{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1875,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0016,
      "grad_norm": 83.96888576839658,
      "learning_rate": 0.0,
      "loss": 4.327,
      "step": 1
    },
    {
      "epoch": 0.0032,
      "grad_norm": 91.31966566399686,
      "learning_rate": 5.319148936170213e-08,
      "loss": 4.419,
      "step": 2
    },
    {
      "epoch": 0.0048,
      "grad_norm": 87.56909279314893,
      "learning_rate": 1.0638297872340426e-07,
      "loss": 4.4313,
      "step": 3
    },
    {
      "epoch": 0.0064,
      "grad_norm": 91.33433750263646,
      "learning_rate": 1.5957446808510638e-07,
      "loss": 4.5427,
      "step": 4
    },
    {
      "epoch": 0.008,
      "grad_norm": 85.41395050941443,
      "learning_rate": 2.1276595744680852e-07,
      "loss": 4.3076,
      "step": 5
    },
    {
      "epoch": 0.0096,
      "grad_norm": 87.08422955423524,
      "learning_rate": 2.6595744680851066e-07,
      "loss": 4.5089,
      "step": 6
    },
    {
      "epoch": 0.0112,
      "grad_norm": 86.2412872709995,
      "learning_rate": 3.1914893617021275e-07,
      "loss": 4.4956,
      "step": 7
    },
    {
      "epoch": 0.0128,
      "grad_norm": 83.0695196331119,
      "learning_rate": 3.723404255319149e-07,
      "loss": 4.3874,
      "step": 8
    },
    {
      "epoch": 0.0144,
      "grad_norm": 81.77858364312108,
      "learning_rate": 4.2553191489361704e-07,
      "loss": 4.3019,
      "step": 9
    },
    {
      "epoch": 0.016,
      "grad_norm": 75.26118182998104,
      "learning_rate": 4.787234042553192e-07,
      "loss": 4.3324,
      "step": 10
    },
    {
      "epoch": 0.0176,
      "grad_norm": 64.62857845160552,
      "learning_rate": 5.319148936170213e-07,
      "loss": 4.1263,
      "step": 11
    },
    {
      "epoch": 0.0192,
      "grad_norm": 57.33506805921625,
      "learning_rate": 5.851063829787235e-07,
      "loss": 3.9066,
      "step": 12
    },
    {
      "epoch": 0.0208,
      "grad_norm": 59.38514555941613,
      "learning_rate": 6.382978723404255e-07,
      "loss": 4.018,
      "step": 13
    },
    {
      "epoch": 0.0224,
      "grad_norm": 35.5947487067906,
      "learning_rate": 6.914893617021278e-07,
      "loss": 3.5317,
      "step": 14
    },
    {
      "epoch": 0.024,
      "grad_norm": 34.94172923757436,
      "learning_rate": 7.446808510638298e-07,
      "loss": 3.5982,
      "step": 15
    },
    {
      "epoch": 0.0256,
      "grad_norm": 31.246590853045284,
      "learning_rate": 7.97872340425532e-07,
      "loss": 3.5303,
      "step": 16
    },
    {
      "epoch": 0.0272,
      "grad_norm": 29.27836558828987,
      "learning_rate": 8.510638297872341e-07,
      "loss": 3.4568,
      "step": 17
    },
    {
      "epoch": 0.0288,
      "grad_norm": 29.1494976526524,
      "learning_rate": 9.042553191489363e-07,
      "loss": 3.562,
      "step": 18
    },
    {
      "epoch": 0.0304,
      "grad_norm": 26.374868525420776,
      "learning_rate": 9.574468085106384e-07,
      "loss": 3.2067,
      "step": 19
    },
    {
      "epoch": 0.032,
      "grad_norm": 26.389758806684526,
      "learning_rate": 1.0106382978723404e-06,
      "loss": 3.1893,
      "step": 20
    },
    {
      "epoch": 0.0336,
      "grad_norm": 24.50976330922104,
      "learning_rate": 1.0638297872340427e-06,
      "loss": 3.2951,
      "step": 21
    },
    {
      "epoch": 0.0352,
      "grad_norm": 16.156569876029472,
      "learning_rate": 1.1170212765957447e-06,
      "loss": 2.9994,
      "step": 22
    },
    {
      "epoch": 0.0368,
      "grad_norm": 11.07857574112629,
      "learning_rate": 1.170212765957447e-06,
      "loss": 3.034,
      "step": 23
    },
    {
      "epoch": 0.0384,
      "grad_norm": 10.423592151563629,
      "learning_rate": 1.223404255319149e-06,
      "loss": 3.0882,
      "step": 24
    },
    {
      "epoch": 0.04,
      "grad_norm": 9.745928651307173,
      "learning_rate": 1.276595744680851e-06,
      "loss": 3.0375,
      "step": 25
    },
    {
      "epoch": 0.0416,
      "grad_norm": 10.587416814520605,
      "learning_rate": 1.3297872340425533e-06,
      "loss": 2.9481,
      "step": 26
    },
    {
      "epoch": 0.0432,
      "grad_norm": 8.69596290080208,
      "learning_rate": 1.3829787234042555e-06,
      "loss": 2.799,
      "step": 27
    },
    {
      "epoch": 0.0448,
      "grad_norm": 7.289191783884944,
      "learning_rate": 1.4361702127659578e-06,
      "loss": 2.8441,
      "step": 28
    },
    {
      "epoch": 0.0464,
      "grad_norm": 5.6938349004593904,
      "learning_rate": 1.4893617021276596e-06,
      "loss": 2.7122,
      "step": 29
    },
    {
      "epoch": 0.048,
      "grad_norm": 5.816199805585433,
      "learning_rate": 1.5425531914893618e-06,
      "loss": 2.6885,
      "step": 30
    },
    {
      "epoch": 0.0496,
      "grad_norm": 5.609983115735322,
      "learning_rate": 1.595744680851064e-06,
      "loss": 2.6463,
      "step": 31
    },
    {
      "epoch": 0.0512,
      "grad_norm": 6.1315839471558276,
      "learning_rate": 1.648936170212766e-06,
      "loss": 2.7066,
      "step": 32
    },
    {
      "epoch": 0.0528,
      "grad_norm": 4.961340103312237,
      "learning_rate": 1.7021276595744682e-06,
      "loss": 2.6177,
      "step": 33
    },
    {
      "epoch": 0.0544,
      "grad_norm": 4.286174413021592,
      "learning_rate": 1.7553191489361704e-06,
      "loss": 2.5035,
      "step": 34
    },
    {
      "epoch": 0.056,
      "grad_norm": 4.254151912943163,
      "learning_rate": 1.8085106382978727e-06,
      "loss": 2.4787,
      "step": 35
    },
    {
      "epoch": 0.0576,
      "grad_norm": 4.315128285216844,
      "learning_rate": 1.8617021276595745e-06,
      "loss": 2.5775,
      "step": 36
    },
    {
      "epoch": 0.0592,
      "grad_norm": 4.128893245623374,
      "learning_rate": 1.9148936170212767e-06,
      "loss": 2.449,
      "step": 37
    },
    {
      "epoch": 0.0608,
      "grad_norm": 3.840745047207333,
      "learning_rate": 1.968085106382979e-06,
      "loss": 2.3346,
      "step": 38
    },
    {
      "epoch": 0.0624,
      "grad_norm": 3.429047051403677,
      "learning_rate": 2.021276595744681e-06,
      "loss": 2.4862,
      "step": 39
    },
    {
      "epoch": 0.064,
      "grad_norm": 3.108863864048659,
      "learning_rate": 2.074468085106383e-06,
      "loss": 2.3626,
      "step": 40
    },
    {
      "epoch": 0.0656,
      "grad_norm": 3.261830759723144,
      "learning_rate": 2.1276595744680853e-06,
      "loss": 2.3703,
      "step": 41
    },
    {
      "epoch": 0.0672,
      "grad_norm": 3.022000583705631,
      "learning_rate": 2.1808510638297876e-06,
      "loss": 2.3006,
      "step": 42
    },
    {
      "epoch": 0.0688,
      "grad_norm": 2.723041480829612,
      "learning_rate": 2.2340425531914894e-06,
      "loss": 2.1945,
      "step": 43
    },
    {
      "epoch": 0.0704,
      "grad_norm": 2.632409842982361,
      "learning_rate": 2.2872340425531916e-06,
      "loss": 2.1298,
      "step": 44
    },
    {
      "epoch": 0.072,
      "grad_norm": 2.4834400463035045,
      "learning_rate": 2.340425531914894e-06,
      "loss": 2.0982,
      "step": 45
    },
    {
      "epoch": 0.0736,
      "grad_norm": 2.8489054946951584,
      "learning_rate": 2.393617021276596e-06,
      "loss": 2.2291,
      "step": 46
    },
    {
      "epoch": 0.0752,
      "grad_norm": 2.563918505938661,
      "learning_rate": 2.446808510638298e-06,
      "loss": 2.1438,
      "step": 47
    },
    {
      "epoch": 0.0768,
      "grad_norm": 2.4635396842713533,
      "learning_rate": 2.5e-06,
      "loss": 2.0942,
      "step": 48
    },
    {
      "epoch": 0.0784,
      "grad_norm": 2.197327586988563,
      "learning_rate": 2.553191489361702e-06,
      "loss": 2.0969,
      "step": 49
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1766581335619195,
      "learning_rate": 2.6063829787234047e-06,
      "loss": 2.087,
      "step": 50
    },
    {
      "epoch": 0.0816,
      "grad_norm": 2.314758018941928,
      "learning_rate": 2.6595744680851065e-06,
      "loss": 2.0107,
      "step": 51
    },
    {
      "epoch": 0.0832,
      "grad_norm": 2.3169109648365507,
      "learning_rate": 2.7127659574468084e-06,
      "loss": 1.99,
      "step": 52
    },
    {
      "epoch": 0.0848,
      "grad_norm": 2.0666471945780294,
      "learning_rate": 2.765957446808511e-06,
      "loss": 2.0287,
      "step": 53
    },
    {
      "epoch": 0.0864,
      "grad_norm": 2.195248594321617,
      "learning_rate": 2.819148936170213e-06,
      "loss": 2.0223,
      "step": 54
    },
    {
      "epoch": 0.088,
      "grad_norm": 2.473465623800088,
      "learning_rate": 2.8723404255319155e-06,
      "loss": 1.8464,
      "step": 55
    },
    {
      "epoch": 0.0896,
      "grad_norm": 2.176792757932094,
      "learning_rate": 2.9255319148936174e-06,
      "loss": 1.9284,
      "step": 56
    },
    {
      "epoch": 0.0912,
      "grad_norm": 2.1119137115797417,
      "learning_rate": 2.978723404255319e-06,
      "loss": 1.9631,
      "step": 57
    },
    {
      "epoch": 0.0928,
      "grad_norm": 2.703635123087661,
      "learning_rate": 3.031914893617022e-06,
      "loss": 1.9444,
      "step": 58
    },
    {
      "epoch": 0.0944,
      "grad_norm": 2.083313296620025,
      "learning_rate": 3.0851063829787237e-06,
      "loss": 1.8896,
      "step": 59
    },
    {
      "epoch": 0.096,
      "grad_norm": 2.190754900064518,
      "learning_rate": 3.1382978723404255e-06,
      "loss": 1.8359,
      "step": 60
    },
    {
      "epoch": 0.0976,
      "grad_norm": 2.044933834535774,
      "learning_rate": 3.191489361702128e-06,
      "loss": 1.7869,
      "step": 61
    },
    {
      "epoch": 0.0992,
      "grad_norm": 7.919400156140475,
      "learning_rate": 3.24468085106383e-06,
      "loss": 1.8209,
      "step": 62
    },
    {
      "epoch": 0.1008,
      "grad_norm": 2.2585404188512324,
      "learning_rate": 3.297872340425532e-06,
      "loss": 1.8121,
      "step": 63
    },
    {
      "epoch": 0.1024,
      "grad_norm": 1.8832284444469338,
      "learning_rate": 3.3510638297872345e-06,
      "loss": 1.8777,
      "step": 64
    },
    {
      "epoch": 0.104,
      "grad_norm": 1.8431868728363672,
      "learning_rate": 3.4042553191489363e-06,
      "loss": 1.7776,
      "step": 65
    },
    {
      "epoch": 0.1056,
      "grad_norm": 2.0310031878527015,
      "learning_rate": 3.457446808510639e-06,
      "loss": 1.7204,
      "step": 66
    },
    {
      "epoch": 0.1072,
      "grad_norm": 1.6970884138416462,
      "learning_rate": 3.510638297872341e-06,
      "loss": 1.7811,
      "step": 67
    },
    {
      "epoch": 0.1088,
      "grad_norm": 2.0565071641253727,
      "learning_rate": 3.5638297872340426e-06,
      "loss": 1.7645,
      "step": 68
    },
    {
      "epoch": 0.1104,
      "grad_norm": 1.72779039815946,
      "learning_rate": 3.6170212765957453e-06,
      "loss": 1.7506,
      "step": 69
    },
    {
      "epoch": 0.112,
      "grad_norm": 1.6133922309759734,
      "learning_rate": 3.670212765957447e-06,
      "loss": 1.7619,
      "step": 70
    },
    {
      "epoch": 0.1136,
      "grad_norm": 1.7996182185336602,
      "learning_rate": 3.723404255319149e-06,
      "loss": 1.708,
      "step": 71
    },
    {
      "epoch": 0.1152,
      "grad_norm": 1.7733451829141527,
      "learning_rate": 3.7765957446808516e-06,
      "loss": 1.6895,
      "step": 72
    },
    {
      "epoch": 0.1168,
      "grad_norm": 1.749405535595353,
      "learning_rate": 3.8297872340425535e-06,
      "loss": 1.7735,
      "step": 73
    },
    {
      "epoch": 0.1184,
      "grad_norm": 1.5898392483198043,
      "learning_rate": 3.882978723404256e-06,
      "loss": 1.7426,
      "step": 74
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.605553325605672,
      "learning_rate": 3.936170212765958e-06,
      "loss": 1.7431,
      "step": 75
    },
    {
      "epoch": 0.1216,
      "grad_norm": 1.9237208015369838,
      "learning_rate": 3.98936170212766e-06,
      "loss": 1.6873,
      "step": 76
    },
    {
      "epoch": 0.1232,
      "grad_norm": 1.6279059293506584,
      "learning_rate": 4.042553191489362e-06,
      "loss": 1.7326,
      "step": 77
    },
    {
      "epoch": 0.1248,
      "grad_norm": 1.4280947057849065,
      "learning_rate": 4.095744680851064e-06,
      "loss": 1.6423,
      "step": 78
    },
    {
      "epoch": 0.1264,
      "grad_norm": 1.9092525793977773,
      "learning_rate": 4.148936170212766e-06,
      "loss": 1.6589,
      "step": 79
    },
    {
      "epoch": 0.128,
      "grad_norm": 1.5178517248058305,
      "learning_rate": 4.202127659574468e-06,
      "loss": 1.67,
      "step": 80
    },
    {
      "epoch": 0.1296,
      "grad_norm": 1.735394341562396,
      "learning_rate": 4.255319148936171e-06,
      "loss": 1.6433,
      "step": 81
    },
    {
      "epoch": 0.1312,
      "grad_norm": 1.4582394237563958,
      "learning_rate": 4.308510638297873e-06,
      "loss": 1.6346,
      "step": 82
    },
    {
      "epoch": 0.1328,
      "grad_norm": 1.6297163926549845,
      "learning_rate": 4.361702127659575e-06,
      "loss": 1.6732,
      "step": 83
    },
    {
      "epoch": 0.1344,
      "grad_norm": 1.4847835079715939,
      "learning_rate": 4.414893617021277e-06,
      "loss": 1.6679,
      "step": 84
    },
    {
      "epoch": 0.136,
      "grad_norm": 1.379757672500155,
      "learning_rate": 4.468085106382979e-06,
      "loss": 1.6578,
      "step": 85
    },
    {
      "epoch": 0.1376,
      "grad_norm": 1.5203376067469037,
      "learning_rate": 4.521276595744681e-06,
      "loss": 1.6661,
      "step": 86
    },
    {
      "epoch": 0.1392,
      "grad_norm": 1.4456036011508362,
      "learning_rate": 4.574468085106383e-06,
      "loss": 1.7216,
      "step": 87
    },
    {
      "epoch": 0.1408,
      "grad_norm": 1.5195186424950862,
      "learning_rate": 4.6276595744680855e-06,
      "loss": 1.6341,
      "step": 88
    },
    {
      "epoch": 0.1424,
      "grad_norm": 1.4044342253016513,
      "learning_rate": 4.680851063829788e-06,
      "loss": 1.6139,
      "step": 89
    },
    {
      "epoch": 0.144,
      "grad_norm": 1.6437475470249057,
      "learning_rate": 4.73404255319149e-06,
      "loss": 1.6453,
      "step": 90
    },
    {
      "epoch": 0.1456,
      "grad_norm": 1.6031120659689886,
      "learning_rate": 4.787234042553192e-06,
      "loss": 1.6367,
      "step": 91
    },
    {
      "epoch": 0.1472,
      "grad_norm": 1.6310465942800896,
      "learning_rate": 4.840425531914894e-06,
      "loss": 1.6668,
      "step": 92
    },
    {
      "epoch": 0.1488,
      "grad_norm": 1.6120801756903371,
      "learning_rate": 4.893617021276596e-06,
      "loss": 1.6603,
      "step": 93
    },
    {
      "epoch": 0.1504,
      "grad_norm": 1.5441241909256869,
      "learning_rate": 4.946808510638298e-06,
      "loss": 1.6282,
      "step": 94
    },
    {
      "epoch": 0.152,
      "grad_norm": 2.3173816212441243,
      "learning_rate": 5e-06,
      "loss": 1.6666,
      "step": 95
    },
    {
      "epoch": 0.1536,
      "grad_norm": 1.5762215124328085,
      "learning_rate": 5.053191489361703e-06,
      "loss": 1.6257,
      "step": 96
    },
    {
      "epoch": 0.1552,
      "grad_norm": 1.5520418623963455,
      "learning_rate": 5.106382978723404e-06,
      "loss": 1.6181,
      "step": 97
    },
    {
      "epoch": 0.1568,
      "grad_norm": 1.5862064973109657,
      "learning_rate": 5.159574468085107e-06,
      "loss": 1.6899,
      "step": 98
    },
    {
      "epoch": 0.1584,
      "grad_norm": 1.5778690720874962,
      "learning_rate": 5.212765957446809e-06,
      "loss": 1.6082,
      "step": 99
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.422744596591887,
      "learning_rate": 5.265957446808511e-06,
      "loss": 1.6216,
      "step": 100
    },
    {
      "epoch": 0.1616,
      "grad_norm": 1.4754447964978863,
      "learning_rate": 5.319148936170213e-06,
      "loss": 1.6356,
      "step": 101
    },
    {
      "epoch": 0.1632,
      "grad_norm": 1.739535117707715,
      "learning_rate": 5.372340425531915e-06,
      "loss": 1.5467,
      "step": 102
    },
    {
      "epoch": 0.1648,
      "grad_norm": 1.7400625108327397,
      "learning_rate": 5.425531914893617e-06,
      "loss": 1.7226,
      "step": 103
    },
    {
      "epoch": 0.1664,
      "grad_norm": 1.4817858193441444,
      "learning_rate": 5.47872340425532e-06,
      "loss": 1.5972,
      "step": 104
    },
    {
      "epoch": 0.168,
      "grad_norm": 1.4362041455142467,
      "learning_rate": 5.531914893617022e-06,
      "loss": 1.62,
      "step": 105
    },
    {
      "epoch": 0.1696,
      "grad_norm": 1.481775448312164,
      "learning_rate": 5.5851063829787235e-06,
      "loss": 1.5589,
      "step": 106
    },
    {
      "epoch": 0.1712,
      "grad_norm": 1.3425750056225467,
      "learning_rate": 5.638297872340426e-06,
      "loss": 1.5659,
      "step": 107
    },
    {
      "epoch": 0.1728,
      "grad_norm": 1.4529386891284868,
      "learning_rate": 5.691489361702128e-06,
      "loss": 1.5551,
      "step": 108
    },
    {
      "epoch": 0.1744,
      "grad_norm": 1.6805970089871614,
      "learning_rate": 5.744680851063831e-06,
      "loss": 1.5913,
      "step": 109
    },
    {
      "epoch": 0.176,
      "grad_norm": 1.4660163857271982,
      "learning_rate": 5.7978723404255325e-06,
      "loss": 1.5761,
      "step": 110
    },
    {
      "epoch": 0.1776,
      "grad_norm": 1.5862400051538896,
      "learning_rate": 5.851063829787235e-06,
      "loss": 1.5596,
      "step": 111
    },
    {
      "epoch": 0.1792,
      "grad_norm": 1.1716158284584197,
      "learning_rate": 5.904255319148937e-06,
      "loss": 1.538,
      "step": 112
    },
    {
      "epoch": 0.1808,
      "grad_norm": 1.8023925931197025,
      "learning_rate": 5.957446808510638e-06,
      "loss": 1.5467,
      "step": 113
    },
    {
      "epoch": 0.1824,
      "grad_norm": 1.3952411268473197,
      "learning_rate": 6.010638297872341e-06,
      "loss": 1.552,
      "step": 114
    },
    {
      "epoch": 0.184,
      "grad_norm": 1.6972348844876675,
      "learning_rate": 6.063829787234044e-06,
      "loss": 1.5129,
      "step": 115
    },
    {
      "epoch": 0.1856,
      "grad_norm": 1.7256229460002293,
      "learning_rate": 6.117021276595745e-06,
      "loss": 1.5806,
      "step": 116
    },
    {
      "epoch": 0.1872,
      "grad_norm": 1.2833034774113026,
      "learning_rate": 6.170212765957447e-06,
      "loss": 1.5229,
      "step": 117
    },
    {
      "epoch": 0.1888,
      "grad_norm": 1.220917952474459,
      "learning_rate": 6.22340425531915e-06,
      "loss": 1.6007,
      "step": 118
    },
    {
      "epoch": 0.1904,
      "grad_norm": 1.296841374746347,
      "learning_rate": 6.276595744680851e-06,
      "loss": 1.4628,
      "step": 119
    },
    {
      "epoch": 0.192,
      "grad_norm": 1.320177327057943,
      "learning_rate": 6.329787234042554e-06,
      "loss": 1.5671,
      "step": 120
    },
    {
      "epoch": 0.1936,
      "grad_norm": 1.5170023239637727,
      "learning_rate": 6.382978723404256e-06,
      "loss": 1.5728,
      "step": 121
    },
    {
      "epoch": 0.1952,
      "grad_norm": 1.2406669173562086,
      "learning_rate": 6.436170212765958e-06,
      "loss": 1.6057,
      "step": 122
    },
    {
      "epoch": 0.1968,
      "grad_norm": 1.3587840934844246,
      "learning_rate": 6.48936170212766e-06,
      "loss": 1.5581,
      "step": 123
    },
    {
      "epoch": 0.1984,
      "grad_norm": 1.2817660619465792,
      "learning_rate": 6.542553191489362e-06,
      "loss": 1.5411,
      "step": 124
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3230018181829954,
      "learning_rate": 6.595744680851064e-06,
      "loss": 1.5549,
      "step": 125
    },
    {
      "epoch": 0.2016,
      "grad_norm": 1.5978048824907949,
      "learning_rate": 6.648936170212767e-06,
      "loss": 1.5283,
      "step": 126
    },
    {
      "epoch": 0.2032,
      "grad_norm": 1.2880207779152755,
      "learning_rate": 6.702127659574469e-06,
      "loss": 1.5589,
      "step": 127
    },
    {
      "epoch": 0.2048,
      "grad_norm": 1.448341449725927,
      "learning_rate": 6.75531914893617e-06,
      "loss": 1.5361,
      "step": 128
    },
    {
      "epoch": 0.2064,
      "grad_norm": 1.2869914593522047,
      "learning_rate": 6.808510638297873e-06,
      "loss": 1.5775,
      "step": 129
    },
    {
      "epoch": 0.208,
      "grad_norm": 1.2465832863577584,
      "learning_rate": 6.861702127659575e-06,
      "loss": 1.5006,
      "step": 130
    },
    {
      "epoch": 0.2096,
      "grad_norm": 1.6394298541399523,
      "learning_rate": 6.914893617021278e-06,
      "loss": 1.5456,
      "step": 131
    },
    {
      "epoch": 0.2112,
      "grad_norm": 1.350080805533287,
      "learning_rate": 6.968085106382979e-06,
      "loss": 1.4998,
      "step": 132
    },
    {
      "epoch": 0.2128,
      "grad_norm": 1.467520301256843,
      "learning_rate": 7.021276595744682e-06,
      "loss": 1.5881,
      "step": 133
    },
    {
      "epoch": 0.2144,
      "grad_norm": 1.3884253314385266,
      "learning_rate": 7.074468085106384e-06,
      "loss": 1.5612,
      "step": 134
    },
    {
      "epoch": 0.216,
      "grad_norm": 1.2837625374264625,
      "learning_rate": 7.127659574468085e-06,
      "loss": 1.589,
      "step": 135
    },
    {
      "epoch": 0.2176,
      "grad_norm": 1.4667536641730585,
      "learning_rate": 7.1808510638297875e-06,
      "loss": 1.481,
      "step": 136
    },
    {
      "epoch": 0.2192,
      "grad_norm": 1.469927636711873,
      "learning_rate": 7.234042553191491e-06,
      "loss": 1.5424,
      "step": 137
    },
    {
      "epoch": 0.2208,
      "grad_norm": 1.2960615017012136,
      "learning_rate": 7.287234042553192e-06,
      "loss": 1.477,
      "step": 138
    },
    {
      "epoch": 0.2224,
      "grad_norm": 1.3649352128780534,
      "learning_rate": 7.340425531914894e-06,
      "loss": 1.5078,
      "step": 139
    },
    {
      "epoch": 0.224,
      "grad_norm": 1.473969030470594,
      "learning_rate": 7.3936170212765965e-06,
      "loss": 1.5323,
      "step": 140
    },
    {
      "epoch": 0.2256,
      "grad_norm": 1.3727202280561577,
      "learning_rate": 7.446808510638298e-06,
      "loss": 1.5167,
      "step": 141
    },
    {
      "epoch": 0.2272,
      "grad_norm": 1.5584621380385375,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.5504,
      "step": 142
    },
    {
      "epoch": 0.2288,
      "grad_norm": 1.3632850679245647,
      "learning_rate": 7.553191489361703e-06,
      "loss": 1.4836,
      "step": 143
    },
    {
      "epoch": 0.2304,
      "grad_norm": 1.6081697988695212,
      "learning_rate": 7.606382978723405e-06,
      "loss": 1.5801,
      "step": 144
    },
    {
      "epoch": 0.232,
      "grad_norm": 1.3990825386841346,
      "learning_rate": 7.659574468085107e-06,
      "loss": 1.5377,
      "step": 145
    },
    {
      "epoch": 0.2336,
      "grad_norm": 1.5452799498435228,
      "learning_rate": 7.71276595744681e-06,
      "loss": 1.493,
      "step": 146
    },
    {
      "epoch": 0.2352,
      "grad_norm": 1.3709479830976863,
      "learning_rate": 7.765957446808511e-06,
      "loss": 1.4437,
      "step": 147
    },
    {
      "epoch": 0.2368,
      "grad_norm": 1.6802383990187488,
      "learning_rate": 7.819148936170213e-06,
      "loss": 1.5489,
      "step": 148
    },
    {
      "epoch": 0.2384,
      "grad_norm": 1.45086520226594,
      "learning_rate": 7.872340425531916e-06,
      "loss": 1.5173,
      "step": 149
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8109605891827119,
      "learning_rate": 7.925531914893617e-06,
      "loss": 1.5681,
      "step": 150
    },
    {
      "epoch": 0.2416,
      "grad_norm": 1.49792428432427,
      "learning_rate": 7.97872340425532e-06,
      "loss": 1.4722,
      "step": 151
    },
    {
      "epoch": 0.2432,
      "grad_norm": 1.512336804395951,
      "learning_rate": 8.031914893617022e-06,
      "loss": 1.5228,
      "step": 152
    },
    {
      "epoch": 0.2448,
      "grad_norm": 1.9183689979812977,
      "learning_rate": 8.085106382978723e-06,
      "loss": 1.4597,
      "step": 153
    },
    {
      "epoch": 0.2464,
      "grad_norm": 1.3335457674011406,
      "learning_rate": 8.138297872340426e-06,
      "loss": 1.4792,
      "step": 154
    },
    {
      "epoch": 0.248,
      "grad_norm": 1.99097668490106,
      "learning_rate": 8.191489361702128e-06,
      "loss": 1.5139,
      "step": 155
    },
    {
      "epoch": 0.2496,
      "grad_norm": 1.3255715052981873,
      "learning_rate": 8.24468085106383e-06,
      "loss": 1.4915,
      "step": 156
    },
    {
      "epoch": 0.2512,
      "grad_norm": 1.6505612559995027,
      "learning_rate": 8.297872340425532e-06,
      "loss": 1.5014,
      "step": 157
    },
    {
      "epoch": 0.2528,
      "grad_norm": 1.503566350756993,
      "learning_rate": 8.351063829787235e-06,
      "loss": 1.5237,
      "step": 158
    },
    {
      "epoch": 0.2544,
      "grad_norm": 1.518708578724971,
      "learning_rate": 8.404255319148937e-06,
      "loss": 1.5545,
      "step": 159
    },
    {
      "epoch": 0.256,
      "grad_norm": 1.376628003489769,
      "learning_rate": 8.457446808510638e-06,
      "loss": 1.5237,
      "step": 160
    },
    {
      "epoch": 0.2576,
      "grad_norm": 1.325620680422541,
      "learning_rate": 8.510638297872341e-06,
      "loss": 1.5692,
      "step": 161
    },
    {
      "epoch": 0.2592,
      "grad_norm": 1.3933853360022188,
      "learning_rate": 8.563829787234044e-06,
      "loss": 1.5639,
      "step": 162
    },
    {
      "epoch": 0.2608,
      "grad_norm": 1.4540624345326596,
      "learning_rate": 8.617021276595746e-06,
      "loss": 1.5495,
      "step": 163
    },
    {
      "epoch": 0.2624,
      "grad_norm": 1.2591357013701845,
      "learning_rate": 8.670212765957447e-06,
      "loss": 1.5618,
      "step": 164
    },
    {
      "epoch": 0.264,
