {
      "epoch": 0.9823874755381604,
      "grad_norm": 0.177734375,
      "learning_rate": 1.761252446183953e-07,
      "loss": 0.8024,
      "step": 502
    },
    {
      "epoch": 0.9843444227005871,
      "grad_norm": 0.1337890625,
      "learning_rate": 1.5655577299412917e-07,
      "loss": 0.8554,
      "step": 503
    },
    {
      "epoch": 0.9863013698630136,
      "grad_norm": 0.1953125,
      "learning_rate": 1.36986301369863e-07,
      "loss": 0.7178,
      "step": 504
    },
    {
      "epoch": 0.9882583170254403,
      "grad_norm": 0.1533203125,
      "learning_rate": 1.1741682974559687e-07,
      "loss": 0.8381,
      "step": 505
    },
    {
      "epoch": 0.9902152641878669,
      "grad_norm": 0.142578125,
      "learning_rate": 9.784735812133072e-08,
      "loss": 0.8041,
      "step": 506
    },
    {
      "epoch": 0.9921722113502935,
      "grad_norm": 0.1455078125,
      "learning_rate": 7.827788649706459e-08,
      "loss": 0.8308,
      "step": 507
    },
    {
      "epoch": 0.9941291585127201,
      "grad_norm": 0.1630859375,
      "learning_rate": 5.870841487279844e-08,
      "loss": 0.8443,
      "step": 508
    },
    {
      "epoch": 0.9960861056751468,
      "grad_norm": 0.1376953125,
      "learning_rate": 3.9138943248532294e-08,
      "loss": 0.8119,
      "step": 509
    },
    {
      "epoch": 0.9980430528375733,
      "grad_norm": 0.1513671875,
      "learning_rate": 1.9569471624266147e-08,
      "loss": 0.7697,
      "step": 510
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.0,
      "loss": 0.8326,
      "step": 511
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 511,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.6513747864266998e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}