      "grad_norm": 1.379402395204661,
      "learning_rate": 8.72340425531915e-06,
      "loss": 1.5395,
      "step": 165
    },
    {
      "epoch": 0.2656,
      "grad_norm": 1.335402738815002,
      "learning_rate": 8.776595744680852e-06,
      "loss": 1.4545,
      "step": 166
    },
    {
      "epoch": 0.2672,
      "grad_norm": 2.0018446907776513,
      "learning_rate": 8.829787234042555e-06,
      "loss": 1.5012,
      "step": 167
    },
    {
      "epoch": 0.2688,
      "grad_norm": 1.337573590695219,
      "learning_rate": 8.882978723404256e-06,
      "loss": 1.4869,
      "step": 168
    },
    {
      "epoch": 0.2704,
      "grad_norm": 1.4234319587700826,
      "learning_rate": 8.936170212765958e-06,
      "loss": 1.4594,
      "step": 169
    },
    {
      "epoch": 0.272,
      "grad_norm": 1.4172863232944428,
      "learning_rate": 8.98936170212766e-06,
      "loss": 1.4983,
      "step": 170
    },
    {
      "epoch": 0.2736,
      "grad_norm": 1.252139497954569,
      "learning_rate": 9.042553191489362e-06,
      "loss": 1.4547,
      "step": 171
    },
    {
      "epoch": 0.2752,
      "grad_norm": 1.3083298270075756,
      "learning_rate": 9.095744680851063e-06,
      "loss": 1.5275,
      "step": 172
    },
    {
      "epoch": 0.2768,
      "grad_norm": 1.4336988372369537,
      "learning_rate": 9.148936170212767e-06,
      "loss": 1.5198,
      "step": 173
    },
    {
      "epoch": 0.2784,
      "grad_norm": 1.2018074963130918,
      "learning_rate": 9.20212765957447e-06,
      "loss": 1.568,
      "step": 174
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2982508612039867,
      "learning_rate": 9.255319148936171e-06,
      "loss": 1.6258,
      "step": 175
    },
    {
      "epoch": 0.2816,
      "grad_norm": 1.2773153771697932,
      "learning_rate": 9.308510638297872e-06,
      "loss": 1.5519,
      "step": 176
    },
    {
      "epoch": 0.2832,
      "grad_norm": 1.2316506987856843,
      "learning_rate": 9.361702127659576e-06,
      "loss": 1.5152,
      "step": 177
    },
    {
      "epoch": 0.2848,
      "grad_norm": 1.2095583070027007,
      "learning_rate": 9.414893617021279e-06,
      "loss": 1.5393,
      "step": 178
    },
    {
      "epoch": 0.2864,
      "grad_norm": 1.346447226528795,
      "learning_rate": 9.46808510638298e-06,
      "loss": 1.4678,
      "step": 179
    },
    {
      "epoch": 0.288,
      "grad_norm": 1.4503293001717439,
      "learning_rate": 9.521276595744681e-06,
      "loss": 1.4917,
      "step": 180
    },
    {
      "epoch": 0.2896,
      "grad_norm": 1.3941983670870643,
      "learning_rate": 9.574468085106385e-06,
      "loss": 1.5888,
      "step": 181
    },
    {
      "epoch": 0.2912,
      "grad_norm": 1.1707238487998954,
      "learning_rate": 9.627659574468086e-06,
      "loss": 1.5206,
      "step": 182
    },
    {
      "epoch": 0.2928,
      "grad_norm": 1.5047100621156586,
      "learning_rate": 9.680851063829787e-06,
      "loss": 1.4746,
      "step": 183
    },
    {
      "epoch": 0.2944,
      "grad_norm": 1.6261387502358329,
      "learning_rate": 9.73404255319149e-06,
      "loss": 1.4362,
      "step": 184
    },
    {
      "epoch": 0.296,
      "grad_norm": 1.282573453665114,
      "learning_rate": 9.787234042553192e-06,
      "loss": 1.5425,
      "step": 185
    },
    {
      "epoch": 0.2976,
      "grad_norm": 1.1761191480894009,
      "learning_rate": 9.840425531914895e-06,
      "loss": 1.4987,
      "step": 186
    },
    {
      "epoch": 0.2992,
      "grad_norm": 1.463004498075031,
      "learning_rate": 9.893617021276596e-06,
      "loss": 1.5825,
      "step": 187
    },
    {
      "epoch": 0.3008,
      "grad_norm": 1.2477756255099024,
      "learning_rate": 9.946808510638298e-06,
      "loss": 1.5869,
      "step": 188
    },
    {
      "epoch": 0.3024,
      "grad_norm": 1.3889900748498005,
      "learning_rate": 1e-05,
      "loss": 1.471,
      "step": 189
    },
    {
      "epoch": 0.304,
      "grad_norm": 1.3367690295568562,
      "learning_rate": 9.99999133019233e-06,
      "loss": 1.4679,
      "step": 190
    },
    {
      "epoch": 0.3056,
      "grad_norm": 1.032182386403074,
      "learning_rate": 9.999965320799377e-06,
      "loss": 1.4573,
      "step": 191
    },
    {
      "epoch": 0.3072,
      "grad_norm": 1.2607024547821617,
      "learning_rate": 9.999921971911345e-06,
      "loss": 1.4764,
      "step": 192
    },
    {
      "epoch": 0.3088,
      "grad_norm": 1.3127177016200606,
      "learning_rate": 9.999861283678563e-06,
      "loss": 1.4838,
      "step": 193
    },
    {
      "epoch": 0.3104,
      "grad_norm": 1.1872166933130794,
      "learning_rate": 9.999783256311494e-06,
      "loss": 1.5124,
      "step": 194
    },
    {
      "epoch": 0.312,
      "grad_norm": 1.251057243117382,
      "learning_rate": 9.99968789008073e-06,
      "loss": 1.4313,
      "step": 195
    },
    {
      "epoch": 0.3136,
      "grad_norm": 1.2965767409592521,
      "learning_rate": 9.999575185316994e-06,
      "loss": 1.5007,
      "step": 196
    },
    {
      "epoch": 0.3152,
      "grad_norm": 1.14627370215376,
      "learning_rate": 9.999445142411139e-06,
      "loss": 1.4856,
      "step": 197
    },
    {
      "epoch": 0.3168,
      "grad_norm": 1.6957498572089729,
      "learning_rate": 9.99929776181414e-06,
      "loss": 1.5697,
      "step": 198
    },
    {
      "epoch": 0.3184,
      "grad_norm": 1.2843826779539114,
      "learning_rate": 9.999133044037107e-06,
      "loss": 1.484,
      "step": 199
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2737606864227375,
      "learning_rate": 9.998950989651261e-06,
      "loss": 1.5205,
      "step": 200
    },
    {
      "epoch": 0.3216,
      "grad_norm": 1.4960625232798468,
      "learning_rate": 9.99875159928796e-06,
      "loss": 1.4353,
      "step": 201
    },
    {
      "epoch": 0.3232,
      "grad_norm": 1.1997336664649583,
      "learning_rate": 9.99853487363867e-06,
      "loss": 1.4438,
      "step": 202
    },
    {
      "epoch": 0.3248,
      "grad_norm": 1.2384862270425072,
      "learning_rate": 9.998300813454981e-06,
      "loss": 1.5768,
      "step": 203
    },
    {
      "epoch": 0.3264,
      "grad_norm": 1.0300631820288284,
      "learning_rate": 9.998049419548597e-06,
      "loss": 1.427,
      "step": 204
    },
    {
      "epoch": 0.328,
      "grad_norm": 1.2056009292376055,
      "learning_rate": 9.997780692791329e-06,
      "loss": 1.445,
      "step": 205
    },
    {
      "epoch": 0.3296,
      "grad_norm": 1.2531217535386505,
      "learning_rate": 9.997494634115101e-06,
      "loss": 1.4574,
      "step": 206
    },
    {
      "epoch": 0.3312,
      "grad_norm": 1.161298025950752,
      "learning_rate": 9.997191244511947e-06,
      "loss": 1.4919,
      "step": 207
    },
    {
      "epoch": 0.3328,
      "grad_norm": 1.1111774831016628,
      "learning_rate": 9.996870525033994e-06,
      "loss": 1.4666,
      "step": 208
    },
    {
      "epoch": 0.3344,
      "grad_norm": 1.5596709538772309,
      "learning_rate": 9.996532476793475e-06,
      "loss": 1.4873,
      "step": 209
    },
    {
      "epoch": 0.336,
      "grad_norm": 1.2866935184512227,
      "learning_rate": 9.996177100962714e-06,
      "loss": 1.4853,
      "step": 210
    },
    {
      "epoch": 0.3376,
      "grad_norm": 1.2252201430886978,
      "learning_rate": 9.995804398774129e-06,
      "loss": 1.4803,
      "step": 211
    },
    {
      "epoch": 0.3392,
      "grad_norm": 1.3409253110617978,
      "learning_rate": 9.99541437152022e-06,
      "loss": 1.4566,
      "step": 212
    },
    {
      "epoch": 0.3408,
      "grad_norm": 1.1094662344680752,
      "learning_rate": 9.995007020553572e-06,
      "loss": 1.4736,
      "step": 213
    },
    {
      "epoch": 0.3424,
      "grad_norm": 1.6895155952677046,
      "learning_rate": 9.994582347286849e-06,
      "loss": 1.5706,
      "step": 214
    },
    {
      "epoch": 0.344,
      "grad_norm": 1.2525486549598668,
      "learning_rate": 9.994140353192782e-06,
      "loss": 1.5328,
      "step": 215
    },
    {
      "epoch": 0.3456,
      "grad_norm": 1.4011604333530476,
      "learning_rate": 9.993681039804176e-06,
      "loss": 1.4955,
      "step": 216
    },
    {
      "epoch": 0.3472,
      "grad_norm": 1.042104563446532,
      "learning_rate": 9.99320440871389e-06,
      "loss": 1.4814,
      "step": 217
    },
    {
      "epoch": 0.3488,
      "grad_norm": 1.1853009487290047,
      "learning_rate": 9.99271046157485e-06,
      "loss": 1.4698,
      "step": 218
    },
    {
      "epoch": 0.3504,
      "grad_norm": 1.1939288400114438,
      "learning_rate": 9.99219920010002e-06,
      "loss": 1.5322,
      "step": 219
    },
    {
      "epoch": 0.352,
      "grad_norm": 1.5325026028620865,
      "learning_rate": 9.991670626062422e-06,
      "loss": 1.5648,
      "step": 220
    },
    {
      "epoch": 0.3536,
      "grad_norm": 1.2578831725736794,
      "learning_rate": 9.991124741295106e-06,
      "loss": 1.4894,
      "step": 221
    },
    {
      "epoch": 0.3552,
      "grad_norm": 1.2578393358399829,
      "learning_rate": 9.990561547691159e-06,
      "loss": 1.4998,
      "step": 222
    },
    {
      "epoch": 0.3568,
      "grad_norm": 1.14238258500309,
      "learning_rate": 9.989981047203693e-06,
      "loss": 1.4896,
      "step": 223
    },
    {
      "epoch": 0.3584,
      "grad_norm": 1.1001809094151334,
      "learning_rate": 9.98938324184584e-06,
      "loss": 1.4442,
      "step": 224
    },
    {
      "epoch": 0.36,
      "grad_norm": 3.501998907921889,
      "learning_rate": 9.988768133690741e-06,
      "loss": 1.4898,
      "step": 225
    },
    {
      "epoch": 0.3616,
      "grad_norm": 1.7648646180620982,
      "learning_rate": 9.988135724871546e-06,
      "loss": 1.4558,
      "step": 226
    },
    {
      "epoch": 0.3632,
      "grad_norm": 1.1525189696401832,
      "learning_rate": 9.987486017581401e-06,
      "loss": 1.4802,
      "step": 227
    },
    {
      "epoch": 0.3648,
      "grad_norm": 1.461063115315384,
      "learning_rate": 9.986819014073436e-06,
      "loss": 1.5256,
      "step": 228
    },
    {
      "epoch": 0.3664,
      "grad_norm": 1.1919199601191655,
      "learning_rate": 9.986134716660774e-06,
      "loss": 1.557,
      "step": 229
    },
    {
      "epoch": 0.368,
      "grad_norm": 1.4317599795574656,
      "learning_rate": 9.9854331277165e-06,
      "loss": 1.5147,
      "step": 230
    },
    {
      "epoch": 0.3696,
      "grad_norm": 1.127330020654043,
      "learning_rate": 9.984714249673676e-06,
      "loss": 1.51,
      "step": 231
    },
    {
      "epoch": 0.3712,
      "grad_norm": 1.360683189302506,
      "learning_rate": 9.98397808502531e-06,
      "loss": 1.4991,
      "step": 232
    },
    {
      "epoch": 0.3728,
      "grad_norm": 1.164941910564385,
      "learning_rate": 9.983224636324369e-06,
      "loss": 1.4934,
      "step": 233
    },
    {
      "epoch": 0.3744,
      "grad_norm": 1.3942950311600224,
      "learning_rate": 9.982453906183754e-06,
      "loss": 1.4818,
      "step": 234
    },
    {
      "epoch": 0.376,
      "grad_norm": 1.197370801122074,
      "learning_rate": 9.981665897276298e-06,
      "loss": 1.5015,
      "step": 235
    },
    {
      "epoch": 0.3776,
      "grad_norm": 1.3355450458659575,
      "learning_rate": 9.980860612334753e-06,
      "loss": 1.5117,
      "step": 236
    },
    {
      "epoch": 0.3792,
      "grad_norm": 1.2908793207953804,
      "learning_rate": 9.980038054151789e-06,
      "loss": 1.5106,
      "step": 237
    },
    {
      "epoch": 0.3808,
      "grad_norm": 1.1616399116560787,
      "learning_rate": 9.979198225579968e-06,
      "loss": 1.4799,
      "step": 238
    },
    {
      "epoch": 0.3824,
      "grad_norm": 1.110099361969353,
      "learning_rate": 9.97834112953176e-06,
      "loss": 1.5091,
      "step": 239
    },
    {
      "epoch": 0.384,
      "grad_norm": 1.2869830542427836,
      "learning_rate": 9.9774667689795e-06,
      "loss": 1.4767,
      "step": 240
    },
    {
      "epoch": 0.3856,
      "grad_norm": 1.0836141387112168,
      "learning_rate": 9.976575146955409e-06,
      "loss": 1.4891,
      "step": 241
    },
    {
      "epoch": 0.3872,
      "grad_norm": 1.1511839534145174,
      "learning_rate": 9.97566626655156e-06,
      "loss": 1.4543,
      "step": 242
    },
    {
      "epoch": 0.3888,
      "grad_norm": 1.1656368296244786,
      "learning_rate": 9.974740130919883e-06,
      "loss": 1.3977,
      "step": 243
    },
    {
      "epoch": 0.3904,
      "grad_norm": 1.2593975820991925,
      "learning_rate": 9.973796743272141e-06,
      "loss": 1.4821,
      "step": 244
    },
    {
      "epoch": 0.392,
      "grad_norm": 1.2772905709431284,
      "learning_rate": 9.972836106879936e-06,
      "loss": 1.4478,
      "step": 245
    },
    {
      "epoch": 0.3936,
      "grad_norm": 1.278782110593137,
      "learning_rate": 9.971858225074672e-06,
      "loss": 1.4721,
      "step": 246
    },
    {
      "epoch": 0.3952,
      "grad_norm": 1.348516217253955,
      "learning_rate": 9.970863101247578e-06,
      "loss": 1.4351,
      "step": 247
    },
    {
      "epoch": 0.3968,
      "grad_norm": 0.9566816969820353,
      "learning_rate": 9.96985073884966e-06,
      "loss": 1.5496,
      "step": 248
    },
    {
      "epoch": 0.3984,
      "grad_norm": 1.3513364089022382,
      "learning_rate": 9.968821141391716e-06,
      "loss": 1.5421,
      "step": 249
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2449677992485972,
      "learning_rate": 9.96777431244431e-06,
      "loss": 1.4933,
      "step": 250
    },
    {
      "epoch": 0.4016,
      "grad_norm": 1.3298295054633305,
      "learning_rate": 9.966710255637764e-06,
      "loss": 1.4436,
      "step": 251
    },
    {
      "epoch": 0.4032,
      "grad_norm": 1.1614890524320938,
      "learning_rate": 9.965628974662145e-06,
      "loss": 1.4431,
      "step": 252
    },
    {
      "epoch": 0.4048,
      "grad_norm": 1.4217859684413299,
      "learning_rate": 9.964530473267253e-06,
      "loss": 1.5169,
      "step": 253
    },
    {
      "epoch": 0.4064,
      "grad_norm": 1.0219745937548286,
      "learning_rate": 9.963414755262606e-06,
      "loss": 1.5114,
      "step": 254
    },
    {
      "epoch": 0.408,
      "grad_norm": 1.390281775403407,
      "learning_rate": 9.962281824517427e-06,
      "loss": 1.5661,
      "step": 255
    },
    {
      "epoch": 0.4096,
      "grad_norm": 1.3402031536429548,
      "learning_rate": 9.961131684960635e-06,
      "loss": 1.5401,
      "step": 256
    },
    {
      "epoch": 0.4112,
      "grad_norm": 1.050798431164902,
      "learning_rate": 9.959964340580823e-06,
      "loss": 1.4636,
      "step": 257
    },
    {
      "epoch": 0.4128,
      "grad_norm": 1.7679094927969832,
      "learning_rate": 9.958779795426253e-06,
      "loss": 1.5274,
      "step": 258
    },
    {
      "epoch": 0.4144,
      "grad_norm": 1.1649105295543003,
      "learning_rate": 9.957578053604837e-06,
      "loss": 1.4672,
      "step": 259
    },
    {
      "epoch": 0.416,
      "grad_norm": 1.356303520242325,
      "learning_rate": 9.956359119284123e-06,
      "loss": 1.4511,
      "step": 260
    },
    {
      "epoch": 0.4176,
      "grad_norm": 1.3167672439185454,
      "learning_rate": 9.955122996691278e-06,
      "loss": 1.4627,
      "step": 261
    },
    {
      "epoch": 0.4192,
      "grad_norm": 1.1628895304653968,
      "learning_rate": 9.953869690113085e-06,
      "loss": 1.5492,
      "step": 262
    },
    {
      "epoch": 0.4208,
      "grad_norm": 1.407692864632045,
      "learning_rate": 9.952599203895912e-06,
      "loss": 1.559,
      "step": 263
    },
    {
      "epoch": 0.4224,
      "grad_norm": 15.983519127883453,
      "learning_rate": 9.95131154244571e-06,
      "loss": 1.5658,
      "step": 264
    },
    {
      "epoch": 0.424,
      "grad_norm": 2.3306892765400957,
      "learning_rate": 9.950006710227986e-06,
      "loss": 1.4904,
      "step": 265
    },
    {
      "epoch": 0.4256,
      "grad_norm": 1.3119707481162175,
      "learning_rate": 9.9486847117678e-06,
      "loss": 1.5044,
      "step": 266
    },
    {
      "epoch": 0.4272,
      "grad_norm": 2.1168169875867004,
      "learning_rate": 9.947345551649741e-06,
      "loss": 1.4575,
      "step": 267
    },
    {
      "epoch": 0.4288,
      "grad_norm": 1.7678881952865002,
      "learning_rate": 9.945989234517913e-06,
      "loss": 1.4726,
      "step": 268
    },
    {
      "epoch": 0.4304,
      "grad_norm": 2.058899504919541,
      "learning_rate": 9.94461576507592e-06,
      "loss": 1.5455,
      "step": 269
    },
    {
      "epoch": 0.432,
      "grad_norm": 1.8759780502198073,
      "learning_rate": 9.943225148086846e-06,
      "loss": 1.4909,
      "step": 270
    },
    {
      "epoch": 0.4336,
      "grad_norm": 1.5868949774650993,
      "learning_rate": 9.941817388373248e-06,
      "loss": 1.4455,
      "step": 271
    },
    {
      "epoch": 0.4352,
      "grad_norm": 1.502827160287061,
      "learning_rate": 9.940392490817124e-06,
      "loss": 1.486,
      "step": 272
    },
    {
      "epoch": 0.4368,
      "grad_norm": 1.7071866467458545,
      "learning_rate": 9.938950460359912e-06,
      "loss": 1.5032,
      "step": 273
    },
    {
      "epoch": 0.4384,
      "grad_norm": 1.3952028137774504,
      "learning_rate": 9.937491302002462e-06,
      "loss": 1.4347,
      "step": 274
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7072605161967411,
      "learning_rate": 9.936015020805022e-06,
      "loss": 1.4529,
      "step": 275
    },
    {
      "epoch": 0.4416,
      "grad_norm": 1.4234449247460264,
      "learning_rate": 9.934521621887223e-06,
      "loss": 1.4236,
      "step": 276
    },
    {
      "epoch": 0.4432,
      "grad_norm": 1.7235718810406606,
      "learning_rate": 9.933011110428058e-06,
      "loss": 1.4629,
      "step": 277
    },
    {
      "epoch": 0.4448,
      "grad_norm": 1.349053275473019,
      "learning_rate": 9.93148349166586e-06,
      "loss": 1.5471,
      "step": 278
    },
    {
      "epoch": 0.4464,
      "grad_norm": 1.7968793031326296,
      "learning_rate": 9.929938770898299e-06,
      "loss": 1.4786,
      "step": 279
    },
    {
      "epoch": 0.448,
      "grad_norm": 1.2703286012746722,
      "learning_rate": 9.928376953482343e-06,
      "loss": 1.4338,
      "step": 280
    },
    {
      "epoch": 0.4496,
      "grad_norm": 1.7237229071715365,
      "learning_rate": 9.926798044834261e-06,
      "loss": 1.4948,
      "step": 281
    },
    {
      "epoch": 0.4512,
      "grad_norm": 1.4267297852407117,
      "learning_rate": 9.92520205042958e-06,
      "loss": 1.5282,
      "step": 282
    },
    {
      "epoch": 0.4528,
      "grad_norm": 1.6524448242262395,
      "learning_rate": 9.92358897580309e-06,
      "loss": 1.4058,
      "step": 283
    },
    {
      "epoch": 0.4544,
      "grad_norm": 1.4592086913158095,
      "learning_rate": 9.921958826548808e-06,
      "loss": 1.4242,
      "step": 284
    },
    {
      "epoch": 0.456,
      "grad_norm": 1.770270488193743,
      "learning_rate": 9.920311608319968e-06,
      "loss": 1.4256,
      "step": 285
    },
    {
      "epoch": 0.4576,
      "grad_norm": 1.419355951771947,
      "learning_rate": 9.918647326828993e-06,
      "loss": 1.3681,
      "step": 286
    },
    {
      "epoch": 0.4592,
      "grad_norm": 1.6865876941096876,
      "learning_rate": 9.916965987847485e-06,
      "loss": 1.4186,
      "step": 287
    },
    {
      "epoch": 0.4608,
      "grad_norm": 1.4454930793944292,
      "learning_rate": 9.915267597206198e-06,
      "loss": 1.4461,
      "step": 288
    },
    {
      "epoch": 0.4624,
      "grad_norm": 1.7894433709325641,
      "learning_rate": 9.913552160795022e-06,
      "loss": 1.4272,
      "step": 289
    },
    {
      "epoch": 0.464,
      "grad_norm": 1.568569898861861,
      "learning_rate": 9.911819684562954e-06,
      "loss": 1.4484,
      "step": 290
    },
    {
      "epoch": 0.4656,
      "grad_norm": 1.7794466526773005,
      "learning_rate": 9.910070174518093e-06,
      "loss": 1.4777,
      "step": 291
    },
    {
      "epoch": 0.4672,
      "grad_norm": 1.6760814737298007,
      "learning_rate": 9.908303636727604e-06,
      "loss": 1.4702,
      "step": 292
    },
    {
      "epoch": 0.4688,
      "grad_norm": 1.4953455792307826,
      "learning_rate": 9.9065200773177e-06,
      "loss": 1.4475,
      "step": 293
    },
    {
      "epoch": 0.4704,
      "grad_norm": 1.4953180753096649,
      "learning_rate": 9.904719502473635e-06,
      "loss": 1.4441,
      "step": 294
    },
    {
      "epoch": 0.472,
      "grad_norm": 1.458387798824813,
      "learning_rate": 9.902901918439658e-06,
      "loss": 1.4171,
      "step": 295
    },
    {
      "epoch": 0.4736,
      "grad_norm": 1.2798721535891207,
      "learning_rate": 9.901067331519013e-06,
      "loss": 1.5099,
      "step": 296
    },
    {
      "epoch": 0.4752,
      "grad_norm": 1.4534399592991287,
      "learning_rate": 9.899215748073906e-06,
      "loss": 1.5149,
      "step": 297
    },
    {
      "epoch": 0.4768,
      "grad_norm": 1.2436007449472803,
      "learning_rate": 9.897347174525487e-06,
      "loss": 1.5343,
      "step": 298
    },
    {
      "epoch": 0.4784,
      "grad_norm": 1.4564194826146166,
      "learning_rate": 9.895461617353823e-06,
      "loss": 1.466,
      "step": 299
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.191855527534418,
      "learning_rate": 9.893559083097885e-06,
      "loss": 1.4968,
      "step": 300
    },
    {
      "epoch": 0.4816,
      "grad_norm": 1.3626858034565652,
      "learning_rate": 9.891639578355511e-06,
      "loss": 1.4033,
      "step": 301
    },
    {
      "epoch": 0.4832,
      "grad_norm": 1.144188489973957,
      "learning_rate": 9.8897031097834e-06,
      "loss": 1.4367,
      "step": 302
    },
    {
      "epoch": 0.4848,
      "grad_norm": 1.5507167804019006,
      "learning_rate": 9.887749684097072e-06,
      "loss": 1.4347,
      "step": 303
    },
    {
      "epoch": 0.4864,
      "grad_norm": 1.2033492438789368,
      "learning_rate": 9.88577930807086e-06,
      "loss": 1.4581,
      "step": 304
    },
    {
      "epoch": 0.488,
      "grad_norm": 1.8277701503976098,
      "learning_rate": 9.883791988537874e-06,
      "loss": 1.4543,
      "step": 305
    },
    {
      "epoch": 0.4896,
      "grad_norm": 1.5176094701793028,
      "learning_rate": 9.881787732389987e-06,
      "loss": 1.449,
      "step": 306
    },
    {
      "epoch": 0.4912,
      "grad_norm": 1.636692790334047,
      "learning_rate": 9.879766546577805e-06,
      "loss": 1.3924,
      "step": 307
    },
    {
      "epoch": 0.4928,
      "grad_norm": 1.6224196737905565,
      "learning_rate": 9.877728438110645e-06,
      "loss": 1.4413,
      "step": 308
    },
    {
      "epoch": 0.4944,
      "grad_norm": 1.4259177047249683,
      "learning_rate": 9.87567341405651e-06,
      "loss": 1.4558,
      "step": 309
    },
    {
      "epoch": 0.496,
      "grad_norm": 1.3769524398407151,
      "learning_rate": 9.873601481542065e-06,
      "loss": 1.5189,
      "step": 310
    },
    {
      "epoch": 0.4976,
      "grad_norm": 1.3479895400054256,
      "learning_rate": 9.871512647752612e-06,
      "loss": 1.4936,
      "step": 311
    },
    {
      "epoch": 0.4992,
      "grad_norm": 1.178941113251283,
      "learning_rate": 9.86940691993207e-06,
      "loss": 1.4526,
      "step": 312
    },
    {
      "epoch": 0.5008,
      "grad_norm": 1.2038157632134066,
      "learning_rate": 9.867284305382936e-06,
      "loss": 1.4327,
      "step": 313
    },
    {
      "epoch": 0.5024,
      "grad_norm": 1.0976220232678495,
      "learning_rate": 9.865144811466275e-06,
      "loss": 1.4296,
      "step": 314
    },
    {
      "epoch": 0.504,
      "grad_norm": 1.3116607428265081,
      "learning_rate": 9.86298844560169e-06,
      "loss": 1.4451,
      "step": 315
    },
    {
      "epoch": 0.5056,
      "grad_norm": 1.1385114240925454,
      "learning_rate": 9.860815215267288e-06,
      "loss": 1.4391,
      "step": 316
    },
    {
      "epoch": 0.5072,
      "grad_norm": 1.1128722464893872,
      "learning_rate": 9.858625127999668e-06,
      "loss": 1.4652,
      "step": 317
    },
    {
      "epoch": 0.5088,
      "grad_norm": 0.8781661632416442,
      "learning_rate": 9.856418191393881e-06,
      "loss": 1.3921,
      "step": 318
    },
    {
      "epoch": 0.5104,
      "grad_norm": 1.173053963576615,
      "learning_rate": 9.854194413103418e-06,
      "loss": 1.5433,
      "step": 319
    },
    {
      "epoch": 0.512,
      "grad_norm": 0.9468990246715718,
      "learning_rate": 9.851953800840166e-06,
      "loss": 1.4309,
      "step": 320
    },
    {
      "epoch": 0.5136,
      "grad_norm": 1.083128743731162,
      "learning_rate": 9.849696362374399e-06,
      "loss": 1.3884,
      "step": 321
    },
    {
      "epoch": 0.5152,
      "grad_norm": 1.152784974999538,
      "learning_rate": 9.847422105534739e-06,
      "loss": 1.4203,
      "step": 322
    },
    {
      "epoch": 0.5168,
      "grad_norm": 0.9896366987266405,
      "learning_rate": 9.845131038208135e-06,
      "loss": 1.4688,
      "step": 323
    },
    {
      "epoch": 0.5184,
      "grad_norm": 1.109630384260666,
      "learning_rate": 9.84282316833983e-06,
      "loss": 1.4213,
      "step": 324
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.074416650144299,
      "learning_rate": 9.84049850393334e-06,
      "loss": 1.382,
      "step": 325
    },
    {
      "epoch": 0.5216,
      "grad_norm": 1.0688394999024706,
      "learning_rate": 9.838157053050423e-06,
      "loss": 1.4328,
      "step": 326
    },
    {
      "epoch": 0.5232,
      "grad_norm": 1.006958925411963,
      "learning_rate": 9.83579882381105e-06,
      "loss": 1.4838,
      "step": 327
    },
    {
      "epoch": 0.5248,
      "grad_norm": 1.0840568782094824,
      "learning_rate": 9.83342382439338e-06,
      "loss": 1.4563,
      "step": 328
    },
    {
      "epoch": 0.5264,
      "grad_norm": 0.9763853683088985,
      "learning_rate": 9.831032063033726e-06,
      "loss": 1.4335,
      "step": 329
    },
    {
      "epoch": 0.528,
      "grad_norm": 0.9983371328482389,
      "learning_rate": 9.828623548026533e-06,
      "loss": 1.4302,
      "step": 330
    },
    {
      "epoch": 0.5296,
      "grad_norm": 0.9305552732374393,
      "learning_rate": 9.826198287724346e-06,
      "loss": 1.4545,
      "step": 331
    },
    {
      "epoch": 0.5312,
      "grad_norm": 1.0146960460839802,
      "learning_rate": 9.823756290537783e-06,
      "loss": 1.5465,
      "step": 332
    },
    {
      "epoch": 0.5328,
      "grad_norm": 0.9807456666436033,
      "learning_rate": 9.821297564935499e-06,
      "loss": 1.5365,
      "step": 333
    },
    {
      "epoch": 0.5344,
      "grad_norm": 1.110930356367486,
      "learning_rate": 9.81882211944417e-06,
      "loss": 1.4362,
      "step": 334
    },
    {
      "epoch": 0.536,
      "grad_norm": 1.0000187578508193,
      "learning_rate": 9.816329962648444e-06,
      "loss": 1.472,
      "step": 335
    },
    {
      "epoch": 0.5376,
      "grad_norm": 0.9133512069602816,
      "learning_rate": 9.813821103190932e-06,
      "loss": 1.4248,
      "step": 336
    },
    {
      "epoch": 0.5392,
      "grad_norm": 0.9731054791448231,
      "learning_rate": 9.811295549772169e-06,
      "loss": 1.4422,
      "step": 337
    },
    {
      "epoch": 0.5408,
      "grad_norm": 1.0197595143419849,
      "learning_rate": 9.808753311150575e-06,
      "loss": 1.5193,
      "step": 338
    },
    {
      "epoch": 0.5424,
      "grad_norm": 0.9916402058019893,
      "learning_rate": 9.80619439614244e-06,
      "loss": 1.4324,
      "step": 339
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.9715939401172263,
      "learning_rate": 9.803618813621885e-06,
      "loss": 1.3858,
      "step": 340
    },
    {
      "epoch": 0.5456,
      "grad_norm": 1.0376563331719462,
      "learning_rate": 9.801026572520832e-06,
      "loss": 1.4663,
      "step": 341
    },
    {
      "epoch": 0.5472,
      "grad_norm": 1.0670290171579633,
      "learning_rate": 9.798417681828972e-06,
      "loss": 1.4294,
      "step": 342
    },
    {
      "epoch": 0.5488,
      "grad_norm": 0.9732448358663861,
      "learning_rate": 9.795792150593739e-06,
      "loss": 1.4686,
      "step": 343
    },
    {
      "epoch": 0.5504,
      "grad_norm": 0.9227311759548037,
      "learning_rate": 9.793149987920273e-06,
      "loss": 1.41,
      "step": 344
    },
    {
      "epoch": 0.552,
      "grad_norm": 1.0181243064093515,
      "learning_rate": 9.79049120297139e-06,
      "loss": 1.5179,
      "step": 345
    },
    {
      "epoch": 0.5536,
      "grad_norm": 1.03342679673426,
      "learning_rate": 9.787815804967554e-06,
      "loss": 1.4179,
      "step": 346
    },
    {
      "epoch": 0.5552,
      "grad_norm": 0.9910415495735191,
      "learning_rate": 9.785123803186834e-06,
      "loss": 1.4425,
      "step": 347
    },
    {
      "epoch": 0.5568,
      "grad_norm": 0.9553897800466262,
      "learning_rate": 9.782415206964892e-06,
      "loss": 1.4948,
      "step": 348
    },
    {
      "epoch": 0.5584,
      "grad_norm": 0.84967122237312,
      "learning_rate": 9.779690025694926e-06,
      "loss": 1.4514,
      "step": 349
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9751697043649451,
      "learning_rate": 9.776948268827658e-06,
      "loss": 1.4478,
      "step": 350
    },
    {
      "epoch": 0.5616,
      "grad_norm": 0.8333802230392668,
      "learning_rate": 9.77418994587129e-06,
      "loss": 1.4452,
      "step": 351
    },
    {
      "epoch": 0.5632,
      "grad_norm": 0.9837861001181029,
      "learning_rate": 9.771415066391473e-06,
      "loss": 1.4508,
      "step": 352
    },
    {
      "epoch": 0.5648,
      "grad_norm": 0.9303378628951637,
      "learning_rate": 9.768623640011272e-06,
      "loss": 1.4329,
      "step": 353
    },
    {
      "epoch": 0.5664,
      "grad_norm": 1.0069729655617146,
      "learning_rate": 9.765815676411145e-06,
      "loss": 1.4383,
      "step": 354
    },
    {
      "epoch": 0.568,
      "grad_norm": 1.023948233460435,
      "learning_rate": 9.762991185328891e-06,
      "loss": 1.4959,
      "step": 355
    },
    {
      "epoch": 0.5696,
      "grad_norm": 0.9233963911558033,
      "learning_rate": 9.760150176559627e-06,
      "loss": 1.3882,
      "step": 356
    },
    {
      "epoch": 0.5712,
      "grad_norm": 1.114042642119405,
      "learning_rate": 9.757292659955755e-06,
      "loss": 1.5125,
      "step": 357
    },
    {
      "epoch": 0.5728,
      "grad_norm": 1.0434436771909195,
      "learning_rate": 9.754418645426919e-06,
      "loss": 1.4135,
      "step": 358
    },
    {
      "epoch": 0.5744,
      "grad_norm": 1.4040933007466314,
      "learning_rate": 9.751528142939986e-06,
      "loss": 1.5146,
      "step": 359
    },
    {
      "epoch": 0.576,
      "grad_norm": 1.5056404985198253,
      "learning_rate": 9.74862116251899e-06,
      "loss": 1.4859,
      "step": 360
    },
    {
      "epoch": 0.5776,
      "grad_norm": 1.1086909620785568,
      "learning_rate": 9.74569771424512e-06,
      "loss": 1.4675,
      "step": 361
    },
    {
      "epoch": 0.5792,
      "grad_norm": 1.2424534080790788,
      "learning_rate": 9.742757808256667e-06,
      "loss": 1.5197,
      "step": 362
    },
    {
      "epoch": 0.5808,
      "grad_norm": 1.1460612577352152,
      "learning_rate": 9.739801454749e-06,
      "loss": 1.4471,
      "step": 363
    },
    {
      "epoch": 0.5824,
      "grad_norm": 1.105780043020093,
      "learning_rate": 9.736828663974527e-06,
      "loss": 1.504,
      "step": 364
    },
    {
      "epoch": 0.584,
      "grad_norm": 1.3307075805856143,
      "learning_rate": 9.733839446242655e-06,
      "loss": 1.4941,
      "step": 365
    },
    {
      "epoch": 0.5856,
      "grad_norm": 1.0864596499138872,
      "learning_rate": 9.730833811919763e-06,
      "loss": 1.5042,
      "step": 366
    },
    {
      "epoch": 0.5872,
      "grad_norm": 1.3088419657737063,
      "learning_rate": 9.727811771429158e-06,
      "loss": 1.4558,
      "step": 367
    },
    {
      "epoch": 0.5888,
      "grad_norm": 1.0040762608860703,
      "learning_rate": 9.724773335251046e-06,
      "loss": 1.4786,
      "step": 368
    },
    {
      "epoch": 0.5904,
      "grad_norm": 1.3745248333188465,
      "learning_rate": 9.721718513922488e-06,
      "loss": 1.4049,
      "step": 369
    },
    {
      "epoch": 0.592,
      "grad_norm": 1.1755741007630596,
      "learning_rate": 9.71864731803737e-06,
      "loss": 1.5065,
      "step": 370
    },
    {
      "epoch": 0.5936,
      "grad_norm": 1.3992604591191051,
      "learning_rate": 9.715559758246363e-06,
      "loss": 1.5223,
      "step": 371
    },
    {
      "epoch": 0.5952,
      "grad_norm": 1.270379474852935,
      "learning_rate": 9.712455845256888e-06,
      "loss": 1.4828,
      "step": 372
    },
    {
      "epoch": 0.5968,
      "grad_norm": 1.59510974889637,
      "learning_rate": 9.709335589833076e-06,
      "loss": 1.4412,
      "step": 373
    },
    {
      "epoch": 0.5984,
      "grad_norm": 1.298821216905264,
      "learning_rate": 9.70619900279573e-06,
      "loss": 1.4856,
      "step": 374
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.580895082716664,
      "learning_rate": 9.703046095022297e-06,
      "loss": 1.4627,
      "step": 375
    },
    {
      "epoch": 0.6016,
      "grad_norm": 1.521059236022838,
      "learning_rate": 9.699876877446815e-06,
      "loss": 1.4171,
      "step": 376
    },
    {
      "epoch": 0.6032,
      "grad_norm": 1.4630570834072918,
      "learning_rate": 9.696691361059886e-06,
      "loss": 1.4507,
      "step": 377
    },
    {
      "epoch": 0.6048,
      "grad_norm": 1.395289206111259,
      "learning_rate": 9.693489556908641e-06,
      "loss": 1.427,
      "step": 378
    },
    {
      "epoch": 0.6064,
      "grad_norm": 1.229893500369459,
      "learning_rate": 9.690271476096686e-06,
      "loss": 1.476,
      "step": 379
    },
    {
      "epoch": 0.608,
      "grad_norm": 1.1245090723342341,
      "learning_rate": 9.68703712978408e-06,
      "loss": 1.4299,
      "step": 380
    },
    {
      "epoch": 0.6096,
      "grad_norm": 1.3072136839500712,
      "learning_rate": 9.683786529187287e-06,
      "loss": 1.4896,
      "step": 381
    },
    {
      "epoch": 0.6112,
      "grad_norm": 1.1185460850078062,
      "learning_rate": 9.680519685579137e-06,
      "loss": 1.4549,
      "step": 382
    },
    {
      "epoch": 0.6128,
      "grad_norm": 1.1410031331675459,
      "learning_rate": 9.677236610288797e-06,
      "loss": 1.4546,
      "step": 383
    },
    {
      "epoch": 0.6144,
      "grad_norm": 1.1156561447560143,
      "learning_rate": 9.673937314701714e-06,
      "loss": 1.4781,
      "step": 384
    },
    {
      "epoch": 0.616,
      "grad_norm": 1.1563161911085673,
      "learning_rate": 9.670621810259596e-06,
      "loss": 1.4021,
      "step": 385
    },
    {
      "epoch": 0.6176,
      "grad_norm": 1.4345640776648152,
      "learning_rate": 9.667290108460354e-06,
      "loss": 1.4368,
      "step": 386
    },
    {
      "epoch": 0.6192,
      "grad_norm": 1.0427302725176528,
      "learning_rate": 9.663942220858075e-06,
      "loss": 1.4431,
      "step": 387
    },
    {
      "epoch": 0.6208,
      "grad_norm": 1.0090238116280525,
      "learning_rate": 9.660578159062977e-06,
      "loss": 1.4426,
      "step": 388
    },
    {
      "epoch": 0.6224,
      "grad_norm": 0.9175331690022491,
      "learning_rate": 9.657197934741366e-06,
      "loss": 1.5014,
      "step": 389
    },
    {
      "epoch": 0.624,
      "grad_norm": 2.0318786881968802,
      "learning_rate": 9.6538015596156e-06,
      "loss": 1.4215,
      "step": 390
    },
    {
      "epoch": 0.6256,
      "grad_norm": 1.0649570442331067,
      "learning_rate": 9.650389045464046e-06,
      "loss": 1.4518,
      "step": 391
    },
    {
      "epoch": 0.6272,
      "grad_norm": 1.0304488557946732,
      "learning_rate": 9.646960404121042e-06,
      "loss": 1.4232,
      "step": 392
    },
    {
      "epoch": 0.6288,
      "grad_norm": 1.0092061927982725,
      "learning_rate": 9.643515647476851e-06,
      "loss": 1.4455,
      "step": 393
    },
    {
      "epoch": 0.6304,
      "grad_norm": 1.0909259690569286,
      "learning_rate": 9.640054787477626e-06,
      "loss": 1.3963,
      "step": 394
    },
    {
      "epoch": 0.632,
      "grad_norm": 0.9956268018383412,
      "learning_rate": 9.63657783612536e-06,
      "loss": 1.389,
      "step": 395
    },
    {
      "epoch": 0.6336,
      "grad_norm": 1.048056880066633,
      "learning_rate": 9.633084805477857e-06,
      "loss": 1.4542,
      "step": 396
    },
    {
      "epoch": 0.6352,
      "grad_norm": 0.8690028044861864,
      "learning_rate": 9.629575707648675e-06,
      "loss": 1.496,
      "step": 397
    },
    {
      "epoch": 0.6368,
      "grad_norm": 1.0365143459066577,
      "learning_rate": 9.626050554807096e-06,
      "loss": 1.4601,
      "step": 398
    },
    {
      "epoch": 0.6384,
      "grad_norm": 1.02258016279251,
      "learning_rate": 9.62250935917808e-06,
      "loss": 1.45,
      "step": 399
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0813691026074215,
      "learning_rate": 9.618952133042223e-06,
      "loss": 1.428,
      "step": 400
    },
    {
      "epoch": 0.6416,
      "grad_norm": 0.9419229669126369,
      "learning_rate": 9.615378888735706e-06,
      "loss": 1.4818,
      "step": 401
    },
    {
      "epoch": 0.6432,
      "grad_norm": 1.1280744161754672,
      "learning_rate": 9.611789638650269e-06,
      "loss": 1.4642,
      "step": 402
    },
    {
      "epoch": 0.6448,
      "grad_norm": 0.8786887801383031,
      "learning_rate": 9.608184395233156e-06,
      "loss": 1.4558,
      "step": 403
    },
    {
      "epoch": 0.6464,
      "grad_norm": 1.0869817585306052,
      "learning_rate": 9.604563170987072e-06,
      "loss": 1.391,
      "step": 404
    },
    {
      "epoch": 0.648,
      "grad_norm": 0.9734018042960012,
      "learning_rate": 9.600925978470143e-06,
      "loss": 1.4608,
      "step": 405
    },
    {
      "epoch": 0.6496,
      "grad_norm": 0.9584933451141865,
      "learning_rate": 9.597272830295877e-06,
      "loss": 1.4281,
      "step": 406
    },
    {
      "epoch": 0.6512,
      "grad_norm": 1.0674721589298504,
      "learning_rate": 9.593603739133105e-06,
      "loss": 1.4344,
      "step": 407
    },
    {
      "epoch": 0.6528,
      "grad_norm": 0.8907204191759808,
      "learning_rate": 9.589918717705957e-06,
      "loss": 1.3768,
      "step": 408
    },
    {
      "epoch": 0.6544,
      "grad_norm": 0.9151437337950384,
      "learning_rate": 9.586217778793804e-06,
      "loss": 1.4517,
      "step": 409
    },
    {
      "epoch": 0.656,
      "grad_norm": 0.9750615850665619,
      "learning_rate": 9.582500935231215e-06,
      "loss": 1.4137,
      "step": 410
    },
    {
      "epoch": 0.6576,
      "grad_norm": 0.8289933891307314,
      "learning_rate": 9.57876819990792e-06,
      "loss": 1.4067,
      "step": 411
    },
    {
      "epoch": 0.6592,
      "grad_norm": 0.9546842080706989,
      "learning_rate": 9.575019585768758e-06,
      "loss": 1.5355,
      "step": 412
    },
    {
      "epoch": 0.6608,
      "grad_norm": 0.9505305578939003,
      "learning_rate": 9.571255105813632e-06,
      "loss": 1.472,
      "step": 413
    },
    {
      "epoch": 0.6624,
      "grad_norm": 0.8950566025057659,
      "learning_rate": 9.567474773097469e-06,
      "loss": 1.4621,
      "step": 414
    },
    {
      "epoch": 0.664,
      "grad_norm": 0.8917669064258923,
      "learning_rate": 9.563678600730175e-06,
      "loss": 1.4236,
      "step": 415
    },
    {
      "epoch": 0.6656,
      "grad_norm": 0.8509191680429908,
      "learning_rate": 9.559866601876581e-06,
      "loss": 1.4288,
      "step": 416
    },
    {
      "epoch": 0.6672,
      "grad_norm": 0.9368850217137291,
      "learning_rate": 9.556038789756407e-06,
      "loss": 1.4754,
      "step": 417
    },
    {
      "epoch": 0.6688,
      "grad_norm": 1.046246145910134,
      "learning_rate": 9.55219517764421e-06,
      "loss": 1.4894,
      "step": 418
    },
    {
      "epoch": 0.6704,
      "grad_norm": 0.9355256881748993,
      "learning_rate": 9.548335778869342e-06,
      "loss": 1.4163,
      "step": 419
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.9383887016108632,
      "learning_rate": 9.544460606815901e-06,
      "loss": 1.4484,
      "step": 420
    },
    {
      "epoch": 0.6736,
      "grad_norm": 0.8327978580700858,
      "learning_rate": 9.540569674922685e-06,
      "loss": 1.4306,
      "step": 421
    },
    {
      "epoch": 0.6752,
      "grad_norm": 0.8189865300318745,
      "learning_rate": 9.536662996683146e-06,
      "loss": 1.4172,
      "step": 422
    },
    {
      "epoch": 0.6768,
      "grad_norm": 0.9463066907369695,
      "learning_rate": 9.532740585645346e-06,
      "loss": 1.4274,
      "step": 423
    },
    {
      "epoch": 0.6784,
      "grad_norm": 0.784097895458344,
      "learning_rate": 9.528802455411902e-06,
      "loss": 1.4265,
      "step": 424
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8308442525474724,
      "learning_rate": 9.52484861963995e-06,
      "loss": 1.4571,
      "step": 425
    },
    {
      "epoch": 0.6816,
      "grad_norm": 0.7854521896269024,
      "learning_rate": 9.520879092041085e-06,
      "loss": 1.3966,
      "step": 426
    },
    {
      "epoch": 0.6832,
      "grad_norm": 0.8493067791372543,
      "learning_rate": 9.516893886381324e-06,
      "loss": 1.4604,
      "step": 427
    },
    {
      "epoch": 0.6848,
      "grad_norm": 0.8187863872444242,
      "learning_rate": 9.512893016481053e-06,
      "loss": 1.3874,
      "step": 428
    },
    {
      "epoch": 0.6864,
      "grad_norm": 0.7887228536937237,
      "learning_rate": 9.508876496214983e-06,
      "loss": 1.3625,
      "step": 429
    },
    {
      "epoch": 0.688,
      "grad_norm": 0.8877304302593954,
      "learning_rate": 9.504844339512096e-06,
      "loss": 1.4929,
      "step": 430
    },
    {
      "epoch": 0.6896,
      "grad_norm": 0.7590872023667599,
      "learning_rate": 9.500796560355603e-06,
      "loss": 1.3869,
      "step": 431
    },
    {
      "epoch": 0.6912,
      "grad_norm": 0.8932122215741385,
      "learning_rate": 9.496733172782889e-06,
      "loss": 1.4812,
      "step": 432
    },
    {
      "epoch": 0.6928,
      "grad_norm": 0.7780351389397473,
      "learning_rate": 9.492654190885469e-06,
      "loss": 1.417,
      "step": 433
    },
    {
      "epoch": 0.6944,
      "grad_norm": 0.8289048950371796,
      "learning_rate": 9.488559628808939e-06,
      "loss": 1.4728,
      "step": 434
    },
    {
      "epoch": 0.696,
      "grad_norm": 0.8435585048071734,
      "learning_rate": 9.484449500752927e-06,
      "loss": 1.3785,
      "step": 435
    },
    {
      "epoch": 0.6976,
      "grad_norm": 0.8335544689216257,
      "learning_rate": 9.480323820971039e-06,
      "loss": 1.4724,
      "step": 436
    },
    {
      "epoch": 0.6992,
      "grad_norm": 0.8677414817796826,
      "learning_rate": 9.476182603770814e-06,
      "loss": 1.4497,
      "step": 437
    },
    {
      "epoch": 0.7008,
      "grad_norm": 0.8359358772215948,
      "learning_rate": 9.472025863513676e-06,
      "loss": 1.4705,
      "step": 438
    },
    {
      "epoch": 0.7024,
      "grad_norm": 0.8441968366335426,
      "learning_rate": 9.467853614614883e-06,
      "loss": 1.4538,
      "step": 439
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.841032491493606,
      "learning_rate": 9.46366587154347e-06,
      "loss": 1.4197,
      "step": 440
    },
    {
      "epoch": 0.7056,
      "grad_norm": 0.8371166956228127,
      "learning_rate": 9.459462648822209e-06,
      "loss": 1.4365,
      "step": 441
    },
    {
      "epoch": 0.7072,
      "grad_norm": 0.8269804289181598,
      "learning_rate": 9.45524396102755e-06,
      "loss": 1.3849,
      "step": 442
    },
    {
      "epoch": 0.7088,
      "grad_norm": 0.8443883127852159,
      "learning_rate": 9.451009822789583e-06,
      "loss": 1.5057,
      "step": 443
    },
    {
      "epoch": 0.7104,
      "grad_norm": 5.072552426987265,
      "learning_rate": 9.44676024879197e-06,
      "loss": 1.4942,
      "step": 444
    },
    {
      "epoch": 0.712,
      "grad_norm": 1.313849861717472,
      "learning_rate": 9.442495253771909e-06,
      "loss": 1.4757,
      "step": 445
    },
    {
      "epoch": 0.7136,
      "grad_norm": 0.8983636398835028,
      "learning_rate": 9.438214852520073e-06,
      "loss": 1.4424,
      "step": 446
    },
    {
      "epoch": 0.7152,
      "grad_norm": 0.9771026564763301,
      "learning_rate": 9.433919059880564e-06,
      "loss": 1.4663,
      "step": 447
    },
    {
      "epoch": 0.7168,
      "grad_norm": 1.0107102942413724,
      "learning_rate": 9.429607890750863e-06,
      "loss": 1.4604,
      "step": 448
    },
    {
      "epoch": 0.7184,
      "grad_norm": 0.9092168464461536,
      "learning_rate": 9.425281360081769e-06,
      "loss": 1.4448,
      "step": 449
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.143733803281509,
      "learning_rate": 9.420939482877359e-06,
      "loss": 1.342,
      "step": 450
    },
    {
      "epoch": 0.7216,
      "grad_norm": 0.9112279120828453,
      "learning_rate": 9.416582274194929e-06,
      "loss": 1.3804,
      "step": 451
    },
    {
      "epoch": 0.7232,
      "grad_norm": 1.0654654751994028,
      "learning_rate": 9.412209749144947e-06,
      "loss": 1.4213,
      "step": 452
    },
    {
      "epoch": 0.7248,
      "grad_norm": 1.1099404316416661,
      "learning_rate": 9.40782192289099e-06,
      "loss": 1.4244,
      "step": 453
    },
    {
      "epoch": 0.7264,
      "grad_norm": 2.406412330536763,
      "learning_rate": 9.4034188106497e-06,
      "loss": 1.4053,
      "step": 454
    },
    {
      "epoch": 0.728,
      "grad_norm": 1.3868493828488384,
      "learning_rate": 9.399000427690736e-06,
      "loss": 1.4666,
      "step": 455
    },
    {
      "epoch": 0.7296,
      "grad_norm": 0.8805374171270947,
      "learning_rate": 9.394566789336707e-06,
      "loss": 1.4821,
      "step": 456
    },
    {
      "epoch": 0.7312,
      "grad_norm": 1.2184097684163755,
      "learning_rate": 9.390117910963132e-06,
      "loss": 1.4652,
      "step": 457
    },
    {
      "epoch": 0.7328,
      "grad_norm": 1.007076402926211,
      "learning_rate": 9.385653807998376e-06,
      "loss": 1.4796,
      "step": 458
    },
    {
      "epoch": 0.7344,
      "grad_norm": 1.0069912122685885,
      "learning_rate": 9.381174495923608e-06,
      "loss": 1.363,
      "step": 459
    },
    {
      "epoch": 0.736,
      "grad_norm": 1.1977446043856277,
      "learning_rate": 9.376679990272736e-06,
      "loss": 1.4195,
      "step": 460
    },
    {
      "epoch": 0.7376,
      "grad_norm": 0.9650450058391333,
      "learning_rate": 9.37217030663236e-06,
      "loss": 1.4264,
      "step": 461
    },
    {
      "epoch": 0.7392,
      "grad_norm": 1.0768825976796637,
      "learning_rate": 9.367645460641716e-06,
      "loss": 1.396,
      "step": 462
    },
    {
      "epoch": 0.7408,
      "grad_norm": 0.9257492254463009,
      "learning_rate": 9.36310546799262e-06,
      "loss": 1.447,
      "step": 463
    },
    {
      "epoch": 0.7424,
      "grad_norm": 1.0105335541928313,
      "learning_rate": 9.358550344429421e-06,
      "loss": 1.4183,
      "step": 464
    },
    {
      "epoch": 0.744,
      "grad_norm": 1.1292141669877076,
      "learning_rate": 9.353980105748934e-06,
      "loss": 1.4689,
      "step": 465
    },
    {
      "epoch": 0.7456,
      "grad_norm": 0.98287672393378,
      "learning_rate": 9.349394767800397e-06,
      "loss": 1.5354,
      "step": 466
    },
    {
      "epoch": 0.7472,
      "grad_norm": 0.9224924951065196,
      "learning_rate": 9.344794346485408e-06,
      "loss": 1.3593,
      "step": 467
    },
    {
      "epoch": 0.7488,
      "grad_norm": 0.9461761294115515,
      "learning_rate": 9.340178857757876e-06,
      "loss": 1.402,
      "step": 468
    },
    {
      "epoch": 0.7504,
      "grad_norm": 0.930138697849644,
      "learning_rate": 9.335548317623957e-06,
      "loss": 1.4175,
      "step": 469
    },
    {
      "epoch": 0.752,
      "grad_norm": 0.9618354176004978,
      "learning_rate": 9.330902742142013e-06,
      "loss": 1.409,
      "step": 470
    },
    {
      "epoch": 0.7536,
      "grad_norm": 1.193913257974537,
      "learning_rate": 9.326242147422538e-06,
      "loss": 1.4742,
      "step": 471
    },
    {
      "epoch": 0.7552,
      "grad_norm": 1.1177232389455984,
      "learning_rate": 9.321566549628118e-06,
      "loss": 1.3997,
      "step": 472
    },
    {
      "epoch": 0.7568,
      "grad_norm": 0.9885879438158058,
      "learning_rate": 9.316875964973366e-06,
      "loss": 1.4753,
      "step": 473
    },
    {
      "epoch": 0.7584,
      "grad_norm": 0.9475265308278713,
      "learning_rate": 9.31217040972487e-06,
      "loss": 1.3791,
      "step": 474
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0390044584118383,
      "learning_rate": 9.307449900201132e-06,
      "loss": 1.4801,
      "step": 475
    },
    {
      "epoch": 0.7616,
      "grad_norm": 1.000320386336902,
      "learning_rate": 9.302714452772515e-06,
      "loss": 1.3948,
      "step": 476
    },
    {
      "epoch": 0.7632,
      "grad_norm": 1.1392891576335316,
      "learning_rate": 9.29796408386119e-06,
      "loss": 1.3997,
      "step": 477
    },
    {
      "epoch": 0.7648,
      "grad_norm": 0.9652564877046603,
      "learning_rate": 9.293198809941067e-06,
      "loss": 1.407,
      "step": 478
    },
    {
      "epoch": 0.7664,
      "grad_norm": 1.1281908300707557,
      "learning_rate": 9.288418647537752e-06,
      "loss": 1.4303,
      "step": 479
    },
    {
      "epoch": 0.768,
      "grad_norm": 0.9271032557859589,
      "learning_rate": 9.283623613228479e-06,
      "loss": 1.4939,
      "step": 480
    },
    {
      "epoch": 0.7696,
      "grad_norm": 1.0187144939660993,
      "learning_rate": 9.27881372364206e-06,
      "loss": 1.4265,
      "step": 481
    },
    {
      "epoch": 0.7712,
      "grad_norm": 0.8839345302952286,
      "learning_rate": 9.27398899545882e-06,
      "loss": 1.4207,
      "step": 482
    },
    {
      "epoch": 0.7728,
      "grad_norm": 1.0336150873645362,
      "learning_rate": 9.269149445410545e-06,
      "loss": 1.44,
      "step": 483
    },
    {
      "epoch": 0.7744,
      "grad_norm": 1.084450501724458,
      "learning_rate": 9.264295090280424e-06,
      "loss": 1.4683,
      "step": 484
    },
    {
      "epoch": 0.776,
      "grad_norm": 0.9803112083731588,
      "learning_rate": 9.259425946902987e-06,
      "loss": 1.4483,
      "step": 485
    },
    {
      "epoch": 0.7776,
      "grad_norm": 1.2931600514167292,
      "learning_rate": 9.254542032164047e-06,
      "loss": 1.4196,
      "step": 486
    },
    {
      "epoch": 0.7792,
      "grad_norm": 1.0306084440896797,
      "learning_rate": 9.249643363000645e-06,
      "loss": 1.4589,
      "step": 487
    },
    {
      "epoch": 0.7808,
      "grad_norm": 1.4140277510044232,
      "learning_rate": 9.24472995640099e-06,
      "loss": 1.417,
      "step": 488
    },
    {
      "epoch": 0.7824,
      "grad_norm": 1.0572066110257337,
      "learning_rate": 9.239801829404396e-06,
      "loss": 1.4592,
      "step": 489
    },
    {
      "epoch": 0.784,
      "grad_norm": 1.3080594348387935,
      "learning_rate": 9.234858999101232e-06,
      "loss": 1.4974,
      "step": 490
    },
    {
      "epoch": 0.7856,
      "grad_norm": 1.098233791423182,
      "learning_rate": 9.22990148263285e-06,
      "loss": 1.4815,
      "step": 491
    },
    {
      "epoch": 0.7872,
      "grad_norm": 1.172167140729455,
      "learning_rate": 9.224929297191536e-06,
      "loss": 1.3924,
      "step": 492
    },
    {
      "epoch": 0.7888,
      "grad_norm": 1.2121474524902365,
      "learning_rate": 9.219942460020447e-06,
      "loss": 1.421,
      "step": 493
    },
    {
      "epoch": 0.7904,
      "grad_norm": 1.3028493490047062,
      "learning_rate": 9.214940988413552e-06,
      "loss": 1.3966,
      "step": 494
    },
    {
      "epoch": 0.792,
      "grad_norm": 1.1894379009405245,
      "learning_rate": 9.20992489971557e-06,
      "loss": 1.407,
      "step": 495
    },
    {
      "epoch": 0.7936,
      "grad_norm": 1.2246585161984351,
      "learning_rate": 9.204894211321906e-06,
      "loss": 1.4396,
      "step": 496
    },
    {
      "epoch": 0.7952,
      "grad_norm": 1.2373389069678633,
      "learning_rate": 9.199848940678607e-06,
      "loss": 1.3859,
      "step": 497
    },
    {
      "epoch": 0.7968,
      "grad_norm": 1.1828195039737146,
      "learning_rate": 9.194789105282277e-06,
      "loss": 1.3913,
      "step": 498
    },
    {
      "epoch": 0.7984,
      "grad_norm": 1.2993835900566866,
      "learning_rate": 9.189714722680041e-06,
      "loss": 1.4433,
      "step": 499
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2366256034010579,
      "learning_rate": 9.184625810469468e-06,
      "loss": 1.3813,
      "step": 500
    },
    {
      "epoch": 0.8016,
      "grad_norm": 0.9745308324707477,
      "learning_rate": 9.179522386298508e-06,
      "loss": 1.447,
      "step": 501
    },
    {
      "epoch": 0.8032,
      "grad_norm": 1.09481831151002,
      "learning_rate": 9.174404467865447e-06,
      "loss": 1.4465,
      "step": 502
    },
    {
      "epoch": 0.8048,
      "grad_norm": 0.9148535733258608,
      "learning_rate": 9.169272072918834e-06,
      "loss": 1.4433,
      "step": 503
    },
    {
      "epoch": 0.8064,
      "grad_norm": 1.151369104598539,
      "learning_rate": 9.164125219257419e-06,
      "loss": 1.4729,
      "step": 504
    },
    {
      "epoch": 0.808,
      "grad_norm": 0.924145974964028,
      "learning_rate": 9.158963924730092e-06,
      "loss": 1.429,
      "step": 505
    },
    {
      "epoch": 0.8096,
      "grad_norm": 1.3958149932614126,
      "learning_rate": 9.153788207235827e-06,
      "loss": 1.4189,
      "step": 506
    },
    {
      "epoch": 0.8112,
      "grad_norm": 1.0581861191052884,
      "learning_rate": 9.148598084723615e-06,
      "loss": 1.4058,
      "step": 507
    },
    {
      "epoch": 0.8128,
      "grad_norm": 1.451267970349791,
      "learning_rate": 9.143393575192402e-06,
      "loss": 1.4,
      "step": 508
    },
    {
      "epoch": 0.8144,
      "grad_norm": 1.285126355263028,
      "learning_rate": 9.138174696691025e-06,
      "loss": 1.4194,
      "step": 509
    },
    {
      "epoch": 0.816,
      "grad_norm": 1.1941963488260712,
      "learning_rate": 9.132941467318152e-06,
      "loss": 1.4012,
      "step": 510
    },
    {
      "epoch": 0.8176,
      "grad_norm": 1.376720804452895,
      "learning_rate": 9.127693905222223e-06,
      "loss": 1.3897,
      "step": 511
    },
    {
      "epoch": 0.8192,
      "grad_norm": 1.1145383510068851,
      "learning_rate": 9.122432028601377e-06,
      "loss": 1.4166,
      "step": 512
    },
    {
      "epoch": 0.8208,
      "grad_norm": 1.3134161794925896,
      "learning_rate": 9.1171558557034e-06,
      "loss": 1.4814,
      "step": 513
    },
    {
      "epoch": 0.8224,
      "grad_norm": 1.1596338664142773,
      "learning_rate": 9.111865404825652e-06,
      "loss": 1.4769,
      "step": 514
    },
    {
      "epoch": 0.824,
      "grad_norm": 1.1393714409130529,
      "learning_rate": 9.10656069431501e-06,
      "loss": 1.4003,
      "step": 515
    },
    {
      "epoch": 0.8256,
      "grad_norm": 1.1423740243748612,
      "learning_rate": 9.101241742567802e-06,
      "loss": 1.4443,
      "step": 516
    },
    {
      "epoch": 0.8272,
      "grad_norm": 1.1415504449932985,
      "learning_rate": 9.095908568029741e-06,
      "loss": 1.4585,
      "step": 517
    },
    {
      "epoch": 0.8288,
      "grad_norm": 1.049568386852133,
      "learning_rate": 9.09056118919587e-06,
      "loss": 1.4688,
      "step": 518
    },
    {
      "epoch": 0.8304,
      "grad_norm": 0.9972838867146476,
      "learning_rate": 9.085199624610486e-06,
      "loss": 1.3807,
      "step": 519
    },
    {
      "epoch": 0.832,
      "grad_norm": 1.042633321050248,
      "learning_rate": 9.079823892867083e-06,
      "loss": 1.4721,
      "step": 520
    },
    {
      "epoch": 0.8336,
      "grad_norm": 0.866272005191185,
      "learning_rate": 9.074434012608282e-06,
      "loss": 1.4538,
      "step": 521
    },
    {
      "epoch": 0.8352,
      "grad_norm": 0.9476857799730387,
      "learning_rate": 9.069030002525777e-06,
      "loss": 1.3394,
      "step": 522
    },
    {
      "epoch": 0.8368,
      "grad_norm": 0.8209368300049602,
      "learning_rate": 9.063611881360258e-06,
      "loss": 1.3815,
      "step": 523
    },
    {
      "epoch": 0.8384,
      "grad_norm": 0.9370773109000528,
      "learning_rate": 9.05817966790135e-06,
      "loss": 1.3825,
      "step": 524
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8791287626166582,
      "learning_rate": 9.052733380987555e-06,
      "loss": 1.52,
      "step": 525
    },
    {
      "epoch": 0.8416,
      "grad_norm": 0.8332595327963986,
      "learning_rate": 9.047273039506174e-06,
      "loss": 1.398,
      "step": 526
    },
    {
      "epoch": 0.8432,
      "grad_norm": 1.000885209969858,
      "learning_rate": 9.041798662393255e-06,
      "loss": 1.4414,
      "step": 527
    },
    {
      "epoch": 0.8448,
      "grad_norm": 0.7679291386133303,
      "learning_rate": 9.036310268633515e-06,
      "loss": 1.4534,
      "step": 528
    },
    {
      "epoch": 0.8464,
      "grad_norm": 1.0417821362072528,
      "learning_rate": 9.030807877260278e-06,
      "loss": 1.4259,
      "step": 529
    },
    {
      "epoch": 0.848,
      "grad_norm": 0.7367636640107398,
      "learning_rate": 9.025291507355419e-06,
      "loss": 1.2978,
      "step": 530
    },
    {
      "epoch": 0.8496,
      "grad_norm": 0.9785461029128765,
      "learning_rate": 9.01976117804928e-06,
      "loss": 1.3771,
      "step": 531
    },
    {
      "epoch": 0.8512,
      "grad_norm": 0.7632164468772593,
      "learning_rate": 9.014216908520619e-06,
      "loss": 1.4157,
      "step": 532
    },
    {
      "epoch": 0.8528,
      "grad_norm": 0.9479644034550154,
      "learning_rate": 9.008658717996538e-06,
      "loss": 1.4516,
      "step": 533
    },
    {
      "epoch": 0.8544,
      "grad_norm": 0.8961502909294885,
      "learning_rate": 9.003086625752414e-06,
      "loss": 1.5473,
      "step": 534
    },
    {
      "epoch": 0.856,
      "grad_norm": 0.8451831205375874,
      "learning_rate": 8.997500651111833e-06,
      "loss": 1.4925,
      "step": 535
    },
    {
      "epoch": 0.8576,
      "grad_norm": 1.2377280829190525,
      "learning_rate": 8.991900813446523e-06,
      "loss": 1.3986,
      "step": 536
    },
    {
      "epoch": 0.8592,
      "grad_norm": 0.7772474727921547,
      "learning_rate": 8.986287132176295e-06,
      "loss": 1.4372,
      "step": 537
    },
    {
      "epoch": 0.8608,
      "grad_norm": 0.7794287246862142,
      "learning_rate": 8.980659626768961e-06,
      "loss": 1.4325,
      "step": 538
    },
    {
      "epoch": 0.8624,
      "grad_norm": 0.7528026068960167,
      "learning_rate": 8.975018316740278e-06,
      "loss": 1.411,
      "step": 539
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.7783293283615662,
      "learning_rate": 8.969363221653875e-06,
      "loss": 1.4935,
      "step": 540
    },
    {
      "epoch": 0.8656,
      "grad_norm": 0.8001375092804403,
      "learning_rate": 8.963694361121186e-06,
      "loss": 1.4264,
      "step": 541
    },
    {
      "epoch": 0.8672,
      "grad_norm": 0.8862567018921613,
      "learning_rate": 8.958011754801383e-06,
      "loss": 1.4138,
      "step": 542
    },
    {
      "epoch": 0.8688,
      "grad_norm": 0.8631343709404485,
      "learning_rate": 8.952315422401307e-06,
      "loss": 1.4568,
      "step": 543
    },
    {
      "epoch": 0.8704,
      "grad_norm": 0.9083642743531506,
      "learning_rate": 8.946605383675403e-06,
      "loss": 1.4133,
      "step": 544
    },
    {
      "epoch": 0.872,
      "grad_norm": 0.9405909420813873,
      "learning_rate": 8.940881658425645e-06,
      "loss": 1.4056,
      "step": 545
    },
    {
      "epoch": 0.8736,
      "grad_norm": 0.8393098971185528,
      "learning_rate": 8.93514426650147e-06,
      "loss": 1.4667,
      "step": 546
    },
    {
      "epoch": 0.8752,
      "grad_norm": 0.9748048319020342,
      "learning_rate": 8.929393227799715e-06,
      "loss": 1.4448,
      "step": 547
    },
    {
      "epoch": 0.8768,
      "grad_norm": 0.8178085991225351,
      "learning_rate": 8.923628562264536e-06,
      "loss": 1.425,
      "step": 548
    },
    {
      "epoch": 0.8784,
      "grad_norm": 0.8760795384204181,
      "learning_rate": 8.917850289887353e-06,
      "loss": 1.4145,
      "step": 549
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9513624630284034,
      "learning_rate": 8.91205843070677e-06,
      "loss": 1.4598,
      "step": 550
    },
    {
      "epoch": 0.8816,
      "grad_norm": 0.8170257625216413,
      "learning_rate": 8.906253004808506e-06,
      "loss": 1.4638,
      "step": 551
    },
    {
      "epoch": 0.8832,
      "grad_norm": 0.8625323824515049,
      "learning_rate": 8.900434032325332e-06,
      "loss": 1.3552,
      "step": 552
    },
    {
      "epoch": 0.8848,
      "grad_norm": 0.8122077736112968,
      "learning_rate": 8.894601533437e-06,
      "loss": 1.4747,
      "step": 553
    },
    {
      "epoch": 0.8864,
      "grad_norm": 0.8905909666762591,
      "learning_rate": 8.888755528370163e-06,
      "loss": 1.3474,
      "step": 554
    },
    {
      "epoch": 0.888,
      "grad_norm": 0.7904245772629086,
      "learning_rate": 8.882896037398322e-06,
      "loss": 1.3642,
      "step": 555
    },
    {
      "epoch": 0.8896,
      "grad_norm": 0.8435262219745708,
      "learning_rate": 8.877023080841739e-06,
      "loss": 1.5116,
      "step": 556
    },
    {
      "epoch": 0.8912,
      "grad_norm": 0.9445282544681075,
      "learning_rate": 8.871136679067372e-06,
      "loss": 1.472,
      "step": 557
    },
    {
      "epoch": 0.8928,
      "grad_norm": 0.7744361533435973,
      "learning_rate": 8.865236852488813e-06,
      "loss": 1.4572,
      "step": 558
    },
    {
      "epoch": 0.8944,
      "grad_norm": 0.9496145232125562,
      "learning_rate": 8.859323621566207e-06,
      "loss": 1.3675,
      "step": 559
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.7691164665305542,
      "learning_rate": 8.853397006806183e-06,
      "loss": 1.3989,
      "step": 560
    },
    {
      "epoch": 0.8976,
      "grad_norm": 0.9434243555271439,
      "learning_rate": 8.847457028761783e-06,
      "loss": 1.4457,
      "step": 561
    },
    {
      "epoch": 0.8992,
      "grad_norm": 0.7722594905143625,
      "learning_rate": 8.841503708032398e-06,
      "loss": 1.3938,
      "step": 562
    },
    {
      "epoch": 0.9008,
      "grad_norm": 0.8649042489711487,
      "learning_rate": 8.835537065263684e-06,
      "loss": 1.4124,
      "step": 563
    },
    {
      "epoch": 0.9024,
      "grad_norm": 0.8593567073525227,
      "learning_rate": 8.829557121147499e-06,
      "loss": 1.3882,
      "step": 564
    },
    {
      "epoch": 0.904,
      "grad_norm": 0.8566369347990131,
      "learning_rate": 8.82356389642183e-06,
      "loss": 1.3844,
      "step": 565
    },
    {
      "epoch": 0.9056,
      "grad_norm": 1.0099283971032698,
      "learning_rate": 8.817557411870717e-06,
      "loss": 1.3964,
      "step": 566
    },
    {
      "epoch": 0.9072,
      "grad_norm": 0.7611344180047375,
      "learning_rate": 8.811537688324187e-06,
      "loss": 1.4086,
      "step": 567
    },
    {
      "epoch": 0.9088,
      "grad_norm": 0.9718535119933895,
      "learning_rate": 8.805504746658183e-06,
      "loss": 1.4834,
      "step": 568
    },
    {
      "epoch": 0.9104,
      "grad_norm": 0.7574171638328504,
      "learning_rate": 8.799458607794476e-06,
      "loss": 1.4424,
      "step": 569
    },
    {
      "epoch": 0.912,
      "grad_norm": 0.8791698015632617,
      "learning_rate": 8.793399292700616e-06,
      "loss": 1.4357,
      "step": 570
    },
    {
      "epoch": 0.9136,
      "grad_norm": 0.846652256175992,
      "learning_rate": 8.787326822389836e-06,
      "loss": 1.4486,
      "step": 571
    },
    {
      "epoch": 0.9152,
      "grad_norm": 0.8192548713410595,
      "learning_rate": 8.781241217921e-06,
      "loss": 1.3608,
      "step": 572
    },
    {
      "epoch": 0.9168,
      "grad_norm": 0.9765317603430138,
      "learning_rate": 8.775142500398513e-06,
      "loss": 1.4045,
      "step": 573
    },
    {
      "epoch": 0.9184,
      "grad_norm": 0.7087364637103916,
      "learning_rate": 8.769030690972262e-06,
      "loss": 1.4011,
      "step": 574
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0192809189937944,
      "learning_rate": 8.76290581083753e-06,
      "loss": 1.3918,
      "step": 575
    },
    {
      "epoch": 0.9216,
      "grad_norm": 0.810080404941639,
      "learning_rate": 8.756767881234928e-06,
      "loss": 1.4003,
      "step": 576
    },
    {
      "epoch": 0.9232,
      "grad_norm": 0.8363441373046685,
      "learning_rate": 8.750616923450328e-06,
      "loss": 1.4238,
      "step": 577
    },
    {
      "epoch": 0.9248,
      "grad_norm": 0.8025965552028345,
      "learning_rate": 8.744452958814775e-06,
      "loss": 1.4411,
      "step": 578
    },
    {
      "epoch": 0.9264,
      "grad_norm": 0.8586150883241711,
      "learning_rate": 8.738276008704426e-06,
      "loss": 1.4187,
      "step": 579
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.9198450016259929,
      "learning_rate": 8.732086094540467e-06,
      "loss": 1.3568,
      "step": 580
    },
    {
      "epoch": 0.9296,
      "grad_norm": 0.7697410429994137,
      "learning_rate": 8.725883237789046e-06,
      "loss": 1.369,
      "step": 581
    },
    {
      "epoch": 0.9312,
      "grad_norm": 0.9053528322143917,
      "learning_rate": 8.719667459961191e-06,
      "loss": 1.3693,
      "step": 582
    },
    {
      "epoch": 0.9328,
      "grad_norm": 0.776996033866215,
      "learning_rate": 8.713438782612743e-06,
      "loss": 1.4224,
      "step": 583
    },
    {
      "epoch": 0.9344,
      "grad_norm": 0.8135767155411472,
      "learning_rate": 8.707197227344275e-06,
      "loss": 1.4469,
      "step": 584
    },
    {
      "epoch": 0.936,
      "grad_norm": 0.828500007042032,
      "learning_rate": 8.700942815801023e-06,
      "loss": 1.3983,
      "step": 585
    },
    {
      "epoch": 0.9376,
      "grad_norm": 0.9971368711707599,
      "learning_rate": 8.6946755696728e-06,
      "loss": 1.4216,
      "step": 586
    },
    {
      "epoch": 0.9392,
      "grad_norm": 0.8558161446270199,
      "learning_rate": 8.688395510693939e-06,
      "loss": 1.3925,
      "step": 587
    },
    {
      "epoch": 0.9408,
      "grad_norm": 0.8868587654317946,
      "learning_rate": 8.682102660643196e-06,
      "loss": 1.397,
      "step": 588
    },
    {
      "epoch": 0.9424,
      "grad_norm": 0.9502729942673493,
      "learning_rate": 8.675797041343696e-06,
      "loss": 1.3646,
      "step": 589
    },
    {
      "epoch": 0.944,
      "grad_norm": 0.7727759925305121,
      "learning_rate": 8.669478674662839e-06,
      "loss": 1.4174,
      "step": 590
    },
    {
      "epoch": 0.9456,
      "grad_norm": 0.8879792001283602,
      "learning_rate": 8.663147582512232e-06,
      "loss": 1.4514,
      "step": 591
    },
    {
      "epoch": 0.9472,
      "grad_norm": 0.8004418230780913,
      "learning_rate": 8.65680378684762e-06,
      "loss": 1.4979,
      "step": 592
    },
    {
      "epoch": 0.9488,
      "grad_norm": 0.8607349203190552,
      "learning_rate": 8.6504473096688e-06,
      "loss": 1.4205,
      "step": 593
    },
    {
      "epoch": 0.9504,
      "grad_norm": 0.8335567256318612,
      "learning_rate": 8.64407817301954e-06,
      "loss": 1.4101,
      "step": 594
    },
    {
      "epoch": 0.952,
      "grad_norm": 0.824917609516866,
      "learning_rate": 8.637696398987517e-06,
      "loss": 1.3483,
      "step": 595
    },
    {
      "epoch": 0.9536,
      "grad_norm": 0.8785916871569546,
      "learning_rate": 8.631302009704235e-06,
      "loss": 1.3467,
      "step": 596
    },
    {
      "epoch": 0.9552,
      "grad_norm": 0.818063275052906,
      "learning_rate": 8.624895027344943e-06,
      "loss": 1.4511,
      "step": 597
    },
    {
      "epoch": 0.9568,
      "grad_norm": 0.8392457805761903,
      "learning_rate": 8.618475474128563e-06,
      "loss": 1.4238,
      "step": 598
    },
    {
      "epoch": 0.9584,
      "grad_norm": 0.8739964440365925,
      "learning_rate": 8.61204337231761e-06,
      "loss": 1.4778,
      "step": 599
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7802357771911239,
      "learning_rate": 8.605598744218122e-06,
      "loss": 1.3725,
      "step": 600
    },
    {
      "epoch": 0.9616,
      "grad_norm": 0.7725544319601196,
      "learning_rate": 8.599141612179572e-06,
      "loss": 1.4143,
      "step": 601
    },
    {
      "epoch": 0.9632,
      "grad_norm": 0.7965341891272001,
      "learning_rate": 8.592671998594794e-06,
      "loss": 1.3971,
      "step": 602
    },
    {
      "epoch": 0.9648,
      "grad_norm": 0.8044700115685771,
      "learning_rate": 8.586189925899913e-06,
      "loss": 1.4471,
      "step": 603
    },
    {
      "epoch": 0.9664,
      "grad_norm": 0.7865823846211181,
      "learning_rate": 8.57969541657426e-06,
      "loss": 1.3659,
      "step": 604
    },
    {
      "epoch": 0.968,
      "grad_norm": 1.0337351047624972,
      "learning_rate": 8.57318849314029e-06,
      "loss": 1.3472,
      "step": 605
    },
    {
      "epoch": 0.9696,
      "grad_norm": 0.8232020109867875,
      "learning_rate": 8.566669178163513e-06,
      "loss": 1.3943,
      "step": 606
    },
    {
      "epoch": 0.9712,
      "grad_norm": 0.8551670681507276,
      "learning_rate": 8.560137494252416e-06,
      "loss": 1.4167,
      "step": 607
    },
    {
      "epoch": 0.9728,
      "grad_norm": 0.7853737936454872,
      "learning_rate": 8.553593464058374e-06,
      "loss": 1.4922,
      "step": 608
    },
    {
      "epoch": 0.9744,
      "grad_norm": 0.8092072648390912,
      "learning_rate": 8.54703711027558e-06,
      "loss": 1.3961,
      "step": 609
    },
    {
      "epoch": 0.976,
      "grad_norm": 0.7707660380517899,
      "learning_rate": 8.540468455640964e-06,
      "loss": 1.3937,
      "step": 610
    },
    {
      "epoch": 0.9776,
      "grad_norm": 0.8846994306562658,
      "learning_rate": 8.533887522934114e-06,
      "loss": 1.4764,
      "step": 611
    },
    {
      "epoch": 0.9792,
      "grad_norm": 0.7679050385391923,
      "learning_rate": 8.527294334977201e-06,
      "loss": 1.4481,
      "step": 612
    },
    {
      "epoch": 0.9808,
      "grad_norm": 0.9229259558246335,
      "learning_rate": 8.520688914634894e-06,
      "loss": 1.5459,
      "step": 613
    },
    {
      "epoch": 0.9824,
      "grad_norm": 0.8330683303069205,
      "learning_rate": 8.51407128481428e-06,
      "loss": 1.4854,
      "step": 614
    },
    {
      "epoch": 0.984,
      "grad_norm": 1.0055582012108322,
      "learning_rate": 8.507441468464792e-06,
      "loss": 1.4241,
      "step": 615
    },
    {
      "epoch": 0.9856,
      "grad_norm": 0.7710515386984981,
      "learning_rate": 8.50079948857812e-06,
      "loss": 1.4393,
      "step": 616
    },
    {
      "epoch": 0.9872,
      "grad_norm": 1.1057576473774366,
      "learning_rate": 8.494145368188143e-06,
      "loss": 1.4119,
      "step": 617
    },
    {
      "epoch": 0.9888,
      "grad_norm": 0.8291530818293935,
      "learning_rate": 8.487479130370838e-06,
      "loss": 1.4349,
      "step": 618
    },
    {
      "epoch": 0.9904,
      "grad_norm": 0.9737485921135217,
      "learning_rate": 8.480800798244202e-06,
      "loss": 1.3791,
      "step": 619
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.7927301689953336,
      "learning_rate": 8.47411039496818e-06,
      "loss": 1.4778,
      "step": 620
    },
    {
      "epoch": 0.9936,
      "grad_norm": 1.0587251874576764,
      "learning_rate": 8.467407943744574e-06,
      "loss": 1.3962,
      "step": 621
    },
    {
      "epoch": 0.9952,
      "grad_norm": 0.7542805199117172,
      "learning_rate": 8.460693467816972e-06,
      "loss": 1.3428,
      "step": 622
    },
    {
      "epoch": 0.9968,
      "grad_norm": 0.8840869201365114,
      "learning_rate": 8.453966990470656e-06,
      "loss": 1.502,
      "step": 623
    },
    {
      "epoch": 0.9984,
      "grad_norm": 0.7359064928666454,
      "learning_rate": 8.447228535032536e-06,
      "loss": 1.3571,
      "step": 624
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.812939847511038,
      "learning_rate": 8.440478124871054e-06,
      "loss": 1.4234,
      "step": 625
    },
    {
      "epoch": 1.0016,
      "grad_norm": 0.8351559938505766,
      "learning_rate": 8.433715783396115e-06,
      "loss": 1.3456,
      "step": 626
    },
    {
      "epoch": 1.0032,
      "grad_norm": 0.8578078728282718,
      "learning_rate": 8.426941534058999e-06,
      "loss": 1.3458,
      "step": 627
    },
    {
      "epoch": 1.0048,
      "grad_norm": 0.8512199669067851,
      "learning_rate": 8.420155400352279e-06,
      "loss": 1.3213,
      "step": 628
    },
    {
      "epoch": 1.0064,
      "grad_norm": 0.7805827441715782,
      "learning_rate": 8.413357405809748e-06,
      "loss": 1.3225,
      "step": 629
    },
    {
      "epoch": 1.008,
      "grad_norm": 0.8590849098276044,
      "learning_rate": 8.406547574006326e-06,
      "loss": 1.3881,
      "step": 630
    },
    {
      "epoch": 1.0096,
      "grad_norm": 0.8577562854477256,
      "learning_rate": 8.399725928557985e-06,
      "loss": 1.3978,
      "step": 631
    },
    {
      "epoch": 1.0112,
      "grad_norm": 0.8096587995807837,
      "learning_rate": 8.39289249312167e-06,
      "loss": 1.4136,
      "step": 632
    },
    {
      "epoch": 1.0128,
      "grad_norm": 0.7679924686359145,
      "learning_rate": 8.386047291395208e-06,
      "loss": 1.3072,
      "step": 633
    },
    {
      "epoch": 1.0144,
      "grad_norm": 0.8281119301768063,
      "learning_rate": 8.37919034711723e-06,
      "loss": 1.3499,
      "step": 634
    },
    {
      "epoch": 1.016,
      "grad_norm": 0.8337127047364975,
      "learning_rate": 8.372321684067092e-06,
      "loss": 1.4296,
      "step": 635
    },
    {
      "epoch": 1.0176,
      "grad_norm": 0.7889129032210904,
      "learning_rate": 8.36544132606479e-06,
      "loss": 1.3662,
      "step": 636
    },
    {
      "epoch": 1.0192,
      "grad_norm": 0.8510874154361773,
      "learning_rate": 8.358549296970877e-06,
      "loss": 1.3914,
      "step": 637
    },
    {
      "epoch": 1.0208,
      "grad_norm": 0.7609194551869768,
      "learning_rate": 8.351645620686377e-06,
      "loss": 1.3181,
      "step": 638
    },
    {
      "epoch": 1.0224,
      "grad_norm": 0.8062111898189099,
      "learning_rate": 8.34473032115271e-06,
      "loss": 1.369,
      "step": 639
    },
    {
      "epoch": 1.024,
      "grad_norm": 0.7877653537192939,
      "learning_rate": 8.337803422351602e-06,
      "loss": 1.3406,
      "step": 640
    },
    {
      "epoch": 1.0256,
      "grad_norm": 0.8005611741564197,
      "learning_rate": 8.33086494830501e-06,
      "loss": 1.3502,
      "step": 641
    },
    {
      "epoch": 1.0272,
      "grad_norm": 0.8436082727933392,
      "learning_rate": 8.323914923075018e-06,
      "loss": 1.4,
      "step": 642
    },
    {
      "epoch": 1.0288,
      "grad_norm": 0.8204641790356176,
      "learning_rate": 8.316953370763788e-06,
      "loss": 1.3539,
      "step": 643
    },
    {
      "epoch": 1.0304,
      "grad_norm": 0.8094078266567143,
      "learning_rate": 8.309980315513444e-06,
      "loss": 1.3705,
      "step": 644
    },
    {
      "epoch": 1.032,
      "grad_norm": 0.8952622893057216,
      "learning_rate": 8.302995781506007e-06,
      "loss": 1.3607,
      "step": 645
    },
    {
      "epoch": 1.0336,
      "grad_norm": 0.8926278027649625,
      "learning_rate": 8.295999792963301e-06,
      "loss": 1.3616,
      "step": 646
    },
    {
      "epoch": 1.0352,
      "grad_norm": 0.966031100763654,
      "learning_rate": 8.288992374146878e-06,
      "loss": 1.4129,
      "step": 647
    },
    {
      "epoch": 1.0368,
      "grad_norm": 0.8687871928882839,
      "learning_rate": 8.281973549357927e-06,
      "loss": 1.372,
      "step": 648
    },
    {
      "epoch": 1.0384,
      "grad_norm": 0.9201106247593187,
      "learning_rate": 8.274943342937191e-06,
      "loss": 1.4218,
      "step": 649
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7684222455961462,
      "learning_rate": 8.267901779264889e-06,
      "loss": 1.3289,
      "step": 650
    },
    {
      "epoch": 1.0416,
      "grad_norm": 0.9587964461572251,
      "learning_rate": 8.260848882760616e-06,
      "loss": 1.3437,
      "step": 651
    },
    {
      "epoch": 1.0432,
      "grad_norm": 0.8296921258451506,
      "learning_rate": 8.25378467788328e-06,
      "loss": 1.3622,
      "step": 652
    },
    {
      "epoch": 1.0448,
      "grad_norm": 0.8341975344747726,
      "learning_rate": 8.246709189130997e-06,
      "loss": 1.3106,
      "step": 653
    },
    {
      "epoch": 1.0464,
      "grad_norm": 0.8122466234721465,
      "learning_rate": 8.23962244104102e-06,
      "loss": 1.416,
      "step": 654
    },
    {
      "epoch": 1.048,
      "grad_norm": 0.8445695056541725,
      "learning_rate": 8.232524458189644e-06,
      "loss": 1.342,
      "step": 655
    },
    {
      "epoch": 1.0496,
      "grad_norm": 0.752998665680426,
      "learning_rate": 8.225415265192126e-06,
      "loss": 1.3758,
      "step": 656
    },
    {
      "epoch": 1.0512,
      "grad_norm": 0.8364992199849853,
      "learning_rate": 8.218294886702606e-06,
      "loss": 1.3826,
      "step": 657
    },
    {
      "epoch": 1.0528,
      "grad_norm": 0.7860864085773033,
      "learning_rate": 8.211163347414005e-06,
      "loss": 1.4213,
      "step": 658
    },
    {
      "epoch": 1.0544,
      "grad_norm": 0.8628600460199707,
      "learning_rate": 8.20402067205795e-06,
      "loss": 1.3981,
      "step": 659
    },
    {
      "epoch": 1.056,
      "grad_norm": 0.8346221045142787,
      "learning_rate": 8.196866885404697e-06,
      "loss": 1.3015,
      "step": 660
    },
    {
      "epoch": 1.0576,
      "grad_norm": 0.8354454595962658,
      "learning_rate": 8.18970201226302e-06,
      "loss": 1.3704,
      "step": 661
    },
    {
      "epoch": 1.0592,
      "grad_norm": 0.853483819123678,
      "learning_rate": 8.182526077480153e-06,
      "loss": 1.3375,
      "step": 662
    },
    {
      "epoch": 1.0608,
      "grad_norm": 0.8233482009696034,
      "learning_rate": 8.175339105941685e-06,
      "loss": 1.3478,
      "step": 663
    },
    {
      "epoch": 1.0624,
      "grad_norm": 0.8559434682647006,
      "learning_rate": 8.168141122571478e-06,
      "loss": 1.3799,
      "step": 664
    },
    {
      "epoch": 1.064,
      "grad_norm": 0.7332574570376126,
      "learning_rate": 8.160932152331587e-06,
      "loss": 1.3632,
      "step": 665
    },
    {
      "epoch": 1.0656,
      "grad_norm": 0.8880237693145437,
      "learning_rate": 8.153712220222163e-06,
      "loss": 1.3494,
      "step": 666
    },
    {
      "epoch": 1.0672,
      "grad_norm": 0.7690549958689745,
      "learning_rate": 8.14648135128138e-06,
      "loss": 1.3103,
      "step": 667
    },
    {
      "epoch": 1.0688,
      "grad_norm": 0.8608765719493594,
      "learning_rate": 8.139239570585334e-06,
      "loss": 1.3771,
      "step": 668
    },
    {
      "epoch": 1.0704,
      "grad_norm": 0.7564794162890728,
      "learning_rate": 8.131986903247959e-06,
      "loss": 1.3413,
      "step": 669
    },
    {
      "epoch": 1.072,
      "grad_norm": 0.8309407693805636,
      "learning_rate": 8.124723374420951e-06,
      "loss": 1.3012,
      "step": 670
    },
    {
      "epoch": 1.0735999999999999,
      "grad_norm": 0.7355892390980855,
      "learning_rate": 8.117449009293668e-06,
      "loss": 1.3537,
      "step": 671
    },
    {
      "epoch": 1.0752,
      "grad_norm": 0.8153304419743217,
      "learning_rate": 8.11016383309305e-06,
      "loss": 1.3082,
      "step": 672
    },
    {
      "epoch": 1.0768,
      "grad_norm": 0.7638130506097733,
      "learning_rate": 8.102867871083528e-06,
      "loss": 1.3479,
      "step": 673
    },
    {
      "epoch": 1.0784,
      "grad_norm": 0.8566663431848623,
      "learning_rate": 8.095561148566932e-06,
      "loss": 1.3725,
      "step": 674
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.864856914870548,
      "learning_rate": 8.088243690882421e-06,
      "loss": 1.3501,
      "step": 675
    },
    {
      "epoch": 1.0816,
      "grad_norm": 0.8145579820251042,
      "learning_rate": 8.080915523406371e-06,
      "loss": 1.3299,
      "step": 676
    },
    {
      "epoch": 1.0832,
      "grad_norm": 0.7740780159461662,
      "learning_rate": 8.073576671552303e-06,
      "loss": 1.318,
      "step": 677
    },
    {
      "epoch": 1.0848,
      "grad_norm": 0.8083668407718305,
      "learning_rate": 8.06622716077079e-06,
      "loss": 1.3943,
      "step": 678
    },
    {
      "epoch": 1.0864,
      "grad_norm": 0.8024137380762816,
      "learning_rate": 8.058867016549372e-06,
      "loss": 1.3382,
      "step": 679
    },
    {
      "epoch": 1.088,
      "grad_norm": 0.7836677376892376,
      "learning_rate": 8.051496264412464e-06,
      "loss": 1.3768,
      "step": 680
    },
    {
      "epoch": 1.0896,
      "grad_norm": 0.7395094392246157,
      "learning_rate": 8.044114929921264e-06,
      "loss": 1.3475,
      "step": 681
    },
    {
      "epoch": 1.0912,
      "grad_norm": 0.7331350780141876,
      "learning_rate": 8.036723038673675e-06,
      "loss": 1.3439,
      "step": 682
    },
    {
      "epoch": 1.0928,
      "grad_norm": 0.8164491946025054,
      "learning_rate": 8.029320616304204e-06,
      "loss": 1.316,
      "step": 683
    },
    {
      "epoch": 1.0944,
      "grad_norm": 0.747202719455133,
      "learning_rate": 8.021907688483885e-06,
      "loss": 1.3595,
      "step": 684
    },
    {
      "epoch": 1.096,
      "grad_norm": 0.7736271859108658,
      "learning_rate": 8.01448428092018e-06,
      "loss": 1.3491,
      "step": 685
    },
    {
      "epoch": 1.0976,
      "grad_norm": 0.8183977397275276,
      "learning_rate": 8.007050419356898e-06,
      "loss": 1.3559,
      "step": 686
    },
    {
      "epoch": 1.0992,
      "grad_norm": 0.7928870588512775,
      "learning_rate": 7.999606129574096e-06,
      "loss": 1.4129,
      "step": 687
    },
    {
      "epoch": 1.1008,
      "grad_norm": 0.7926982590285653,
      "learning_rate": 7.992151437387999e-06,
      "loss": 1.4282,
      "step": 688
    },
    {
      "epoch": 1.1024,
      "grad_norm": 0.7693618378305822,
      "learning_rate": 7.984686368650907e-06,
      "loss": 1.3615,
      "step": 689
    },
    {
      "epoch": 1.104,
      "grad_norm": 0.7412141344590744,
      "learning_rate": 7.977210949251102e-06,
      "loss": 1.338,
      "step": 690
    },
    {
      "epoch": 1.1056,
      "grad_norm": 0.7389328302373994,
      "learning_rate": 7.969725205112766e-06,
      "loss": 1.4057,
      "step": 691
    },
    {
      "epoch": 1.1072,
      "grad_norm": 0.7617373078736182,
      "learning_rate": 7.962229162195882e-06,
      "loss": 1.3873,
      "step": 692
    },
    {
      "epoch": 1.1088,
      "grad_norm": 0.729248645775749,
      "learning_rate": 7.95472284649615e-06,
      "loss": 1.3692,
      "step": 693
    },
    {
      "epoch": 1.1104,
      "grad_norm": 0.8604183104354819,
      "learning_rate": 7.947206284044896e-06,
      "loss": 1.3568,
      "step": 694
    },
    {
      "epoch": 1.112,
      "grad_norm": 0.7798548902695377,
      "learning_rate": 7.939679500908982e-06,
      "loss": 1.3771,
      "step": 695
    },
    {
      "epoch": 1.1136,
      "grad_norm": 0.7910556047646463,
      "learning_rate": 7.932142523190711e-06,
      "loss": 1.3518,
      "step": 696
    },
    {
      "epoch": 1.1152,
      "grad_norm": 0.874313637607216,
      "learning_rate": 7.924595377027741e-06,
      "loss": 1.3426,
      "step": 697
    },
    {
      "epoch": 1.1168,
      "grad_norm": 0.7637458501054691,
      "learning_rate": 7.917038088592997e-06,
      "loss": 1.3898,
      "step": 698
    },
    {
      "epoch": 1.1184,
      "grad_norm": 0.9305614617739613,
      "learning_rate": 7.90947068409457e-06,
      "loss": 1.4444,
      "step": 699
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.742542249090471,
      "learning_rate": 7.90189318977564e-06,
      "loss": 1.3873,
      "step": 700
    },
    {
      "epoch": 1.1216,
      "grad_norm": 0.9192954994931757,
      "learning_rate": 7.894305631914373e-06,
      "loss": 1.3535,
      "step": 701
    },
    {
      "epoch": 1.1232,
      "grad_norm": 0.7495806826880576,
      "learning_rate": 7.886708036823838e-06,
      "loss": 1.3988,
      "step": 702
    },
    {
      "epoch": 1.1248,
      "grad_norm": 0.8504204663825817,
      "learning_rate": 7.879100430851907e-06,
      "loss": 1.3436,
      "step": 703
    },
    {
      "epoch": 1.1264,
      "grad_norm": 0.7801427597455125,
      "learning_rate": 7.871482840381174e-06,
      "loss": 1.3448,
      "step": 704
    },
    {
      "epoch": 1.1280000000000001,
      "grad_norm": 0.7631946685156605,
      "learning_rate": 7.863855291828857e-06,
      "loss": 1.375,
      "step": 705
    },
    {
      "epoch": 1.1296,
      "grad_norm": 0.7866244034913704,
      "learning_rate": 7.856217811646707e-06,
      "loss": 1.4204,
      "step": 706
    },
    {
      "epoch": 1.1312,
      "grad_norm": 0.7928455384484502,
      "learning_rate": 7.848570426320918e-06,
      "loss": 1.3388,
      "step": 707
    },
    {
      "epoch": 1.1328,
      "grad_norm": 0.7714435670415402,
      "learning_rate": 7.840913162372032e-06,
      "loss": 1.3771,
      "step": 708
    },
    {
      "epoch": 1.1344,
      "grad_norm": 0.8016213805150575,
      "learning_rate": 7.833246046354856e-06,
      "loss": 1.3749,
      "step": 709
    },
    {
      "epoch": 1.1360000000000001,
      "grad_norm": 0.7123025254119106,
      "learning_rate": 7.825569104858353e-06,
      "loss": 1.3486,
      "step": 710
    },
    {
      "epoch": 1.1376,
      "grad_norm": 0.7906010555766028,
      "learning_rate": 7.81788236450557e-06,
      "loss": 1.4096,
      "step": 711
    },
    {
      "epoch": 1.1392,
      "grad_norm": 0.7814295143668982,
      "learning_rate": 7.810185851953529e-06,
      "loss": 1.4216,
      "step": 712
    },
    {
      "epoch": 1.1408,
      "grad_norm": 0.741264684501341,
      "learning_rate": 7.802479593893142e-06,
      "loss": 1.4602,
      "step": 713
    },
    {
      "epoch": 1.1424,
      "grad_norm": 0.7691575381996882,
      "learning_rate": 7.794763617049124e-06,
      "loss": 1.3517,
      "step": 714
    },
    {
      "epoch": 1.144,
      "grad_norm": 0.7158291664331945,
      "learning_rate": 7.787037948179884e-06,
      "loss": 1.3564,
      "step": 715
    },
    {
      "epoch": 1.1456,
      "grad_norm": 0.7542025273509244,
      "learning_rate": 7.779302614077449e-06,
      "loss": 1.4017,
      "step": 716
    },
    {
      "epoch": 1.1472,
      "grad_norm": 0.7462950617594832,
      "learning_rate": 7.771557641567363e-06,
      "loss": 1.3772,
      "step": 717
    },
    {
      "epoch": 1.1488,
      "grad_norm": 0.7901488820601676,
      "learning_rate": 7.763803057508594e-06,
      "loss": 1.3158,
      "step": 718
    },
    {
      "epoch": 1.1504,
      "grad_norm": 0.7309146558121339,
      "learning_rate": 7.756038888793446e-06,
      "loss": 1.4564,
      "step": 719
    },
    {
      "epoch": 1.152,
      "grad_norm": 0.8426630879791774,
      "learning_rate": 7.748265162347455e-06,
      "loss": 1.4254,
      "step": 720
    },
    {
      "epoch": 1.1536,
      "grad_norm": 0.7248386325192862,
      "learning_rate": 7.740481905129307e-06,
      "loss": 1.3865,
      "step": 721
    },
    {
      "epoch": 1.1552,
      "grad_norm": 0.8182787363415237,
      "learning_rate": 7.732689144130741e-06,
      "loss": 1.384,
      "step": 722
    },
    {
      "epoch": 1.1568,
      "grad_norm": 0.7741630134920995,
      "learning_rate": 7.724886906376451e-06,
      "loss": 1.3618,
      "step": 723
    },
    {
      "epoch": 1.1584,
      "grad_norm": 2.4002187079438624,
      "learning_rate": 7.717075218923998e-06,
      "loss": 1.3236,
      "step": 724
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0318362010721844,
      "learning_rate": 7.709254108863714e-06,
      "loss": 1.4357,
      "step": 725
    },
    {
      "epoch": 1.1616,
      "grad_norm": 0.9116798695452414,
      "learning_rate": 7.701423603318605e-06,
      "loss": 1.3778,
      "step": 726
    },
    {
      "epoch": 1.1632,
      "grad_norm": 0.7845499288767877,
      "learning_rate": 7.693583729444263e-06,
      "loss": 1.3375,
      "step": 727
    },
    {
      "epoch": 1.1648,
      "grad_norm": 1.0276272419690058,
      "learning_rate": 7.685734514428767e-06,
      "loss": 1.3799,
      "step": 728
    },
    {
      "epoch": 1.1663999999999999,
      "grad_norm": 0.7224929020968004,
      "learning_rate": 7.677875985492591e-06,
      "loss": 1.293,
      "step": 729
    },
    {
      "epoch": 1.168,
      "grad_norm": 0.9931955372721557,
      "learning_rate": 7.67000816988851e-06,
      "loss": 1.36,
      "step": 730
    },
    {
      "epoch": 1.1696,
      "grad_norm": 0.7581155424282562,
      "learning_rate": 7.662131094901499e-06,
      "loss": 1.3531,
      "step": 731
    },
    {
      "epoch": 1.1712,
      "grad_norm": 0.872222412601772,
      "learning_rate": 7.654244787848655e-06,
      "loss": 1.3309,
      "step": 732
    },
    {
      "epoch": 1.1728,
      "grad_norm": 0.7761560909084766,
      "learning_rate": 7.646349276079079e-06,
      "loss": 1.3233,
      "step": 733
    },
    {
      "epoch": 1.1743999999999999,
      "grad_norm": 0.7467073512857008,
      "learning_rate": 7.6384445869738e-06,
      "loss": 1.3195,
      "step": 734
    },
    {
      "epoch": 1.176,
      "grad_norm": 0.7818789468408512,
      "learning_rate": 7.630530747945672e-06,
      "loss": 1.3969,
      "step": 735
    },
    {
      "epoch": 1.1776,
      "grad_norm": 0.7445546873526807,
      "learning_rate": 7.622607786439279e-06,
      "loss": 1.3904,
      "step": 736
    },
    {
      "epoch": 1.1792,
      "grad_norm": 0.7099516262415123,
      "learning_rate": 7.6146757299308406e-06,
      "loss": 1.3777,
      "step": 737
    },
    {
      "epoch": 1.1808,
      "grad_norm": 0.7646974116797374,
      "learning_rate": 7.606734605928123e-06,
      "loss": 1.3876,
      "step": 738
    },
    {
      "epoch": 1.1824,
      "grad_norm": 0.7232498502052798,
      "learning_rate": 7.598784441970329e-06,
      "loss": 1.3809,
      "step": 739
    },
    {
      "epoch": 1.184,
      "grad_norm": 0.7590941178255343,
      "learning_rate": 7.590825265628019e-06,
      "loss": 1.3716,
      "step": 740
    },
    {
      "epoch": 1.1856,
      "grad_norm": 0.7563588932293994,
      "learning_rate": 7.5828571045030005e-06,
      "loss": 1.3238,
      "step": 741
    },
    {
      "epoch": 1.1872,
      "grad_norm": 0.7565596720587426,
      "learning_rate": 7.574879986228245e-06,
      "loss": 1.2789,
      "step": 742
    },
    {
      "epoch": 1.1888,
      "grad_norm": 0.759219376803243,
      "learning_rate": 7.566893938467788e-06,
      "loss": 1.3831,
      "step": 743
    },
    {
      "epoch": 1.1904,
      "grad_norm": 0.775665975744835,
      "learning_rate": 7.558898988916624e-06,
      "loss": 1.3161,
      "step": 744
    },
    {
      "epoch": 1.192,
      "grad_norm": 0.7280735283848119,
      "learning_rate": 7.550895165300626e-06,
      "loss": 1.3959,
      "step": 745
    },
    {
      "epoch": 1.1936,
      "grad_norm": 0.8013573877744476,
      "learning_rate": 7.542882495376437e-06,
      "loss": 1.3658,
      "step": 746
    },
    {
      "epoch": 1.1952,
      "grad_norm": 0.7432656292226191,
      "learning_rate": 7.5348610069313795e-06,
      "loss": 1.325,
      "step": 747
    },
    {
      "epoch": 1.1968,
      "grad_norm": 0.7468088904053286,
      "learning_rate": 7.5268307277833605e-06,
      "loss": 1.3927,
      "step": 748
    },
    {
      "epoch": 1.1984,
      "grad_norm": 0.7325979522285958,
      "learning_rate": 7.518791685780769e-06,
      "loss": 1.3435,
      "step": 749
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7278360201192645,
      "learning_rate": 7.5107439088023845e-06,
      "loss": 1.3892,
      "step": 750
    },
    {
      "epoch": 1.2016,
      "grad_norm": 0.741773423991486,
      "learning_rate": 7.502687424757278e-06,
      "loss": 1.342,
      "step": 751
    },
    {
      "epoch": 1.2032,
      "grad_norm": 0.7454249140298557,
      "learning_rate": 7.4946222615847165e-06,
      "loss": 1.3623,
      "step": 752
    },
    {
      "epoch": 1.2048,
      "grad_norm": 0.7104091453479872,
      "learning_rate": 7.486548447254065e-06,
      "loss": 1.3226,
      "step": 753
    },
    {
      "epoch": 1.2064,
      "grad_norm": 0.7077202115803977,
      "learning_rate": 7.478466009764692e-06,
      "loss": 1.3566,
      "step": 754
    },
    {
      "epoch": 1.208,
      "grad_norm": 0.7147369509782032,
      "learning_rate": 7.470374977145867e-06,
      "loss": 1.3323,
      "step": 755
    },
    {
      "epoch": 1.2096,
      "grad_norm": 0.7271776024431341,
      "learning_rate": 7.462275377456671e-06,
      "loss": 1.3953,
      "step": 756
    },
    {
      "epoch": 1.2112,
      "grad_norm": 1.0931331049637616,
      "learning_rate": 7.4541672387858895e-06,
      "loss": 1.387,
      "step": 757
    },
    {
      "epoch": 1.2128,
      "grad_norm": 0.7545219407391343,
      "learning_rate": 7.446050589251928e-06,
      "loss": 1.2817,
      "step": 758
    },
    {
      "epoch": 1.2144,
      "grad_norm": 0.7445715061573082,
      "learning_rate": 7.437925457002697e-06,
      "loss": 1.448,
      "step": 759
    },
    {
      "epoch": 1.216,
      "grad_norm": 0.7529341261415462,
      "learning_rate": 7.429791870215535e-06,
      "loss": 1.3423,
      "step": 760
    },
    {
      "epoch": 1.2176,
      "grad_norm": 0.7779486398267906,
      "learning_rate": 7.421649857097092e-06,
      "loss": 1.4256,
      "step": 761
    },
    {
      "epoch": 1.2192,
      "grad_norm": 0.8450495500262457,
      "learning_rate": 7.413499445883245e-06,
      "loss": 1.3624,
      "step": 762
    },
    {
      "epoch": 1.2208,
      "grad_norm": 0.778354957309376,
      "learning_rate": 7.405340664838994e-06,
      "loss": 1.4209,
      "step": 763
    },
    {
      "epoch": 1.2224,
      "grad_norm": 0.8324403651705291,
      "learning_rate": 7.39717354225836e-06,
      "loss": 1.3573,
      "step": 764
    },
    {
      "epoch": 1.224,
      "grad_norm": 0.7817850870158165,
      "learning_rate": 7.3889981064643e-06,
      "loss": 1.3889,
      "step": 765
    },
    {
      "epoch": 1.2256,
      "grad_norm": 0.8329571654274978,
      "learning_rate": 7.380814385808594e-06,
      "loss": 1.3488,
      "step": 766
    },
    {
      "epoch": 1.2272,
      "grad_norm": 0.7845592542276563,
      "learning_rate": 7.372622408671757e-06,
      "loss": 1.3955,
      "step": 767
    },
    {
      "epoch": 1.2288000000000001,
      "grad_norm": 0.8934833279038007,
      "learning_rate": 7.364422203462935e-06,
      "loss": 1.3823,
      "step": 768
    },
    {
      "epoch": 1.2304,
      "grad_norm": 0.7994241585395614,
      "learning_rate": 7.3562137986198065e-06,
      "loss": 1.3632,
      "step": 769
    },
    {
      "epoch": 1.232,
      "grad_norm": 0.8925232683800116,
      "learning_rate": 7.3479972226084925e-06,
      "loss": 1.3679,
      "step": 770
    },
    {
      "epoch": 1.2336,
      "grad_norm": 0.7512662160308877,
      "learning_rate": 7.339772503923445e-06,
      "loss": 1.3652,
      "step": 771
    },
    {
      "epoch": 1.2352,
      "grad_norm": 0.7718626190679205,
      "learning_rate": 7.331539671087353e-06,
      "loss": 1.353,
      "step": 772
    },
    {
      "epoch": 1.2368000000000001,
      "grad_norm": 0.804369153204895,
      "learning_rate": 7.32329875265105e-06,
      "loss": 1.3398,
      "step": 773
    },
    {
      "epoch": 1.2384,
      "grad_norm": 0.7669373016486996,
      "learning_rate": 7.315049777193407e-06,
      "loss": 1.3526,
      "step": 774
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.75792281131365,
      "learning_rate": 7.306792773321234e-06,
      "loss": 1.381,
      "step": 775
    },
    {
      "epoch": 1.2416,
      "grad_norm": 0.751817080221283,
      "learning_rate": 7.298527769669188e-06,
      "loss": 1.4017,
      "step": 776
    },
    {
      "epoch": 1.2432,
      "grad_norm": 0.8592753261116628,
      "learning_rate": 7.290254794899665e-06,
      "loss": 1.3435,
      "step": 777
    },
    {
      "epoch": 1.2448,
      "grad_norm": 0.7212824217202826,
      "learning_rate": 7.281973877702705e-06,
      "loss": 1.3618,
      "step": 778
    },
    {
      "epoch": 1.2464,
      "grad_norm": 0.8323386249737852,
      "learning_rate": 7.2736850467958905e-06,
      "loss": 1.3815,
      "step": 779
    },
    {
      "epoch": 1.248,
      "grad_norm": 0.7314717978027695,
      "learning_rate": 7.26538833092425e-06,
      "loss": 1.3551,
      "step": 780
    },
    {
      "epoch": 1.2496,
      "grad_norm": 0.7727914343659431,
      "learning_rate": 7.257083758860159e-06,
      "loss": 1.3432,
      "step": 781
    },
    {
      "epoch": 1.2511999999999999,
      "grad_norm": 0.7911191291810185,
      "learning_rate": 7.248771359403231e-06,
      "loss": 1.4356,
      "step": 782
    },
    {
      "epoch": 1.2528000000000001,
      "grad_norm": 0.823102926814678,
      "learning_rate": 7.240451161380226e-06,
      "loss": 1.2746,
      "step": 783
    },
    {
      "epoch": 1.2544,
      "grad_norm": 0.7748014580789487,
      "learning_rate": 7.232123193644957e-06,
      "loss": 1.3576,
      "step": 784
    },
    {
      "epoch": 1.256,
      "grad_norm": 0.8036140514185841,
      "learning_rate": 7.22378748507817e-06,
      "loss": 1.4015,
      "step": 785
    },
    {
      "epoch": 1.2576,
      "grad_norm": 0.8358193135779624,
      "learning_rate": 7.215444064587462e-06,
      "loss": 1.4042,
      "step": 786
    },
    {
      "epoch": 1.2591999999999999,
      "grad_norm": 0.8797390059883692,
      "learning_rate": 7.207092961107176e-06,
      "loss": 1.3196,
      "step": 787
    },
    {
      "epoch": 1.2608,
      "grad_norm": 0.8840652233922214,
      "learning_rate": 7.198734203598294e-06,
      "loss": 1.4043,
      "step": 788
    },
    {
      "epoch": 1.2624,
      "grad_norm": 0.8538805119034071,
      "learning_rate": 7.190367821048346e-06,
      "loss": 1.409,
      "step": 789
    },
    {
      "epoch": 1.264,
      "grad_norm": 0.7854726465136568,
      "learning_rate": 7.181993842471301e-06,
      "loss": 1.3445,
      "step": 790
    },
    {
      "epoch": 1.2656,
      "grad_norm": 0.8183149080812712,
      "learning_rate": 7.173612296907473e-06,
      "loss": 1.3607,
      "step": 791
    },
    {
      "epoch": 1.2671999999999999,
      "grad_norm": 0.777660129469255,
      "learning_rate": 7.165223213423416e-06,
      "loss": 1.3877,
      "step": 792
    },
    {
      "epoch": 1.2688,
      "grad_norm": 0.8574468206985288,
      "learning_rate": 7.15682662111183e-06,
      "loss": 1.4017,
      "step": 793
    },
    {
      "epoch": 1.2704,
      "grad_norm": 0.806747276732166,
      "learning_rate": 7.148422549091447e-06,
      "loss": 1.4534,
      "step": 794
    },
    {
      "epoch": 1.272,
      "grad_norm": 0.8120798907269753,
      "learning_rate": 7.140011026506945e-06,
      "loss": 1.3056,
      "step": 795
    },
    {
      "epoch": 1.2736,
      "grad_norm": 0.8300395352324531,
      "learning_rate": 7.131592082528837e-06,
      "loss": 1.3398,
      "step": 796
    },
    {
      "epoch": 1.2752,
      "grad_norm": 0.7767168200780726,
      "learning_rate": 7.12316574635337e-06,
      "loss": 1.3944,
      "step": 797
    },
    {
      "epoch": 1.2768,
      "grad_norm": 0.7067847369009882,
      "learning_rate": 7.114732047202433e-06,
      "loss": 1.2888,
      "step": 798
    },
    {
      "epoch": 1.2784,
      "grad_norm": 0.8199210485983225,
      "learning_rate": 7.106291014323445e-06,
      "loss": 1.2914,
      "step": 799
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7157213731351526,
      "learning_rate": 7.0978426769892585e-06,
      "loss": 1.316,
      "step": 800
    },
    {
      "epoch": 1.2816,
      "grad_norm": 0.6970526940589632,
      "learning_rate": 7.089387064498057e-06,
      "loss": 1.3469,
      "step": 801
    },
    {
      "epoch": 1.2832,
      "grad_norm": 0.7708469235535266,
      "learning_rate": 7.080924206173253e-06,
      "loss": 1.3436,
      "step": 802
    },
    {
      "epoch": 1.2848,
      "grad_norm": 0.6898944038374157,
      "learning_rate": 7.072454131363391e-06,
      "loss": 1.371,
      "step": 803
    },
    {
      "epoch": 1.2864,
      "grad_norm": 0.7534866875319569,
      "learning_rate": 7.063976869442037e-06,
      "loss": 1.3652,
      "step": 804
    },
    {
      "epoch": 1.288,
      "grad_norm": 0.7148983703622245,
      "learning_rate": 7.055492449807684e-06,
      "loss": 1.291,
      "step": 805
    },
    {
      "epoch": 1.2896,
      "grad_norm": 0.6976330176177933,
      "learning_rate": 7.047000901883646e-06,
      "loss": 1.441,
      "step": 806
    },
    {
      "epoch": 1.2912,
      "grad_norm": 0.7091074861341291,
      "learning_rate": 7.038502255117957e-06,
      "loss": 1.4306,
      "step": 807
    },
    {
      "epoch": 1.2928,
      "grad_norm": 0.6957121070969913,
      "learning_rate": 7.029996538983273e-06,
      "loss": 1.272,
      "step": 808
    },
    {
      "epoch": 1.2944,
      "grad_norm": 0.7096065507573124,
      "learning_rate": 7.021483782976759e-06,
      "loss": 1.4224,
      "step": 809
    },
    {
      "epoch": 1.296,
      "grad_norm": 0.6992297917886426,
      "learning_rate": 7.012964016620002e-06,
      "loss": 1.3632,
      "step": 810
    },
    {
      "epoch": 1.2976,
      "grad_norm": 0.721687177825011,
      "learning_rate": 7.004437269458894e-06,
      "loss": 1.4201,
      "step": 811
    },
    {
      "epoch": 1.2992,
      "grad_norm": 0.6973114387227252,
      "learning_rate": 6.995903571063541e-06,
      "loss": 1.3713,
      "step": 812
    },
    {
      "epoch": 1.3008,
      "grad_norm": 0.7564383241152022,
      "learning_rate": 6.987362951028147e-06,
      "loss": 1.3535,
      "step": 813
    },
    {
      "epoch": 1.3024,
      "grad_norm": 0.6685741926106691,
      "learning_rate": 6.97881543897093e-06,
      "loss": 1.3557,
      "step": 814
    },
    {
      "epoch": 1.304,
      "grad_norm": 0.726003594420146,
      "learning_rate": 6.970261064534003e-06,
      "loss": 1.3304,
      "step": 815
    },
    {
      "epoch": 1.3056,
      "grad_norm": 0.6750529519744409,
      "learning_rate": 6.961699857383279e-06,
      "loss": 1.3409,
      "step": 816
    },
    {
      "epoch": 1.3072,
      "grad_norm": 0.7852242223889824,
      "learning_rate": 6.953131847208365e-06,
      "loss": 1.381,
      "step": 817
    },
    {
      "epoch": 1.3088,
      "grad_norm": 0.6995750668225412,
      "learning_rate": 6.944557063722459e-06,
      "loss": 1.3911,
      "step": 818
    },
    {
      "epoch": 1.3104,
      "grad_norm": 0.6873299109124195,
      "learning_rate": 6.935975536662254e-06,
      "loss": 1.3044,
      "step": 819
    },
    {
      "epoch": 1.312,
      "grad_norm": 0.7593297114762613,
      "learning_rate": 6.9273872957878255e-06,
      "loss": 1.3231,
      "step": 820
    },
    {
      "epoch": 1.3136,
      "grad_norm": 0.6645022236307521,
      "learning_rate": 6.91879237088253e-06,
      "loss": 1.3259,
      "step": 821
    },
    {
      "epoch": 1.3152,
      "grad_norm": 0.7164031201199415,
      "learning_rate": 6.910190791752907e-06,
      "loss": 1.2565,
      "step": 822
    },
    {
      "epoch": 1.3168,
      "grad_norm": 0.7345749106388253,
      "learning_rate": 6.90158258822857e-06,
      "loss": 1.3424,
      "step": 823
    },
    {
      "epoch": 1.3184,
      "grad_norm": 0.680184125497553,
      "learning_rate": 6.892967790162109e-06,
      "loss": 1.3683,
      "step": 824
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.6788959698100342,
      "learning_rate": 6.884346427428978e-06,
      "loss": 1.3622,
      "step": 825
    },
    {
      "epoch": 1.3216,
      "grad_norm": 0.7211854388706463,
      "learning_rate": 6.875718529927404e-06,
      "loss": 1.377,
      "step": 826
    },
    {
      "epoch": 1.3232,
      "grad_norm": 0.7403104391655243,
      "learning_rate": 6.867084127578267e-06,
      "loss": 1.3403,
      "step": 827
    },
    {
      "epoch": 1.3248,
      "grad_norm": 0.7059079669560991,
      "learning_rate": 6.858443250325013e-06,
      "loss": 1.3576,
      "step": 828
    },
    {
      "epoch": 1.3264,
      "grad_norm": 0.7043290103274691,
      "learning_rate": 6.849795928133538e-06,
      "loss": 1.4027,
      "step": 829
    },
    {
      "epoch": 1.328,
      "grad_norm": 0.713622154064674,
      "learning_rate": 6.841142190992092e-06,
      "loss": 1.3763,
      "step": 830
    },
    {
      "epoch": 1.3296000000000001,
      "grad_norm": 0.7287400621877576,
      "learning_rate": 6.832482068911167e-06,
      "loss": 1.2791,
      "step": 831
    },
    {
      "epoch": 1.3312,
      "grad_norm": 0.7212105240393936,
      "learning_rate": 6.823815591923402e-06,
      "loss": 1.3398,
      "step": 832
    },
    {
      "epoch": 1.3328,
      "grad_norm": 0.7531466460783768,
      "learning_rate": 6.815142790083473e-06,
      "loss": 1.3338,
      "step": 833
    },
    {
      "epoch": 1.3344,
      "grad_norm": 0.7587316562394139,
      "learning_rate": 6.8064636934679885e-06,
      "loss": 1.3455,
      "step": 834
    },
    {
      "epoch": 1.336,
      "grad_norm": 0.6810905479386465,
      "learning_rate": 6.797778332175387e-06,
      "loss": 1.3048,
      "step": 835
    },
    {
      "epoch": 1.3376000000000001,
      "grad_norm": 0.7353895967508766,
      "learning_rate": 6.789086736325834e-06,
      "loss": 1.2948,
      "step": 836
    },
    {
      "epoch": 1.3392,
      "grad_norm": 0.697751559311748,
      "learning_rate": 6.780388936061118e-06,
      "loss": 1.3051,
      "step": 837
    },
    {
      "epoch": 1.3408,
      "grad_norm": 0.6793385547756493,
      "learning_rate": 6.771684961544537e-06,
      "loss": 1.3154,
      "step": 838
    },
    {
      "epoch": 1.3424,
      "grad_norm": 0.7282034335134697,
      "learning_rate": 6.7629748429608076e-06,
      "loss": 1.3434,
      "step": 839
    },
    {
      "epoch": 1.3439999999999999,
      "grad_norm": 0.7158689286546648,
      "learning_rate": 6.754258610515949e-06,
      "loss": 1.3062,
      "step": 840
    },
    {
      "epoch": 1.3456000000000001,
      "grad_norm": 0.7557816107697163,
      "learning_rate": 6.745536294437187e-06,
      "loss": 1.4072,
      "step": 841
    },
    {
      "epoch": 1.3472,
      "grad_norm": 0.7135476568863266,
      "learning_rate": 6.736807924972841e-06,
      "loss": 1.2773,
      "step": 842
    },
    {
      "epoch": 1.3488,
      "grad_norm": 0.6738951825160933,
      "learning_rate": 6.728073532392226e-06,
      "loss": 1.3768,
      "step": 843
    },
    {
      "epoch": 1.3504,
      "grad_norm": 0.689584833973898,
      "learning_rate": 6.719333146985544e-06,
      "loss": 1.3506,
      "step": 844
    },
    {
      "epoch": 1.3519999999999999,
      "grad_norm": 0.726017517359451,
      "learning_rate": 6.710586799063777e-06,
      "loss": 1.3394,
      "step": 845
    },
    {
      "epoch": 1.3536000000000001,
      "grad_norm": 0.7205585523282458,
      "learning_rate": 6.701834518958587e-06,
      "loss": 1.3608,
      "step": 846
    },
    {
      "epoch": 1.3552,
      "grad_norm": 0.7260632758182276,
      "learning_rate": 6.6930763370222104e-06,
      "loss": 1.3246,
      "step": 847
    },
    {
      "epoch": 1.3568,
      "grad_norm": 0.7083743169897397,
      "learning_rate": 6.684312283627348e-06,
      "loss": 1.4151,
      "step": 848
    },
    {
      "epoch": 1.3584,
      "grad_norm": 0.7168398366841028,
      "learning_rate": 6.6755423891670605e-06,
      "loss": 1.3088,
      "step": 849
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 0.6477882462701957,
      "learning_rate": 6.6667666840546685e-06,
      "loss": 1.3812,
      "step": 850
    },
    {
      "epoch": 1.3616,
      "grad_norm": 0.7301144687064014,
      "learning_rate": 6.6579851987236435e-06,
      "loss": 1.3588,
      "step": 851
    },
    {
      "epoch": 1.3632,
      "grad_norm": 0.6878904250976322,
      "learning_rate": 6.649197963627497e-06,
      "loss": 1.3327,
      "step": 852
    },
    {
      "epoch": 1.3648,
      "grad_norm": 0.6921448715420281,
      "learning_rate": 6.640405009239689e-06,
      "loss": 1.4181,
      "step": 853
    },
    {
      "epoch": 1.3664,
      "grad_norm": 0.6879827136786149,
      "learning_rate": 6.631606366053507e-06,
      "loss": 1.3325,
      "step": 854
    },
    {
      "epoch": 1.3679999999999999,
      "grad_norm": 0.6872472649689411,
      "learning_rate": 6.622802064581968e-06,
      "loss": 1.3996,
      "step": 855
    },
    {
      "epoch": 1.3696,
      "grad_norm": 0.7185254981641624,
      "learning_rate": 6.613992135357713e-06,
      "loss": 1.3766,
      "step": 856
    },
    {
      "epoch": 1.3712,
      "grad_norm": 0.6912943762378405,
      "learning_rate": 6.605176608932897e-06,
      "loss": 1.3075,
      "step": 857
    },
    {
      "epoch": 1.3728,
      "grad_norm": 0.7469813189256428,
      "learning_rate": 6.596355515879091e-06,
      "loss": 1.3839,
      "step": 858
    },
    {
      "epoch": 1.3744,
      "grad_norm": 0.7127267206968712,
      "learning_rate": 6.587528886787165e-06,
      "loss": 1.3435,
      "step": 859
    },
    {
      "epoch": 1.376,
      "grad_norm": 0.6961482690013907,
      "learning_rate": 6.578696752267189e-06,
      "loss": 1.3529,
      "step": 860
    },
    {
      "epoch": 1.3776,
      "grad_norm": 0.674717373603138,
      "learning_rate": 6.5698591429483286e-06,
      "loss": 1.3442,
      "step": 861
    },
    {
      "epoch": 1.3792,
      "grad_norm": 0.6961293242877993,
      "learning_rate": 6.5610160894787275e-06,
      "loss": 1.3074,
      "step": 862
    },
    {
      "epoch": 1.3808,
      "grad_norm": 0.6689252330525438,
      "learning_rate": 6.552167622525421e-06,
      "loss": 1.3627,
      "step": 863
    },
    {
      "epoch": 1.3824,
      "grad_norm": 0.8104731165409661,
      "learning_rate": 6.543313772774209e-06,
      "loss": 1.3386,
      "step": 864
    },
    {
      "epoch": 1.384,
      "grad_norm": 0.666617680454017,
      "learning_rate": 6.534454570929563e-06,
      "loss": 1.3656,
      "step": 865
    },
    {
      "epoch": 1.3856,
      "grad_norm": 0.7464447078927323,
      "learning_rate": 6.52559004771451e-06,
      "loss": 1.3697,
      "step": 866
    },
    {
      "epoch": 1.3872,
      "grad_norm": 0.6541947297935203,
      "learning_rate": 6.516720233870538e-06,
      "loss": 1.3597,
      "step": 867
    },
    {
      "epoch": 1.3888,
      "grad_norm": 0.7891058402786534,
      "learning_rate": 6.507845160157476e-06,
      "loss": 1.4137,
      "step": 868
    },
    {
      "epoch": 1.3904,
      "grad_norm": 0.6870982686774288,
      "learning_rate": 6.498964857353401e-06,
      "loss": 1.3741,
      "step": 869
    },
    {
      "epoch": 1.392,
      "grad_norm": 0.7702955114808983,
      "learning_rate": 6.4900793562545165e-06,
      "loss": 1.3846,
      "step": 870
    },
    {
      "epoch": 1.3936,
      "grad_norm": 0.6503739688435445,
      "learning_rate": 6.481188687675057e-06,
      "loss": 1.3172,
      "step": 871
    },
    {
      "epoch": 1.3952,
      "grad_norm": 0.7357237050500962,
      "learning_rate": 6.47229288244718e-06,
      "loss": 1.3414,
      "step": 872
    },
    {
      "epoch": 1.3968,
      "grad_norm": 0.7572214279980704,
      "learning_rate": 6.46339197142085e-06,
      "loss": 1.3863,
      "step": 873
    },
    {
      "epoch": 1.3984,
      "grad_norm": 0.7443026770636113,
      "learning_rate": 6.454485985463742e-06,
      "loss": 1.3861,
      "step": 874
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7177320973083492,
      "learning_rate": 6.445574955461134e-06,
      "loss": 1.3694,
      "step": 875
    },
    {
      "epoch": 1.4016,
      "grad_norm": 0.734102706338454,
      "learning_rate": 6.436658912315789e-06,
      "loss": 1.3992,
      "step": 876
    },
    {
      "epoch": 1.4032,
      "grad_norm": 0.6831886160183082,
      "learning_rate": 6.427737886947859e-06,
      "loss": 1.3691,
      "step": 877
    },
    {
      "epoch": 1.4048,
      "grad_norm": 0.6855082363610441,
      "learning_rate": 6.418811910294776e-06,
      "loss": 1.3754,
      "step": 878
    },
    {
      "epoch": 1.4064,
      "grad_norm": 0.6887410601192074,
      "learning_rate": 6.409881013311136e-06,
      "loss": 1.3309,
      "step": 879
    },
    {
      "epoch": 1.408,
      "grad_norm": 0.7021251933253706,
      "learning_rate": 6.400945226968607e-06,
      "loss": 1.3191,
      "step": 880
    },
    {
      "epoch": 1.4096,
      "grad_norm": 0.7747924062782494,
      "learning_rate": 6.392004582255807e-06,
      "loss": 1.3837,
      "step": 881
    },
    {
      "epoch": 1.4112,
      "grad_norm": 0.6737494145220029,
      "learning_rate": 6.383059110178205e-06,
      "loss": 1.3491,
      "step": 882
    },
    {
      "epoch": 1.4128,
      "grad_norm": 0.7078319320250652,
      "learning_rate": 6.374108841758006e-06,
      "loss": 1.3244,
      "step": 883
    },
    {
      "epoch": 1.4144,
      "grad_norm": 0.7223266840655397,
      "learning_rate": 6.365153808034057e-06,
      "loss": 1.3623,
      "step": 884
    },
    {
      "epoch": 1.416,
      "grad_norm": 0.7667532584393245,
      "learning_rate": 6.356194040061725e-06,
      "loss": 1.3562,
      "step": 885
    },
    {
      "epoch": 1.4176,
      "grad_norm": 0.7111267001664232,
      "learning_rate": 6.3472295689127946e-06,
      "loss": 1.3145,
      "step": 886
    },
    {
      "epoch": 1.4192,
      "grad_norm": 0.7443627073231872,
      "learning_rate": 6.338260425675365e-06,
      "loss": 1.2655,
      "step": 887
    },
    {
      "epoch": 1.4208,
      "grad_norm": 0.739016687895461,
      "learning_rate": 6.329286641453729e-06,
      "loss": 1.372,
      "step": 888
    },
    {
      "epoch": 1.4224,
      "grad_norm": 0.6758251787185652,
      "learning_rate": 6.320308247368285e-06,
      "loss": 1.353,
      "step": 889
    },
    {
      "epoch": 1.424,
      "grad_norm": 0.8280580997794049,
      "learning_rate": 6.311325274555413e-06,
      "loss": 1.3351,
      "step": 890
    },
    {
      "epoch": 1.4256,
      "grad_norm": 0.6534083117117712,
      "learning_rate": 6.302337754167369e-06,
      "loss": 1.3904,
      "step": 891
    },
    {
      "epoch": 1.4272,
      "grad_norm": 0.85199355458959,
      "learning_rate": 6.2933457173721855e-06,
      "loss": 1.3965,
      "step": 892
    },
    {
      "epoch": 1.4288,
      "grad_norm": 0.6716053333293076,
      "learning_rate": 6.2843491953535515e-06,
      "loss": 1.3693,
      "step": 893
    },
    {
      "epoch": 1.4304000000000001,
      "grad_norm": 0.7452582273569118,
      "learning_rate": 6.275348219310715e-06,
      "loss": 1.3257,
      "step": 894
    },
    {
      "epoch": 1.432,
      "grad_norm": 0.6807691193839405,
      "learning_rate": 6.266342820458366e-06,
      "loss": 1.3796,
      "step": 895
    },
    {
      "epoch": 1.4336,
      "grad_norm": 0.8614239834856463,
      "learning_rate": 6.2573330300265375e-06,
      "loss": 1.3708,
      "step": 896
    },
    {
      "epoch": 1.4352,
      "grad_norm": 0.7180714227829532,
      "learning_rate": 6.248318879260488e-06,
      "loss": 1.4107,
      "step": 897
    },
    {
      "epoch": 1.4368,
      "grad_norm": 0.7419330800794228,
      "learning_rate": 6.239300399420601e-06,
      "loss": 1.3341,
      "step": 898
    },
    {
      "epoch": 1.4384000000000001,
      "grad_norm": 0.7262467564754542,
      "learning_rate": 6.230277621782269e-06,
      "loss": 1.3604,
      "step": 899
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7462383724568051,
      "learning_rate": 6.221250577635791e-06,
      "loss": 1.3427,
      "step": 900
    },
    {
      "epoch": 1.4416,
      "grad_norm": 0.7062295507694772,
      "learning_rate": 6.2122192982862615e-06,
      "loss": 1.4328,
      "step": 901
    },
    {
      "epoch": 1.4432,
      "grad_norm": 0.752516493939586,
      "learning_rate": 6.203183815053463e-06,
      "loss": 1.3313,
      "step": 902
    },
    {
      "epoch": 1.4447999999999999,
      "grad_norm": 0.6755664704604702,
      "learning_rate": 6.1941441592717564e-06,
      "loss": 1.3472,
      "step": 903
    },
    {
      "epoch": 1.4464000000000001,
      "grad_norm": 0.7695737898592934,
      "learning_rate": 6.185100362289972e-06,
      "loss": 1.4209,
      "step": 904
    },
    {
      "epoch": 1.448,
      "grad_norm": 0.6742707148219441,
      "learning_rate": 6.176052455471302e-06,
      "loss": 1.3267,
      "step": 905
    },
    {
      "epoch": 1.4496,
      "grad_norm": 0.7194231618799011,
      "learning_rate": 6.167000470193189e-06,
      "loss": 1.3269,
      "step": 906
    },
    {
      "epoch": 1.4512,
      "grad_norm": 0.7532899859946868,
      "learning_rate": 6.157944437847226e-06,
      "loss": 1.3077,
      "step": 907
    },
    {
      "epoch": 1.4527999999999999,
      "grad_norm": 0.6968487197832061,
      "learning_rate": 6.148884389839035e-06,
      "loss": 1.3615,
      "step": 908
    },
    {
      "epoch": 1.4544000000000001,
      "grad_norm": 0.7480552976737648,
      "learning_rate": 6.1398203575881645e-06,
      "loss": 1.3173,
      "step": 909
    },
    {
      "epoch": 1.456,
      "grad_norm": 0.7513145676580403,
      "learning_rate": 6.130752372527981e-06,
      "loss": 1.3581,
      "step": 910
    },
    {
      "epoch": 1.4576,
      "grad_norm": 0.6669267770224192,
      "learning_rate": 6.121680466105559e-06,
      "loss": 1.3398,
      "step": 911
    },
    {
      "epoch": 1.4592,
      "grad_norm": 0.7795059801974266,
      "learning_rate": 6.112604669781572e-06,
      "loss": 1.3498,
      "step": 912
    },
    {
      "epoch": 1.4607999999999999,
      "grad_norm": 0.7233254447152381,
      "learning_rate": 6.1035250150301864e-06,
      "loss": 1.3252,
      "step": 913
    },
    {
      "epoch": 1.4624,
      "grad_norm": 0.7820881901592643,
      "learning_rate": 6.0944415333389405e-06,
      "loss": 1.4227,
      "step": 914
    },
    {
      "epoch": 1.464,
      "grad_norm": 0.7296633772658974,
      "learning_rate": 6.085354256208655e-06,
      "loss": 1.3233,
      "step": 915
    },
    {
      "epoch": 1.4656,
      "grad_norm": 0.7387342915150338,
      "learning_rate": 6.076263215153308e-06,
      "loss": 1.3679,
      "step": 916
    },
    {
      "epoch": 1.4672,
      "grad_norm": 0.7008168272470795,
      "learning_rate": 6.067168441699927e-06,
      "loss": 1.3117,
      "step": 917
    },
    {
      "epoch": 1.4687999999999999,
      "grad_norm": 0.730763862598194,
      "learning_rate": 6.058069967388489e-06,
      "loss": 1.2831,
      "step": 918
    },
    {
      "epoch": 1.4704,
      "grad_norm": 0.7292236653149498,
      "learning_rate": 6.048967823771802e-06,
      "loss": 1.3062,
      "step": 919
    },
    {
      "epoch": 1.472,
      "grad_norm": 0.68705379451331,
      "learning_rate": 6.039862042415401e-06,
      "loss": 1.3823,
      "step": 920
    },
    {
      "epoch": 1.4736,
      "grad_norm": 0.8233886381221293,
      "learning_rate": 6.030752654897435e-06,
      "loss": 1.4124,
      "step": 921
    },
    {
      "epoch": 1.4752,
      "grad_norm": 0.6918919135486455,
      "learning_rate": 6.021639692808558e-06,
      "loss": 1.2692,
      "step": 922
    },
    {
      "epoch": 1.4768,
      "grad_norm": 0.7103826996563717,
      "learning_rate": 6.0125231877518205e-06,
      "loss": 1.3187,
      "step": 923
    },
    {
      "epoch": 1.4784,
      "grad_norm": 0.76047299692411,
      "learning_rate": 6.0034031713425636e-06,
      "loss": 1.3643,
      "step": 924
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.6627487133036887,
      "learning_rate": 5.994279675208302e-06,
      "loss": 1.3424,
      "step": 925
    },
    {
      "epoch": 1.4816,
      "grad_norm": 0.7308752667287031,
      "learning_rate": 5.985152730988617e-06,
      "loss": 1.3637,
      "step": 926
    },
    {
      "epoch": 1.4832,
      "grad_norm": 0.76051661136634,
      "learning_rate": 5.9760223703350495e-06,
      "loss": 1.3579,
      "step": 927
    },
    {
      "epoch": 1.4848,
      "grad_norm": 0.7311188690585912,
      "learning_rate": 5.966888624910989e-06,
      "loss": 1.307,
      "step": 928
    },
    {
      "epoch": 1.4864,
      "grad_norm": 0.6555309914404092,
      "learning_rate": 5.957751526391558e-06,
      "loss": 1.2612,
      "step": 929
    },
    {
      "epoch": 1.488,
      "grad_norm": 0.7317508785570481,
      "learning_rate": 5.948611106463518e-06,
      "loss": 1.2755,
      "step": 930
    },
    {
      "epoch": 1.4896,
      "grad_norm": 0.6848815400395859,
      "learning_rate": 5.939467396825137e-06,
      "loss": 1.3532,
      "step": 931
    },
    {
      "epoch": 1.4912,
      "grad_norm": 0.7403798958490521,
      "learning_rate": 5.9303204291860975e-06,
      "loss": 1.3269,
      "step": 932
    },
    {
      "epoch": 1.4928,
      "grad_norm": 0.7239752567410714,
      "learning_rate": 5.92117023526738e-06,
      "loss": 1.3435,
      "step": 933
    },
    {
      "epoch": 1.4944,
      "grad_norm": 0.6850687335667321,
      "learning_rate": 5.912016846801153e-06,
      "loss": 1.3888,
      "step": 934
    },
    {
      "epoch": 1.496,
      "grad_norm": 0.7450421078778725,
      "learning_rate": 5.902860295530665e-06,
      "loss": 1.3078,
      "step": 935
    },
    {
      "epoch": 1.4976,
      "grad_norm": 0.6705853194909426,
      "learning_rate": 5.893700613210128e-06,
      "loss": 1.3035,
      "step": 936
    },
    {
      "epoch": 1.4992,
      "grad_norm": 0.7032374829865361,
      "learning_rate": 5.88453783160462e-06,
      "loss": 1.3475,
      "step": 937
    },
    {
      "epoch": 1.5008,
      "grad_norm": 0.7418329915856665,
      "learning_rate": 5.875371982489959e-06,
      "loss": 1.3588,
      "step": 938
    },
    {
      "epoch": 1.5024,
      "grad_norm": 0.6904623721813852,
      "learning_rate": 5.866203097652605e-06,
      "loss": 1.3359,
      "step": 939
    },
    {
      "epoch": 1.504,
      "grad_norm": 0.7530294506283617,
      "learning_rate": 5.857031208889548e-06,
      "loss": 1.4288,
      "step": 940
    },
    {
      "epoch": 1.5056,
      "grad_norm": 0.7625763687455192,
      "learning_rate": 5.847856348008188e-06,
      "loss": 1.3899,
      "step": 941
    },
    {
      "epoch": 1.5072,
      "grad_norm": 0.7243668032942546,
      "learning_rate": 5.838678546826242e-06,
      "loss": 1.3467,
      "step": 942
    },
    {
      "epoch": 1.5088,
      "grad_norm": 0.8101520633774664,
      "learning_rate": 5.829497837171616e-06,
      "loss": 1.3914,
      "step": 943
    },
    {
      "epoch": 1.5104,
      "grad_norm": 0.6815861084005849,
      "learning_rate": 5.820314250882304e-06,
      "loss": 1.3613,
      "step": 944
    },
    {
      "epoch": 1.512,
      "grad_norm": 0.7179581291930351,
      "learning_rate": 5.811127819806277e-06,
      "loss": 1.2721,
      "step": 945
    },
    {
      "epoch": 1.5135999999999998,
      "grad_norm": 0.7891960865857428,
      "learning_rate": 5.801938575801372e-06,
      "loss": 1.4153,
      "step": 946
    },
    {
      "epoch": 1.5152,
      "grad_norm": 0.697382116148076,
      "learning_rate": 5.792746550735182e-06,
      "loss": 1.4095,
      "step": 947
    },
    {
      "epoch": 1.5168,
      "grad_norm": 0.7903869238061654,
      "learning_rate": 5.7835517764849395e-06,
      "loss": 1.4307,
      "step": 948
    },
    {
      "epoch": 1.5184,
      "grad_norm": 0.6523539351829918,
      "learning_rate": 5.7743542849374155e-06,
      "loss": 1.402,
      "step": 949
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7268517185897488,
      "learning_rate": 5.765154107988803e-06,
      "loss": 1.4111,
      "step": 950
    },
    {
      "epoch": 1.5215999999999998,
      "grad_norm": 0.6468014092829909,
      "learning_rate": 5.755951277544607e-06,
      "loss": 1.3726,
      "step": 951
    },
    {
      "epoch": 1.5232,
      "grad_norm": 0.7755987765653763,
      "learning_rate": 5.746745825519539e-06,
      "loss": 1.4393,
      "step": 952
    },
    {
      "epoch": 1.5248,
      "grad_norm": 0.68902650137046,
      "learning_rate": 5.737537783837395e-06,
      "loss": 1.4101,
      "step": 953
    },
    {
      "epoch": 1.5264,
      "grad_norm": 0.7219142089658582,
      "learning_rate": 5.728327184430955e-06,
      "loss": 1.3894,
      "step": 954
    },
    {
      "epoch": 1.528,
      "grad_norm": 0.6803361196669425,
      "learning_rate": 5.719114059241871e-06,
      "loss": 1.3733,
      "step": 955
    },
    {
      "epoch": 1.5295999999999998,
      "grad_norm": 0.6723371204280771,
      "learning_rate": 5.709898440220552e-06,
      "loss": 1.3069,
      "step": 956
    },
    {
      "epoch": 1.5312000000000001,
      "grad_norm": 0.7588698088999906,
      "learning_rate": 5.700680359326055e-06,
      "loss": 1.3912,
      "step": 957
    },
    {
      "epoch": 1.5328,
      "grad_norm": 0.6931551753212686,
      "learning_rate": 5.691459848525977e-06,
      "loss": 1.3412,
      "step": 958
    },
    {
      "epoch": 1.5344,
      "grad_norm": 0.8015532902635476,
      "learning_rate": 5.682236939796337e-06,
      "loss": 1.3619,
      "step": 959
    },
    {
      "epoch": 1.536,
      "grad_norm": 0.6979596875402471,
      "learning_rate": 5.673011665121477e-06,
      "loss": 1.3381,
      "step": 960
    },
    {
      "epoch": 1.5375999999999999,
      "grad_norm": 0.7127711137912374,
      "learning_rate": 5.663784056493936e-06,
      "loss": 1.3718,
      "step": 961
    },
    {
      "epoch": 1.5392000000000001,
      "grad_norm": 0.6944860596203737,
      "learning_rate": 5.6545541459143535e-06,
      "loss": 1.3587,
      "step": 962
    },
    {
      "epoch": 1.5408,
      "grad_norm": 0.7458423265686603,
      "learning_rate": 5.6453219653913495e-06,
      "loss": 1.3749,
      "step": 963
    },
    {
      "epoch": 1.5424,
      "grad_norm": 0.8003713449970572,
      "learning_rate": 5.636087546941413e-06,
      "loss": 1.3962,
      "step": 964
    },
    {
      "epoch": 1.544,
      "grad_norm": 0.7088123900155326,
      "learning_rate": 5.6268509225888005e-06,
      "loss": 1.3156,
      "step": 965
    },
    {
      "epoch": 1.5455999999999999,
      "grad_norm": 0.6810442135300473,
      "learning_rate": 5.617612124365411e-06,
      "loss": 1.343,
      "step": 966
    },
    {
      "epoch": 1.5472000000000001,
      "grad_norm": 0.7210407729175294,
      "learning_rate": 5.608371184310688e-06,
      "loss": 1.3086,
      "step": 967
    },
    {
      "epoch": 1.5488,
      "grad_norm": 0.6753802268119455,
      "learning_rate": 5.5991281344714984e-06,
      "loss": 1.3608,
      "step": 968
    },
    {
      "epoch": 1.5504,
      "grad_norm": 0.6377620228759628,
      "learning_rate": 5.5898830069020325e-06,
      "loss": 1.3562,
      "step": 969
    },
    {
      "epoch": 1.552,
      "grad_norm": 0.7170375200019616,
      "learning_rate": 5.580635833663679e-06,
      "loss": 1.3325,
      "step": 970
    },
    {
      "epoch": 1.5535999999999999,
      "grad_norm": 0.6810405688560527,
      "learning_rate": 5.5713866468249235e-06,
      "loss": 1.3362,
      "step": 971
    },
    {
      "epoch": 1.5552000000000001,
      "grad_norm": 0.6872383889296149,
      "learning_rate": 5.562135478461234e-06,
      "loss": 1.3526,
      "step": 972
    },
    {
      "epoch": 1.5568,
      "grad_norm": 0.6772713058412964,
      "learning_rate": 5.55288236065495e-06,
      "loss": 1.3191,
      "step": 973
    },
    {
      "epoch": 1.5584,
      "grad_norm": 0.6889664234358763,
      "learning_rate": 5.5436273254951734e-06,
      "loss": 1.2968,
      "step": 974
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.652926169156554,
      "learning_rate": 5.5343704050776535e-06,
      "loss": 1.3269,
      "step": 975
    },
    {
      "epoch": 1.5615999999999999,
      "grad_norm": 0.6637282512295308,
      "learning_rate": 5.5251116315046785e-06,
      "loss": 1.406,
      "step": 976
    },
    {
      "epoch": 1.5632000000000001,
      "grad_norm": 0.7318088722748626,
      "learning_rate": 5.515851036884964e-06,
      "loss": 1.3645,
      "step": 977
    },
    {
      "epoch": 1.5648,
      "grad_norm": 0.6769948491227279,
      "learning_rate": 5.5065886533335355e-06,
      "loss": 1.3572,
      "step": 978
    },
    {
      "epoch": 1.5664,
      "grad_norm": 0.6833314355467591,
      "learning_rate": 5.497324512971632e-06,
      "loss": 1.3289,
      "step": 979
    },
    {
      "epoch": 1.568,
      "grad_norm": 0.6474021164843294,
      "learning_rate": 5.4880586479265774e-06,
      "loss": 1.2682,
      "step": 980
    },
    {
      "epoch": 1.5695999999999999,
      "grad_norm": 0.670462402269025,
      "learning_rate": 5.478791090331677e-06,
      "loss": 1.3531,
      "step": 981
    },
    {
      "epoch": 1.5712000000000002,
      "grad_norm": 0.7177114919118484,
      "learning_rate": 5.4695218723261115e-06,
      "loss": 1.3831,
      "step": 982
    },
    {
      "epoch": 1.5728,
      "grad_norm": 0.6611738105965316,
      "learning_rate": 5.46025102605481e-06,
      "loss": 1.4015,
      "step": 983
    },
    {
      "epoch": 1.5744,
      "grad_norm": 0.6670319555400192,
      "learning_rate": 5.4509785836683606e-06,
      "loss": 1.3338,
      "step": 984
    },
    {
      "epoch": 1.576,
      "grad_norm": 0.6806504708537645,
      "learning_rate": 5.441704577322877e-06,
      "loss": 1.3404,
      "step": 985
    },
    {
      "epoch": 1.5776,
      "grad_norm": 0.6812422861120327,
      "learning_rate": 5.4324290391798995e-06,
      "loss": 1.3011,
      "step": 986
    },
    {
      "epoch": 1.5792000000000002,
      "grad_norm": 0.6774689104239141,
      "learning_rate": 5.423152001406282e-06,
      "loss": 1.3537,
      "step": 987
    },
    {
      "epoch": 1.5808,
      "grad_norm": 0.6793035690144824,
      "learning_rate": 5.413873496174077e-06,
      "loss": 1.2821,
      "step": 988
    },
    {
      "epoch": 1.5824,
      "grad_norm": 0.6827038653486691,
      "learning_rate": 5.404593555660424e-06,
      "loss": 1.3254,
      "step": 989
    },
    {
      "epoch": 1.584,
      "grad_norm": 0.6764204353741932,
      "learning_rate": 5.39531221204745e-06,
      "loss": 1.3453,
      "step": 990
    },
    {
      "epoch": 1.5856,
      "grad_norm": 0.7010247709094417,
      "learning_rate": 5.3860294975221335e-06,
      "loss": 1.3224,
      "step": 991
    },
    {
      "epoch": 1.5872000000000002,
      "grad_norm": 0.7221787892859222,
      "learning_rate": 5.376745444276219e-06,
      "loss": 1.4139,
      "step": 992
    },
    {
      "epoch": 1.5888,
      "grad_norm": 0.6970890472598371,
      "learning_rate": 5.3674600845060856e-06,
      "loss": 1.3268,
      "step": 993
    },
    {
      "epoch": 1.5904,
      "grad_norm": 0.6680065280301735,
      "learning_rate": 5.358173450412649e-06,
      "loss": 1.3568,
      "step": 994
    },
    {
      "epoch": 1.592,
      "grad_norm": 0.7367801110671264,
      "learning_rate": 5.34888557420124e-06,
      "loss": 1.3987,
      "step": 995
    },
    {
      "epoch": 1.5936,
      "grad_norm": 0.6949437392577807,
      "learning_rate": 5.339596488081501e-06,
      "loss": 1.3757,
      "step": 996
    },
    {
      "epoch": 1.5952,
      "grad_norm": 0.7807430174357621,
      "learning_rate": 5.330306224267268e-06,
      "loss": 1.3971,
      "step": 997
    },
    {
      "epoch": 1.5968,
      "grad_norm": 0.732557880710717,
      "learning_rate": 5.321014814976459e-06,
      "loss": 1.3966,
      "step": 998
    },
    {
      "epoch": 1.5984,
      "grad_norm": 0.6537746559326066,
      "learning_rate": 5.311722292430966e-06,
      "loss": 1.2646,
      "step": 999
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.6986483149902544,
      "learning_rate": 5.302428688856544e-06,
      "loss": 1.2797,
      "step": 1000
    },
    {
      "epoch": 1.6016,
      "grad_norm": 0.6790122111219713,
      "learning_rate": 5.293134036482697e-06,
      "loss": 1.3257,
      "step": 1001
    },
    {
      "epoch": 1.6032,
      "grad_norm": 0.7278889843188312,
      "learning_rate": 5.283838367542562e-06,
      "loss": 1.3307,
      "step": 1002
    },
    {
      "epoch": 1.6048,
      "grad_norm": 0.677682018021205,
      "learning_rate": 5.274541714272805e-06,
      "loss": 1.3525,
      "step": 1003
    },
    {
      "epoch": 1.6064,
      "grad_norm": 0.7171264511036044,
      "learning_rate": 5.265244108913503e-06,
      "loss": 1.3741,
      "step": 1004
    },
    {
      "epoch": 1.608,
      "grad_norm": 0.6924756345090655,
      "learning_rate": 5.255945583708037e-06,
      "loss": 1.4105,
      "step": 1005
    },
    {
      "epoch": 1.6096,
      "grad_norm": 0.6979067282298962,
      "learning_rate": 5.2466461709029755e-06,
      "loss": 1.3757,
      "step": 1006
    },
    {
      "epoch": 1.6112,
      "grad_norm": 0.6645077586262442,
      "learning_rate": 5.237345902747969e-06,
      "loss": 1.2908,
      "step": 1007
    },
    {
      "epoch": 1.6128,
      "grad_norm": 0.7184795660929841,
      "learning_rate": 5.228044811495632e-06,
      "loss": 1.4162,
      "step": 1008
    },
    {
      "epoch": 1.6143999999999998,
      "grad_norm": 0.6510115228383955,
      "learning_rate": 5.218742929401432e-06,
      "loss": 1.321,
      "step": 1009
    },
    {
      "epoch": 1.616,
      "grad_norm": 0.6605753715284373,
      "learning_rate": 5.2094402887235805e-06,
      "loss": 1.3415,
      "step": 1010
    },
    {
      "epoch": 1.6176,
      "grad_norm": 0.6971631341303121,
      "learning_rate": 5.200136921722919e-06,
      "loss": 1.31,
      "step": 1011
    },
    {
      "epoch": 1.6192,
      "grad_norm": 0.6749472868368925,
      "learning_rate": 5.1908328606628114e-06,
      "loss": 1.3592,
      "step": 1012
    },
    {
      "epoch": 1.6208,
      "grad_norm": 0.6455131925709916,
      "learning_rate": 5.181528137809023e-06,
      "loss": 1.3587,
      "step": 1013
    },
    {
      "epoch": 1.6223999999999998,
      "grad_norm": 0.6847908922453064,
      "learning_rate": 5.1722227854296195e-06,
      "loss": 1.3634,
      "step": 1014
    },
    {
      "epoch": 1.624,
      "grad_norm": 0.6281595079312746,
      "learning_rate": 5.162916835794843e-06,
      "loss": 1.3782,
      "step": 1015
    },
    {
      "epoch": 1.6256,
      "grad_norm": 0.662601458426373,
      "learning_rate": 5.1536103211770135e-06,
      "loss": 1.3187,
      "step": 1016
    },
    {
      "epoch": 1.6272,
      "grad_norm": 0.6671719805159587,
      "learning_rate": 5.14430327385041e-06,
      "loss": 1.353,
      "step": 1017
    },
    {
      "epoch": 1.6288,
      "grad_norm": 0.6892289132264204,
      "learning_rate": 5.134995726091152e-06,
      "loss": 1.384,
      "step": 1018
    },
    {
      "epoch": 1.6303999999999998,
      "grad_norm": 0.650902253918775,
      "learning_rate": 5.1256877101771015e-06,
      "loss": 1.3153,
      "step": 1019
    },
    {
      "epoch": 1.6320000000000001,
      "grad_norm": 0.6324348951111782,
      "learning_rate": 5.116379258387742e-06,
      "loss": 1.3242,
      "step": 1020
    },
    {
      "epoch": 1.6336,
      "grad_norm": 0.6999108476842828,
      "learning_rate": 5.1070704030040675e-06,
      "loss": 1.3269,
      "step": 1021
    },
    {
      "epoch": 1.6352,
      "grad_norm": 0.6445698615538108,
      "learning_rate": 5.097761176308471e-06,
      "loss": 1.3248,
      "step": 1022
    },
    {
      "epoch": 1.6368,
      "grad_norm": 0.6460736641002475,
      "learning_rate": 5.088451610584638e-06,
      "loss": 1.3331,
      "step": 1023
    },
    {
      "epoch": 1.6383999999999999,
      "grad_norm": 0.6679124739675205,
      "learning_rate": 5.079141738117423e-06,
      "loss": 1.2788,
      "step": 1024
    },
    {
      "epoch": 1.6400000000000001,
      "grad_norm": 0.7023529722758919,
      "learning_rate": 5.06983159119275e-06,
      "loss": 1.3854,
      "step": 1025
    },
    {
      "epoch": 1.6416,
      "grad_norm": 0.6667255961671676,
      "learning_rate": 5.060521202097491e-06,
      "loss": 1.3055,
      "step": 1026
    },
    {
      "epoch": 1.6432,
      "grad_norm": 0.6643867194098233,
      "learning_rate": 5.051210603119358e-06,
      "loss": 1.3245,
      "step": 1027
    },
    {
      "epoch": 1.6448,
      "grad_norm": 0.6557568166470471,
      "learning_rate": 5.041899826546791e-06,
      "loss": 1.3932,
      "step": 1028
    },
    {
      "epoch": 1.6463999999999999,
      "grad_norm": 0.6440742834107318,
      "learning_rate": 5.032588904668851e-06,
      "loss": 1.3417,
      "step": 1029
    },
    {
      "epoch": 1.6480000000000001,
      "grad_norm": 0.678894932325513,
      "learning_rate": 5.023277869775097e-06,
      "loss": 1.3874,
      "step": 1030
    },
    {
      "epoch": 1.6496,
      "grad_norm": 0.6131236598823117,
      "learning_rate": 5.013966754155482e-06,
      "loss": 1.3382,
      "step": 1031
    },
    {
      "epoch": 1.6512,
      "grad_norm": 0.6833757878995739,
      "learning_rate": 5.004655590100238e-06,
      "loss": 1.4026,
      "step": 1032
    },
    {
      "epoch": 1.6528,
      "grad_norm": 0.6619341546670373,
      "learning_rate": 4.995344409899764e-06,
      "loss": 1.3269,
      "step": 1033
    },
    {
      "epoch": 1.6543999999999999,
      "grad_norm": 0.6249966838101589,
      "learning_rate": 4.986033245844519e-06,
      "loss": 1.3025,
      "step": 1034
    },
    {
      "epoch": 1.6560000000000001,
      "grad_norm": 0.6457083552913302,
      "learning_rate": 4.976722130224904e-06,
      "loss": 1.3723,
      "step": 1035
    },
    {
      "epoch": 1.6576,
      "grad_norm": 0.6228867753739628,
      "learning_rate": 4.967411095331149e-06,
      "loss": 1.3793,
      "step": 1036
    },
    {
      "epoch": 1.6592,
      "grad_norm": 0.730491086175757,
      "learning_rate": 4.95810017345321e-06,
      "loss": 1.3598,
      "step": 1037
    },
    {
      "epoch": 1.6608,
      "grad_norm": 0.663381091901024,
      "learning_rate": 4.948789396880644e-06,
      "loss": 1.325,
      "step": 1038
    },
    {
      "epoch": 1.6623999999999999,
      "grad_norm": 0.7082711966229142,
      "learning_rate": 4.939478797902512e-06,
      "loss": 1.3191,
      "step": 1039
    },
    {
      "epoch": 1.6640000000000001,
      "grad_norm": 0.6995886026835819,
      "learning_rate": 4.930168408807252e-06,
      "loss": 1.4306,
      "step": 1040
    },
    {
      "epoch": 1.6656,
      "grad_norm": 0.6509011378420538,
      "learning_rate": 4.920858261882578e-06,
      "loss": 1.2868,
      "step": 1041
    },
    {
      "epoch": 1.6672,
      "grad_norm": 0.6706236282822539,
      "learning_rate": 4.911548389415363e-06,
      "loss": 1.3369,
      "step": 1042
    },
    {
      "epoch": 1.6688,
      "grad_norm": 0.6638015648107392,
      "learning_rate": 4.9022388236915306e-06,
      "loss": 1.2778,
      "step": 1043
    },
    {
      "epoch": 1.6703999999999999,
      "grad_norm": 0.6584894458731743,
      "learning_rate": 4.892929596995934e-06,
      "loss": 1.4161,
      "step": 1044
    },
    {
      "epoch": 1.6720000000000002,
      "grad_norm": 0.6796533597117255,
      "learning_rate": 4.883620741612259e-06,
      "loss": 1.3114,
      "step": 1045
    },
    {
      "epoch": 1.6736,
      "grad_norm": 0.6474943043698281,
      "learning_rate": 4.8743122898229e-06,
      "loss": 1.3176,
      "step": 1046
    },
    {
      "epoch": 1.6752,
      "grad_norm": 0.6911038604946017,
      "learning_rate": 4.865004273908851e-06,
      "loss": 1.2984,
      "step": 1047
    },
    {
      "epoch": 1.6768,
      "grad_norm": 0.6859839992882846,
      "learning_rate": 4.855696726149593e-06,
      "loss": 1.3602,
      "step": 1048
    },
    {
      "epoch": 1.6784,
      "grad_norm": 0.6264282139114422,
      "learning_rate": 4.846389678822987e-06,
      "loss": 1.3936,
      "step": 1049
    },
    {
      "epoch": 1.6800000000000002,
      "grad_norm": 0.6378745896727165,
      "learning_rate": 4.837083164205159e-06,
      "loss": 1.3427,
      "step": 1050
    },
    {
      "epoch": 1.6816,
      "grad_norm": 3.0377900280136734,
      "learning_rate": 4.827777214570384e-06,
      "loss": 1.3687,
      "step": 1051
    },
    {
      "epoch": 1.6832,
      "grad_norm": 0.7515494047755383,
      "learning_rate": 4.818471862190979e-06,
      "loss": 1.298,
      "step": 1052
    },
    {
      "epoch": 1.6848,
      "grad_norm": 0.6583577246091852,
      "learning_rate": 4.809167139337191e-06,
      "loss": 1.3085,
      "step": 1053
    },
    {
      "epoch": 1.6864,
      "grad_norm": 0.6657752719153428,
      "learning_rate": 4.799863078277082e-06,
      "loss": 1.3455,
      "step": 1054
    },
    {
      "epoch": 1.688,
      "grad_norm": 0.6732715934067109,
      "learning_rate": 4.790559711276422e-06,
      "loss": 1.2935,
      "step": 1055
    },
    {
      "epoch": 1.6896,
      "grad_norm": 0.654966774586376,
      "learning_rate": 4.781257070598571e-06,
      "loss": 1.3807,
      "step": 1056
    },
    {
      "epoch": 1.6912,
      "grad_norm": 0.7003825962587334,
      "learning_rate": 4.771955188504371e-06,
      "loss": 1.3221,
      "step": 1057
    },
    {
      "epoch": 1.6928,
      "grad_norm": 0.665261616004215,
      "learning_rate": 4.762654097252033e-06,
      "loss": 1.3712,
      "step": 1058
    },
    {
      "epoch": 1.6944,
      "grad_norm": 0.6497946792028855,
      "learning_rate": 4.753353829097025e-06,
      "loss": 1.383,
      "step": 1059
    },
    {
      "epoch": 1.696,
      "grad_norm": 0.6587988349756788,
      "learning_rate": 4.7440544162919645e-06,
      "loss": 1.3807,
      "step": 1060
    },
    {
      "epoch": 1.6976,
      "grad_norm": 0.6435165520848354,
      "learning_rate": 4.734755891086498e-06,
      "loss": 1.3591,
      "step": 1061
    },
    {
      "epoch": 1.6992,
      "grad_norm": 0.6324104637160367,
      "learning_rate": 4.725458285727195e-06,
      "loss": 1.3059,
      "step": 1062
    },
    {
      "epoch": 1.7008,
      "grad_norm": 0.659459548731842,
      "learning_rate": 4.716161632457438e-06,
      "loss": 1.357,
      "step": 1063
    },
    {
      "epoch": 1.7024,
      "grad_norm": 0.6237103637200938,
      "learning_rate": 4.7068659635173034e-06,
      "loss": 1.3354,
      "step": 1064
    },
    {
      "epoch": 1.704,
      "grad_norm": 0.6548618380819098,
      "learning_rate": 4.6975713111434556e-06,
      "loss": 1.3139,
      "step": 1065
    },
    {
      "epoch": 1.7056,
      "grad_norm": 0.6740380457753873,
      "learning_rate": 4.688277707569035e-06,
      "loss": 1.4131,
      "step": 1066
    },
    {
      "epoch": 1.7072,
      "grad_norm": 0.6475384994628477,
      "learning_rate": 4.678985185023542e-06,
      "loss": 1.3105,
      "step": 1067
    },
    {
      "epoch": 1.7088,
      "grad_norm": 0.6387616232316308,
      "learning_rate": 4.669693775732733e-06,
      "loss": 1.319,
      "step": 1068
    },
    {
      "epoch": 1.7104,
      "grad_norm": 0.687689321394054,
      "learning_rate": 4.660403511918499e-06,
      "loss": 1.3782,
      "step": 1069
    },
    {
      "epoch": 1.712,
      "grad_norm": 0.6551885820326636,
      "learning_rate": 4.65111442579876e-06,
      "loss": 1.251,
      "step": 1070
    },
    {
      "epoch": 1.7136,
      "grad_norm": 0.6488402839054503,
      "learning_rate": 4.641826549587352e-06,
      "loss": 1.3798,
      "step": 1071
    },
    {
      "epoch": 1.7151999999999998,
      "grad_norm": 0.6662892682111938,
      "learning_rate": 4.632539915493915e-06,
      "loss": 1.3405,
      "step": 1072
    },
    {
      "epoch": 1.7168,
      "grad_norm": 0.6363903955254434,
      "learning_rate": 4.623254555723783e-06,
      "loss": 1.3326,
      "step": 1073
    },
    {
      "epoch": 1.7184,
      "grad_norm": 0.6176087967915497,
      "learning_rate": 4.613970502477867e-06,
      "loss": 1.2926,
      "step": 1074
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.6903722501155246,
      "learning_rate": 4.604687787952552e-06,
      "loss": 1.3272,
      "step": 1075
    },
    {
      "epoch": 1.7216,
      "grad_norm": 0.647844950950911,
      "learning_rate": 4.5954064443395765e-06,
      "loss": 1.3604,
      "step": 1076
    },
    {
      "epoch": 1.7231999999999998,
      "grad_norm": 0.6859150427916367,
      "learning_rate": 4.586126503825925e-06,
      "loss": 1.2907,
      "step": 1077
    },
    {
      "epoch": 1.7248,
      "grad_norm": 0.6018608141797648,
      "learning_rate": 4.57684799859372e-06,
      "loss": 1.27,
      "step": 1078
    },
    {
      "epoch": 1.7264,
      "grad_norm": 0.6281013923905137,
      "learning_rate": 4.567570960820101e-06,
      "loss": 1.3188,
      "step": 1079
    },
    {
      "epoch": 1.728,
      "grad_norm": 0.7079002118504477,
      "learning_rate": 4.558295422677124e-06,
      "loss": 1.4031,
      "step": 1080
    },
    {
      "epoch": 1.7296,
      "grad_norm": 0.6453739869091973,
      "learning_rate": 4.54902141633164e-06,
      "loss": 1.3176,
      "step": 1081
    },
    {
      "epoch": 1.7311999999999999,
      "grad_norm": 0.693561027149696,
      "learning_rate": 4.539748973945191e-06,
      "loss": 1.3237,
      "step": 1082
    },
    {
      "epoch": 1.7328000000000001,
      "grad_norm": 0.7165005233263607,
      "learning_rate": 4.53047812767389e-06,
      "loss": 1.3202,
      "step": 1083
    },
    {
      "epoch": 1.7344,
      "grad_norm": 0.6567788214276333,
      "learning_rate": 4.5212089096683234e-06,
      "loss": 1.3399,
      "step": 1084
    },
    {
      "epoch": 1.736,
      "grad_norm": 0.6908781343516901,
      "learning_rate": 4.511941352073424e-06,
      "loss": 1.3673,
      "step": 1085
    },
    {
      "epoch": 1.7376,
      "grad_norm": 0.6619932268523898,
      "learning_rate": 4.5026754870283695e-06,
      "loss": 1.3718,
      "step": 1086
    },
    {
      "epoch": 1.7391999999999999,
      "grad_norm": 0.703905812399838,
      "learning_rate": 4.493411346666465e-06,
      "loss": 1.3992,
      "step": 1087
    },
    {
      "epoch": 1.7408000000000001,
      "grad_norm": 0.6447342460548049,
      "learning_rate": 4.484148963115038e-06,
      "loss": 1.3056,
      "step": 1088
    },
    {
      "epoch": 1.7424,
      "grad_norm": 0.6400952588353908,
      "learning_rate": 4.474888368495322e-06,
      "loss": 1.3077,
      "step": 1089
    },
    {
      "epoch": 1.744,
      "grad_norm": 0.6487141871026353,
      "learning_rate": 4.465629594922348e-06,
      "loss": 1.3974,
      "step": 1090
    },
    {
      "epoch": 1.7456,
      "grad_norm": 0.6796158345622526,
      "learning_rate": 4.456372674504828e-06,
      "loss": 1.3451,
      "step": 1091
    },
    {
      "epoch": 1.7471999999999999,
      "grad_norm": 0.6580454888416718,
      "learning_rate": 4.447117639345052e-06,
      "loss": 1.3718,
      "step": 1092
    },
    {
      "epoch": 1.7488000000000001,
      "grad_norm": 0.6446039686253398,
      "learning_rate": 4.437864521538768e-06,
      "loss": 1.3401,
      "step": 1093
    },
    {
      "epoch": 1.7504,
      "grad_norm": 0.7049406502347171,
      "learning_rate": 4.428613353175078e-06,
      "loss": 1.4145,
      "step": 1094
    },
    {
      "epoch": 1.752,
      "grad_norm": 0.6362710087814178,
      "learning_rate": 4.4193641663363214e-06,
      "loss": 1.4086,
      "step": 1095
    },
    {
      "epoch": 1.7536,
      "grad_norm": 1.0531558429898333,
      "learning_rate": 4.410116993097968e-06,
      "loss": 1.3641,
      "step": 1096
    },
    {
      "epoch": 1.7551999999999999,
      "grad_norm": 0.6760897421957139,
      "learning_rate": 4.400871865528502e-06,
      "loss": 1.3811,
      "step": 1097
    },
    {
      "epoch": 1.7568000000000001,
      "grad_norm": 0.7082019247865151,
      "learning_rate": 4.391628815689314e-06,
      "loss": 1.3495,
      "step": 1098
    },
    {
      "epoch": 1.7584,
      "grad_norm": 0.6366810650375123,
      "learning_rate": 4.382387875634592e-06,
      "loss": 1.376,
      "step": 1099
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.6876283529594531,
      "learning_rate": 4.373149077411203e-06,
      "loss": 1.3175,
      "step": 1100
    },
    {
      "epoch": 1.7616,
      "grad_norm": 0.6698579571703096,
      "learning_rate": 4.363912453058589e-06,
      "loss": 1.3263,
      "step": 1101
    },
    {
      "epoch": 1.7631999999999999,
      "grad_norm": 0.6484658022153178,
      "learning_rate": 4.354678034608654e-06,
      "loss": 1.3391,
      "step": 1102
    },
    {
      "epoch": 1.7648000000000001,
      "grad_norm": 0.6678778041022491,
      "learning_rate": 4.345445854085649e-06,
      "loss": 1.3294,
      "step": 1103
    },
    {
      "epoch": 1.7664,
      "grad_norm": 0.6471428343912152,
      "learning_rate": 4.336215943506066e-06,
      "loss": 1.3635,
      "step": 1104
    },
    {
      "epoch": 1.768,
      "grad_norm": 0.6505712648457782,
      "learning_rate": 4.326988334878526e-06,
      "loss": 1.3358,
      "step": 1105
    },
    {
      "epoch": 1.7696,
      "grad_norm": 0.6850183230178106,
      "learning_rate": 4.317763060203665e-06,
      "loss": 1.3175,
      "step": 1106
    },
    {
      "epoch": 1.7711999999999999,
      "grad_norm": 0.6327268022416849,
      "learning_rate": 4.308540151474027e-06,
      "loss": 1.3099,
      "step": 1107
    },
    {
      "epoch": 1.7728000000000002,
      "grad_norm": 0.6396154273912475,
      "learning_rate": 4.299319640673948e-06,
      "loss": 1.453,
      "step": 1108
    },
    {
      "epoch": 1.7744,
      "grad_norm": 0.6952985228262845,
      "learning_rate": 4.290101559779451e-06,
      "loss": 1.3622,
      "step": 1109
    },
    {
      "epoch": 1.776,
      "grad_norm": 0.639042515656194,
      "learning_rate": 4.280885940758131e-06,
      "loss": 1.3769,
      "step": 1110
    },
    {
      "epoch": 1.7776,
      "grad_norm": 0.6546819396429043,
      "learning_rate": 4.271672815569047e-06,
      "loss": 1.3695,
      "step": 1111
    },
    {
      "epoch": 1.7792,
      "grad_norm": 0.6446896991320428,
      "learning_rate": 4.262462216162606e-06,
      "loss": 1.3317,
      "step": 1112
    },
    {
      "epoch": 1.7808000000000002,
      "grad_norm": 0.6038992511181661,
      "learning_rate": 4.253254174480462e-06,
      "loss": 1.3654,
      "step": 1113
    },
    {
      "epoch": 1.7824,
      "grad_norm": 0.6558567243936294,
      "learning_rate": 4.244048722455393e-06,
      "loss": 1.3427,
      "step": 1114
    },
    {
      "epoch": 1.784,
      "grad_norm": 0.6788364506739961,
      "learning_rate": 4.234845892011198e-06,
      "loss": 1.3495,
      "step": 1115
    },
    {
      "epoch": 1.7856,
      "grad_norm": 0.674170189437498,
      "learning_rate": 4.225645715062585e-06,
      "loss": 1.3445,
      "step": 1116
    },
    {
      "epoch": 1.7872,
      "grad_norm": 0.6317117961316667,
      "learning_rate": 4.216448223515061e-06,
      "loss": 1.3644,
      "step": 1117
    },
    {
      "epoch": 1.7888,
      "grad_norm": 0.6413084777388195,
      "learning_rate": 4.2072534492648184e-06,
      "loss": 1.3133,
      "step": 1118
    },
    {
      "epoch": 1.7904,
      "grad_norm": 0.6250406873087505,
      "learning_rate": 4.198061424198627e-06,
      "loss": 1.3314,
      "step": 1119
    },
    {
      "epoch": 1.792,
      "grad_norm": 0.6212547832613983,
      "learning_rate": 4.188872180193723e-06,
      "loss": 1.3746,
      "step": 1120
    },
    {
      "epoch": 1.7936,
      "grad_norm": 0.6567468732402872,
      "learning_rate": 4.179685749117698e-06,
      "loss": 1.3818,
      "step": 1121
    },
    {
      "epoch": 1.7952,
      "grad_norm": 0.64524452893601,
      "learning_rate": 4.170502162828385e-06,
      "loss": 1.2688,
      "step": 1122
    },
    {
      "epoch": 1.7968,
      "grad_norm": 0.6336849022240163,
      "learning_rate": 4.161321453173759e-06,
      "loss": 1.38,
      "step": 1123
    },
    {
      "epoch": 1.7984,
      "grad_norm": 0.650034601957054,
      "learning_rate": 4.152143651991812e-06,
      "loss": 1.27,
      "step": 1124
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.6221824493625211,
      "learning_rate": 4.142968791110455e-06,
      "loss": 1.3401,
      "step": 1125
    },
    {
      "epoch": 1.8016,
      "grad_norm": 0.6581145593590287,
      "learning_rate": 4.133796902347397e-06,
      "loss": 1.3683,
      "step": 1126
    },
    {
      "epoch": 1.8032,
      "grad_norm": 0.6752368594876769,
      "learning_rate": 4.124628017510043e-06,
      "loss": 1.3185,
      "step": 1127
    },
    {
      "epoch": 1.8048,
      "grad_norm": 0.6564974283295307,
      "learning_rate": 4.115462168395382e-06,
      "loss": 1.3263,
      "step": 1128
    },
    {
      "epoch": 1.8064,
      "grad_norm": 0.6717605814407177,
      "learning_rate": 4.106299386789873e-06,
      "loss": 1.3583,
      "step": 1129
    },
    {
      "epoch": 1.808,
      "grad_norm": 0.6399431572826736,
      "learning_rate": 4.097139704469337e-06,
      "loss": 1.3317,
      "step": 1130
    },
    {
      "epoch": 1.8096,
      "grad_norm": 0.640164739975192,
      "learning_rate": 4.0879831531988485e-06,
      "loss": 1.3322,
      "step": 1131
    },
    {
      "epoch": 1.8112,
      "grad_norm": 0.6798297978147348,
      "learning_rate": 4.078829764732621e-06,
      "loss": 1.4627,
      "step": 1132
    },
    {
      "epoch": 1.8128,
      "grad_norm": 0.6454165635948433,
      "learning_rate": 4.069679570813903e-06,
      "loss": 1.4036,
      "step": 1133
    },
    {
      "epoch": 1.8144,
      "grad_norm": 0.700343438871534,
      "learning_rate": 4.060532603174865e-06,
      "loss": 1.3027,
      "step": 1134
    },
    {
      "epoch": 1.8159999999999998,
      "grad_norm": 0.6679769228631405,
      "learning_rate": 4.051388893536484e-06,
      "loss": 1.3356,
      "step": 1135
    },
    {
      "epoch": 1.8176,
      "grad_norm": 0.6626688081749184,
      "learning_rate": 4.042248473608442e-06,
      "loss": 1.3509,
      "step": 1136
    },
    {
      "epoch": 1.8192,
      "grad_norm": 0.7063187450562938,
      "learning_rate": 4.033111375089013e-06,
      "loss": 1.2476,
      "step": 1137
    },
    {
      "epoch": 1.8208,
      "grad_norm": 0.6375241099376013,
      "learning_rate": 4.023977629664951e-06,
      "loss": 1.3383,
      "step": 1138
    },
    {
      "epoch": 1.8224,
      "grad_norm": 0.6714442885178773,
      "learning_rate": 4.0148472690113845e-06,
      "loss": 1.3679,
      "step": 1139
    },
    {
      "epoch": 1.8239999999999998,
      "grad_norm": 12.376466203331214,
      "learning_rate": 4.0057203247917e-06,
      "loss": 1.4963,
      "step": 1140
    },
    {
      "epoch": 1.8256000000000001,
      "grad_norm": 0.8081262964888118,
      "learning_rate": 3.996596828657437e-06,
      "loss": 1.3725,
      "step": 1141
    },
    {
      "epoch": 1.8272,
      "grad_norm": 0.6574422053963478,
      "learning_rate": 3.987476812248181e-06,
      "loss": 1.344,
      "step": 1142
    },
    {
      "epoch": 1.8288,
      "grad_norm": 0.6858796927048543,
      "learning_rate": 3.978360307191444e-06,
      "loss": 1.3085,
      "step": 1143
    },
    {
      "epoch": 1.8304,
      "grad_norm": 0.6240244637535759,
      "learning_rate": 3.969247345102567e-06,
      "loss": 1.4214,
      "step": 1144
    },
    {
      "epoch": 1.8319999999999999,
      "grad_norm": 0.6281495046326401,
      "learning_rate": 3.960137957584601e-06,
      "loss": 1.3574,
      "step": 1145
    },
    {
      "epoch": 1.8336000000000001,
      "grad_norm": 0.6803875365696946,
      "learning_rate": 3.9510321762282e-06,
      "loss": 1.3638,
      "step": 1146
    },
    {
      "epoch": 1.8352,
      "grad_norm": 0.6714943630123164,
      "learning_rate": 3.941930032611513e-06,
      "loss": 1.292,
      "step": 1147
    },
    {
      "epoch": 1.8368,
      "grad_norm": 0.6794232750002699,
      "learning_rate": 3.932831558300074e-06,
      "loss": 1.351,
      "step": 1148
    },
    {
      "epoch": 1.8384,
      "grad_norm": 0.6254609016592831,
      "learning_rate": 3.923736784846693e-06,
      "loss": 1.3551,
      "step": 1149
    },
    {
      "epoch": 1.8399999999999999,
      "grad_norm": 0.7026443739894291,
      "learning_rate": 3.914645743791346e-06,
      "loss": 1.3871,
      "step": 1150
    },
    {
      "epoch": 1.8416000000000001,
      "grad_norm": 0.6605019267665345,
      "learning_rate": 3.90555846666106e-06,
      "loss": 1.337,
      "step": 1151
    },
    {
      "epoch": 1.8432,
      "grad_norm": 0.702294543094394,
      "learning_rate": 3.896474984969817e-06,
      "loss": 1.3669,
      "step": 1152
    },
    {
      "epoch": 1.8448,
      "grad_norm": 0.6964151086705607,
      "learning_rate": 3.887395330218429e-06,
      "loss": 1.3115,
      "step": 1153
    },
    {
      "epoch": 1.8464,
      "grad_norm": 0.628627824432404,
      "learning_rate": 3.878319533894443e-06,
      "loss": 1.3316,
      "step": 1154
    },
    {
      "epoch": 1.8479999999999999,
      "grad_norm": 0.7191772764462545,
      "learning_rate": 3.869247627472021e-06,
      "loss": 1.3294,
      "step": 1155
    },
    {
      "epoch": 1.8496000000000001,
      "grad_norm": 0.6245388979760264,
      "learning_rate": 3.860179642411838e-06,
      "loss": 1.3252,
      "step": 1156
    },
    {
      "epoch": 1.8512,
      "grad_norm": 0.6241177260186508,
      "learning_rate": 3.851115610160967e-06,
      "loss": 1.2875,
      "step": 1157
    },
    {
      "epoch": 1.8528,
      "grad_norm": 0.6599255858981982,
      "learning_rate": 3.842055562152775e-06,
      "loss": 1.3059,
      "step": 1158
    },
    {
      "epoch": 1.8544,
      "grad_norm": 0.6559787324097581,
      "learning_rate": 3.8329995298068114e-06,
      "loss": 1.3471,
      "step": 1159
    },
    {
      "epoch": 1.8559999999999999,
      "grad_norm": 0.6565842531011944,
      "learning_rate": 3.8239475445287015e-06,
      "loss": 1.4238,
      "step": 1160
    },
    {
      "epoch": 1.8576000000000001,
      "grad_norm": 1.5424606748912137,
      "learning_rate": 3.814899637710031e-06,
      "loss": 1.3464,
      "step": 1161
    },
    {
      "epoch": 1.8592,
      "grad_norm": 0.7148939867442236,
      "learning_rate": 3.8058558407282465e-06,
      "loss": 1.321,
      "step": 1162
    },
    {
      "epoch": 1.8608,
      "grad_norm": 0.6517788823305086,
      "learning_rate": 3.7968161849465395e-06,
      "loss": 1.3301,
      "step": 1163
    },
    {
      "epoch": 1.8624,
      "grad_norm": 0.6665285623104507,
      "learning_rate": 3.78778070171374e-06,
      "loss": 1.3298,
      "step": 1164
    },
    {
      "epoch": 1.8639999999999999,
      "grad_norm": 0.6701551630952194,
      "learning_rate": 3.7787494223642096e-06,
      "loss": 1.304,
      "step": 1165
    },
    {
      "epoch": 1.8656000000000001,
      "grad_norm": 0.6711484731915709,
      "learning_rate": 3.7697223782177304e-06,
      "loss": 1.4002,
      "step": 1166
    },
    {
      "epoch": 1.8672,
      "grad_norm": 0.6842476966175096,
      "learning_rate": 3.760699600579399e-06,
      "loss": 1.3462,
      "step": 1167
    },
    {
      "epoch": 1.8688,
      "grad_norm": 0.6930613240985655,
      "learning_rate": 3.7516811207395116e-06,
      "loss": 1.2945,
      "step": 1168
    },
    {
      "epoch": 1.8704,
      "grad_norm": 0.6393252936087012,
      "learning_rate": 3.742666969973463e-06,
      "loss": 1.3642,
      "step": 1169
    },
    {
      "epoch": 1.8719999999999999,
      "grad_norm": 0.6247849516876254,
      "learning_rate": 3.733657179541635e-06,
      "loss": 1.3353,
      "step": 1170
    },
    {
      "epoch": 1.8736000000000002,
      "grad_norm": 0.6612980223953947,
      "learning_rate": 3.724651780689286e-06,
      "loss": 1.3202,
      "step": 1171
    },
    {
      "epoch": 1.8752,
      "grad_norm": 0.6645104650063041,
      "learning_rate": 3.715650804646449e-06,
      "loss": 1.2863,
      "step": 1172
    },
    {
      "epoch": 1.8768,
      "grad_norm": 0.615942640386652,
      "learning_rate": 3.7066542826278153e-06,
      "loss": 1.3004,
      "step": 1173
    },
    {
      "epoch": 1.8784,
      "grad_norm": 0.6452848991277452,
      "learning_rate": 3.6976622458326308e-06,
      "loss": 1.3567,
      "step": 1174
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.6713115504361971,
      "learning_rate": 3.6886747254445877e-06,
      "loss": 1.3248,
      "step": 1175
    },
    {
      "epoch": 1.8816000000000002,
      "grad_norm": 0.636687005515614,
      "learning_rate": 3.6796917526317153e-06,
      "loss": 1.3111,
      "step": 1176
    },
    {
      "epoch": 1.8832,
      "grad_norm": 0.6244271704555026,
      "learning_rate": 3.6707133585462713e-06,
      "loss": 1.3705,
      "step": 1177
    },
    {
      "epoch": 1.8848,
      "grad_norm": 0.6829402219041861,
      "learning_rate": 3.6617395743246375e-06,
      "loss": 1.3533,
      "step": 1178
    },
    {
      "epoch": 1.8864,
      "grad_norm": 0.615986479161409,
      "learning_rate": 3.652770431087206e-06,
      "loss": 1.3259,
      "step": 1179
    },
    {
      "epoch": 1.888,
      "grad_norm": 0.6259717274853193,
      "learning_rate": 3.6438059599382765e-06,
      "loss": 1.3346,
      "step": 1180
    },
    {
      "epoch": 1.8896,
      "grad_norm": 0.6275667326172975,
      "learning_rate": 3.634846191965944e-06,
      "loss": 1.2788,
      "step": 1181
    },
    {
      "epoch": 1.8912,
      "grad_norm": 0.6234577435016704,
      "learning_rate": 3.625891158241994e-06,
      "loss": 1.316,
      "step": 1182
    },
    {
      "epoch": 1.8928,
      "grad_norm": 0.6404847291844787,
      "learning_rate": 3.6169408898217973e-06,
      "loss": 1.3211,
      "step": 1183
    },
    {
      "epoch": 1.8944,
      "grad_norm": 0.6240641631919865,
      "learning_rate": 3.6079954177441945e-06,
      "loss": 1.3503,
      "step": 1184
    },
    {
      "epoch": 1.896,
      "grad_norm": 0.6412056102872291,
      "learning_rate": 3.599054773031394e-06,
      "loss": 1.3813,
      "step": 1185
    },
    {
      "epoch": 1.8976,
      "grad_norm": 0.6392510854690497,
      "learning_rate": 3.5901189866888654e-06,
      "loss": 1.2718,
      "step": 1186
    },
    {
      "epoch": 1.8992,
      "grad_norm": 0.6111856239415533,
      "learning_rate": 3.581188089705226e-06,
      "loss": 1.2854,
      "step": 1187
    },
    {
      "epoch": 1.9008,
      "grad_norm": 0.5962076853861075,
      "learning_rate": 3.572262113052142e-06,
      "loss": 1.2941,
      "step": 1188
    },
    {
      "epoch": 1.9024,
      "grad_norm": 0.6203127547879507,
      "learning_rate": 3.563341087684213e-06,
      "loss": 1.3399,
      "step": 1189
    },
    {
      "epoch": 1.904,
      "grad_norm": 0.6258669247352842,
      "learning_rate": 3.554425044538868e-06,
      "loss": 1.3038,
      "step": 1190
    },
    {
      "epoch": 1.9056,
      "grad_norm": 0.6145089109088188,
      "learning_rate": 3.5455140145362587e-06,
      "loss": 1.3538,
      "step": 1191
    },
    {
      "epoch": 1.9072,
      "grad_norm": 0.6178843237185129,
      "learning_rate": 3.5366080285791516e-06,
      "loss": 1.3276,
      "step": 1192
    },
    {
      "epoch": 1.9088,
      "grad_norm": 0.6318108007644163,
      "learning_rate": 3.527707117552822e-06,
      "loss": 1.329,
      "step": 1193
    },
    {
      "epoch": 1.9104,
      "grad_norm": 0.5875824311505187,
      "learning_rate": 3.5188113123249435e-06,
      "loss": 1.3786,
      "step": 1194
    },
    {
      "epoch": 1.912,
      "grad_norm": 0.6099789899904506,
      "learning_rate": 3.5099206437454852e-06,
      "loss": 1.3503,
      "step": 1195
    },
    {
      "epoch": 1.9136,
      "grad_norm": 0.6022964004044411,
      "learning_rate": 3.5010351426466006e-06,
      "loss": 1.3509,
      "step": 1196
    },
    {
      "epoch": 1.9152,
      "grad_norm": 0.599163634514546,
      "learning_rate": 3.4921548398425246e-06,
      "loss": 1.3047,
      "step": 1197
    },
    {
      "epoch": 1.9167999999999998,
      "grad_norm": 0.6501142050480181,
      "learning_rate": 3.4832797661294633e-06,
      "loss": 1.3283,
      "step": 1198
    },
    {
      "epoch": 1.9184,
      "grad_norm": 0.5918138690729684,
      "learning_rate": 3.4744099522854914e-06,
      "loss": 1.3371,
      "step": 1199
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.65772444864059,
      "learning_rate": 3.4655454290704393e-06,
      "loss": 1.3589,
      "step": 1200
    },
    {
      "epoch": 1.9216,
      "grad_norm": 0.6349770549919541,
      "learning_rate": 3.4566862272257923e-06,
      "loss": 1.3624,
      "step": 1201
    },
    {
      "epoch": 1.9232,
      "grad_norm": 0.5908096027339265,
      "learning_rate": 3.44783237747458e-06,
      "loss": 1.3169,
      "step": 1202
    },
    {
      "epoch": 1.9247999999999998,
      "grad_norm": 1.157324468279487,
      "learning_rate": 3.438983910521273e-06,
      "loss": 1.3929,
      "step": 1203
    },
    {
      "epoch": 1.9264000000000001,
      "grad_norm": 0.6800463676628846,
      "learning_rate": 3.430140857051675e-06,
      "loss": 1.3364,
      "step": 1204
    },
    {
      "epoch": 1.928,
      "grad_norm": 0.6726548323927357,
      "learning_rate": 3.421303247732813e-06,
      "loss": 1.3694,
      "step": 1205
    },
    {
      "epoch": 1.9296,
      "grad_norm": 0.6373691825227287,
      "learning_rate": 3.4124711132128374e-06,
      "loss": 1.3249,
      "step": 1206
    },
    {
      "epoch": 1.9312,
      "grad_norm": 0.6363656641240608,
      "learning_rate": 3.4036444841209113e-06,
      "loss": 1.2503,
      "step": 1207
    },
    {
      "epoch": 1.9327999999999999,
      "grad_norm": 0.6220226143582903,
      "learning_rate": 3.3948233910671036e-06,
      "loss": 1.3452,
      "step": 1208
    },
    {
      "epoch": 1.9344000000000001,
      "grad_norm": 0.6522046010337117,
      "learning_rate": 3.3860078646422894e-06,
      "loss": 1.4131,
      "step": 1209
    },
    {
      "epoch": 1.936,
      "grad_norm": 0.6797447462145116,
      "learning_rate": 3.3771979354180343e-06,
      "loss": 1.3658,
      "step": 1210
    },
    {
      "epoch": 1.9376,
      "grad_norm": 0.6169582926576932,
      "learning_rate": 3.3683936339464957e-06,
      "loss": 1.293,
      "step": 1211
    },
    {
      "epoch": 1.9392,
      "grad_norm": 0.6178000834842585,
      "learning_rate": 3.359594990760313e-06,
      "loss": 1.293,
      "step": 1212
    },
    {
      "epoch": 1.9407999999999999,
      "grad_norm": 0.630754030219247,
      "learning_rate": 3.3508020363725043e-06,
      "loss": 1.3523,
      "step": 1213
    },
    {
      "epoch": 1.9424000000000001,
      "grad_norm": 0.642715182119162,
      "learning_rate": 3.34201480127636e-06,
      "loss": 1.3115,
      "step": 1214
    },
    {
      "epoch": 1.944,
      "grad_norm": 0.6224384125674448,
      "learning_rate": 3.333233315945333e-06,
      "loss": 1.3587,
      "step": 1215
    },
    {
      "epoch": 1.9456,
      "grad_norm": 0.6047167052209338,
      "learning_rate": 3.324457610832942e-06,
      "loss": 1.2812,
      "step": 1216
    },
    {
      "epoch": 1.9472,
      "grad_norm": 0.6350695951057514,
      "learning_rate": 3.315687716372655e-06,
      "loss": 1.3616,
      "step": 1217
    },
    {
      "epoch": 1.9487999999999999,
      "grad_norm": 0.6366793780806884,
      "learning_rate": 3.306923662977789e-06,
      "loss": 1.3557,
      "step": 1218
    },
    {
      "epoch": 1.9504000000000001,
      "grad_norm": 0.6196004730078496,
      "learning_rate": 3.2981654810414128e-06,
      "loss": 1.367,
      "step": 1219
    },
    {
      "epoch": 1.952,
      "grad_norm": 0.5863407824993032,
      "learning_rate": 3.2894132009362245e-06,
      "loss": 1.3092,
      "step": 1220
    },
    {
      "epoch": 1.9536,
      "grad_norm": 0.6256210610608002,
      "learning_rate": 3.280666853014457e-06,
      "loss": 1.3693,
      "step": 1221
    },
    {
      "epoch": 1.9552,
      "grad_norm": 0.7050602915810471,
      "learning_rate": 3.271926467607774e-06,
      "loss": 1.3809,
      "step": 1222
    },
    {
      "epoch": 1.9567999999999999,
      "grad_norm": 0.6181286717172776,
      "learning_rate": 3.2631920750271594e-06,
      "loss": 1.3332,
      "step": 1223
    },
    {
      "epoch": 1.9584000000000001,
      "grad_norm": 0.6671594636027636,
      "learning_rate": 3.2544637055628135e-06,
      "loss": 1.3678,
      "step": 1224
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.6362598270765707,
      "learning_rate": 3.2457413894840516e-06,
      "loss": 1.2772,
      "step": 1225
    },
    {
      "epoch": 1.9616,
      "grad_norm": 0.6252421363145518,
      "learning_rate": 3.2370251570391933e-06,
      "loss": 1.3493,
      "step": 1226
    },
    {
      "epoch": 1.9632,
      "grad_norm": 28.004214604070924,
      "learning_rate": 3.2283150384554642e-06,
      "loss": 1.4639,
      "step": 1227
    },
    {
      "epoch": 1.9647999999999999,
      "grad_norm": 0.750573840899537,
      "learning_rate": 3.219611063938883e-06,
      "loss": 1.3506,
      "step": 1228
    },
    {
      "epoch": 1.9664000000000001,
      "grad_norm": 0.6544448954162048,
      "learning_rate": 3.210913263674166e-06,
      "loss": 1.3127,
      "step": 1229
    },
    {
      "epoch": 1.968,
      "grad_norm": 0.5948026083709533,
      "learning_rate": 3.2022216678246145e-06,
      "loss": 1.3323,
      "step": 1230
    },
    {
      "epoch": 1.9696,
      "grad_norm": 0.7909390664895821,
      "learning_rate": 3.193536306532013e-06,
      "loss": 1.3266,
      "step": 1231
    },
    {
      "epoch": 1.9712,
      "grad_norm": 0.7634441376955958,
      "learning_rate": 3.184857209916528e-06,
      "loss": 1.343,
      "step": 1232
    },
    {
      "epoch": 1.9727999999999999,
      "grad_norm": 0.6495855397862993,
      "learning_rate": 3.1761844080765993e-06,
      "loss": 1.2826,
      "step": 1233
    },
    {
      "epoch": 1.9744000000000002,
      "grad_norm": 0.6367609320649571,
      "learning_rate": 3.1675179310888344e-06,
      "loss": 1.3384,
      "step": 1234
    },
    {
      "epoch": 1.976,
      "grad_norm": 0.6483612875707717,
      "learning_rate": 3.15885780900791e-06,
      "loss": 1.3233,
      "step": 1235
    },
    {
      "epoch": 1.9776,
      "grad_norm": 0.6817415659382478,
      "learning_rate": 3.150204071866464e-06,
      "loss": 1.3203,
      "step": 1236
    },
    {
      "epoch": 1.9792,
      "grad_norm": 0.6364056353889925,
      "learning_rate": 3.141556749674988e-06,
      "loss": 1.2817,
      "step": 1237
    },
    {
      "epoch": 1.9808,
      "grad_norm": 0.606158941570792,
      "learning_rate": 3.132915872421734e-06,
      "loss": 1.2832,
      "step": 1238
    },
    {
      "epoch": 1.9824000000000002,
      "grad_norm": 0.6772470569796374,
      "learning_rate": 3.1242814700725977e-06,
      "loss": 1.2929,
      "step": 1239
    },
    {
      "epoch": 1.984,
      "grad_norm": 0.6646368602346121,
      "learning_rate": 3.1156535725710224e-06,
      "loss": 1.2902,
      "step": 1240
    },
    {
      "epoch": 1.9856,
      "grad_norm": 0.6064780961222773,
      "learning_rate": 3.1070322098378925e-06,
      "loss": 1.3248,
      "step": 1241
    },
    {
      "epoch": 1.9872,
      "grad_norm": 0.6781398484460975,
      "learning_rate": 3.0984174117714306e-06,
      "loss": 1.293,
      "step": 1242
    },
    {
      "epoch": 1.9888,
      "grad_norm": 0.6094767483482506,
      "learning_rate": 3.0898092082470943e-06,
      "loss": 1.3395,
      "step": 1243
    },
    {
      "epoch": 1.9904,
      "grad_norm": 0.5976213745202757,
      "learning_rate": 3.081207629117472e-06,
      "loss": 1.3186,
      "step": 1244
    },
    {
      "epoch": 1.992,
      "grad_norm": 0.6208218161847392,
      "learning_rate": 3.0726127042121766e-06,
      "loss": 1.4383,
      "step": 1245
    },
    {
      "epoch": 1.9936,
      "grad_norm": 0.64025987527674,
      "learning_rate": 3.064024463337747e-06,
      "loss": 1.3705,
      "step": 1246
    },
    {
      "epoch": 1.9952,
      "grad_norm": 0.6105530768215505,
      "learning_rate": 3.0554429362775417e-06,
      "loss": 1.3367,
      "step": 1247
    },
    {
      "epoch": 1.9968,
      "grad_norm": 0.6092786952308857,
      "learning_rate": 3.046868152791638e-06,
      "loss": 1.2827,
      "step": 1248
    },
    {
      "epoch": 1.9984,
      "grad_norm": 0.6381295550680235,
      "learning_rate": 3.038300142616723e-06,
      "loss": 1.353,
      "step": 1249
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.6136779995382629,
      "learning_rate": 3.0297389354659984e-06,
      "loss": 1.3062,
      "step": 1250
    },
    {
      "epoch": 2.0016,
      "grad_norm": 0.8379674429348133,
      "learning_rate": 3.021184561029071e-06,
      "loss": 1.3367,
      "step": 1251
    },
    {
      "epoch": 2.0032,
      "grad_norm": 0.6711180241625848,
      "learning_rate": 3.0126370489718537e-06,
      "loss": 1.3338,
      "step": 1252
    },
    {
      "epoch": 2.0048,
      "grad_norm": 0.7005074366748671,
      "learning_rate": 3.0040964289364618e-06,
      "loss": 1.2804,
      "step": 1253
    },
    {
      "epoch": 2.0064,
      "grad_norm": 0.7076470589524284,
      "learning_rate": 2.9955627305411074e-06,
      "loss": 1.2365,
      "step": 1254
    },
    {
      "epoch": 2.008,
      "grad_norm": 0.6800444036768428,
      "learning_rate": 2.9870359833799994e-06,
      "loss": 1.2179,
      "step": 1255
    },
    {
      "epoch": 2.0096,
      "grad_norm": 0.7059921243132561,
      "learning_rate": 2.978516217023243e-06,
      "loss": 1.2513,
      "step": 1256
    },
    {
      "epoch": 2.0112,
      "grad_norm": 0.7437435185608391,
      "learning_rate": 2.97000346101673e-06,
      "loss": 1.3066,
      "step": 1257
    },
    {
      "epoch": 2.0128,
      "grad_norm": 0.7138885356242721,
      "learning_rate": 2.9614977448820444e-06,
      "loss": 1.3268,
      "step": 1258
    },
    {
      "epoch": 2.0144,
      "grad_norm": 0.6766070330656792,
      "learning_rate": 2.952999098116356e-06,
      "loss": 1.3067,
      "step": 1259
    },
    {
      "epoch": 2.016,
      "grad_norm": 0.705112341334405,
      "learning_rate": 2.944507550192318e-06,
      "loss": 1.3149,
      "step": 1260
    },
    {
      "epoch": 2.0176,
      "grad_norm": 0.6698631587168246,
      "learning_rate": 2.9360231305579645e-06,
      "loss": 1.2426,
      "step": 1261
    },
    {
      "epoch": 2.0192,
      "grad_norm": 0.6825472952542789,
      "learning_rate": 2.9275458686366108e-06,
      "loss": 1.3288,
      "step": 1262
    },
    {
      "epoch": 2.0208,
      "grad_norm": 0.7102143254989264,
      "learning_rate": 2.9190757938267477e-06,
      "loss": 1.2946,
      "step": 1263
    },
    {
      "epoch": 2.0224,
      "grad_norm": 0.6576277394188546,
      "learning_rate": 2.9106129355019464e-06,
      "loss": 1.3191,
      "step": 1264
    },
    {
      "epoch": 2.024,
      "grad_norm": 0.6355260860878179,
      "learning_rate": 2.9021573230107436e-06,
      "loss": 1.3021,
      "step": 1265
    },
    {
      "epoch": 2.0256,
      "grad_norm": 0.6855614591048581,
      "learning_rate": 2.8937089856765564e-06,
      "loss": 1.2264,
      "step": 1266
    },
    {
      "epoch": 2.0272,
      "grad_norm": 0.6611714557145804,
      "learning_rate": 2.885267952797569e-06,
      "loss": 1.3598,
      "step": 1267
    },
    {
      "epoch": 2.0288,
      "grad_norm": 0.6193508217493053,
      "learning_rate": 2.876834253646631e-06,
      "loss": 1.3001,
      "step": 1268
    },
    {
      "epoch": 2.0304,
      "grad_norm": 0.6600041351215179,
      "learning_rate": 2.8684079174711665e-06,
      "loss": 1.2407,
      "step": 1269
    },
    {
      "epoch": 2.032,
      "grad_norm": 0.6575393940119219,
      "learning_rate": 2.8599889734930548e-06,
      "loss": 1.2728,
      "step": 1270
    },
    {
      "epoch": 2.0336,
      "grad_norm": 0.6250514038057702,
      "learning_rate": 2.8515774509085535e-06,
      "loss": 1.3232,
      "step": 1271
    },
    {
      "epoch": 2.0352,
      "grad_norm": 0.6307585560380914,
      "learning_rate": 2.8431733788881703e-06,
      "loss": 1.289,
      "step": 1272
    },
    {
      "epoch": 2.0368,
      "grad_norm": 0.6740498048652726,
      "learning_rate": 2.8347767865765828e-06,
      "loss": 1.3078,
      "step": 1273
    },
    {
      "epoch": 2.0384,
      "grad_norm": 0.6101110761578937,
      "learning_rate": 2.826387703092528e-06,
      "loss": 1.2766,
      "step": 1274
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.6171436099173709,
      "learning_rate": 2.8180061575286995e-06,
      "loss": 1.2437,
      "step": 1275
    },
    {
      "epoch": 2.0416,
      "grad_norm": 0.665621262766703,
      "learning_rate": 2.8096321789516557e-06,
      "loss": 1.2758,
      "step": 1276
    },
    {
      "epoch": 2.0432,
      "grad_norm": 0.6660632907424812,
      "learning_rate": 2.8012657964017056e-06,
      "loss": 1.2579,
      "step": 1277
    },
    {
      "epoch": 2.0448,
      "grad_norm": 0.6799068818571734,
      "learning_rate": 2.792907038892823e-06,
      "loss": 1.3629,
      "step": 1278
    },
    {
      "epoch": 2.0464,
      "grad_norm": 0.6173133859064169,
      "learning_rate": 2.784555935412538e-06,
      "loss": 1.3178,
      "step": 1279
    },
    {
      "epoch": 2.048,
      "grad_norm": 0.6334880045376245,
      "learning_rate": 2.77621251492183e-06,
      "loss": 1.2912,
      "step": 1280
    },
    {
      "epoch": 2.0496,
      "grad_norm": 0.6691227271772218,
      "learning_rate": 2.7678768063550454e-06,
      "loss": 1.1976,
      "step": 1281
    },
    {
      "epoch": 2.0512,
      "grad_norm": 0.6448017271995542,
      "learning_rate": 2.759548838619774e-06,
      "loss": 1.2715,
      "step": 1282
    },
    {
      "epoch": 2.0528,
      "grad_norm": 0.6213914296155125,
      "learning_rate": 2.7512286405967726e-06,
      "loss": 1.2634,
      "step": 1283
    },
    {
      "epoch": 2.0544,
      "grad_norm": 0.630466766816398,
      "learning_rate": 2.742916241139843e-06,
      "loss": 1.2986,
      "step": 1284
    },
    {
      "epoch": 2.056,
      "grad_norm": 0.6020263293212413,
      "learning_rate": 2.7346116690757496e-06,
      "loss": 1.3387,
      "step": 1285
    },
    {
      "epoch": 2.0576,
      "grad_norm": 0.6115337109220004,
      "learning_rate": 2.726314953204111e-06,
      "loss": 1.3033,
      "step": 1286
    },
    {
      "epoch": 2.0592,
      "grad_norm": 1.1402616615113788,
      "learning_rate": 2.718026122297297e-06,
      "loss": 1.3192,
      "step": 1287
    },
    {
      "epoch": 2.0608,
      "grad_norm": 0.6531871571229472,
      "learning_rate": 2.7097452051003375e-06,
      "loss": 1.3038,
      "step": 1288
    },
    {
      "epoch": 2.0624,
      "grad_norm": 0.644715016688103,
      "learning_rate": 2.701472230330813e-06,
      "loss": 1.2165,
      "step": 1289
    },
    {
      "epoch": 2.064,
      "grad_norm": 0.5987579687571938,
      "learning_rate": 2.693207226678767e-06,
      "loss": 1.2706,
      "step": 1290
    },
    {
      "epoch": 2.0656,
      "grad_norm": 0.6737085097029939,
      "learning_rate": 2.684950222806596e-06,
      "loss": 1.2776,
      "step": 1291
    },
    {
      "epoch": 2.0672,
      "grad_norm": 0.6371230889505877,
      "learning_rate": 2.676701247348951e-06,
      "loss": 1.2781,
      "step": 1292
    },
    {
      "epoch": 2.0688,
      "grad_norm": 0.6386818082107468,
      "learning_rate": 2.6684603289126492e-06,
      "loss": 1.301,
      "step": 1293
    },
    {
      "epoch": 2.0704,
      "grad_norm": 0.6406236856231667,
      "learning_rate": 2.660227496076557e-06,
      "loss": 1.3211,
      "step": 1294
    },
    {
      "epoch": 2.072,
      "grad_norm": 0.6500816481347562,
      "learning_rate": 2.6520027773915075e-06,
      "loss": 1.2628,
      "step": 1295
    },
    {
      "epoch": 2.0736,
      "grad_norm": 0.6125113261351928,
      "learning_rate": 2.643786201380194e-06,
      "loss": 1.3185,
      "step": 1296
    },
    {
      "epoch": 2.0752,
      "grad_norm": 0.6184232379847553,
      "learning_rate": 2.6355777965370665e-06,
      "loss": 1.2801,
      "step": 1297
    },
    {
      "epoch": 2.0768,
      "grad_norm": 0.6322118169801048,
      "learning_rate": 2.627377591328245e-06,
      "loss": 1.2858,
      "step": 1298
    },
    {
      "epoch": 2.0784,
      "grad_norm": 0.6497409846095994,
      "learning_rate": 2.6191856141914074e-06,
      "loss": 1.3522,
      "step": 1299
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.6046085939883755,
      "learning_rate": 2.6110018935357005e-06,
      "loss": 1.2789,
      "step": 1300
    },
    {
      "epoch": 2.0816,
      "grad_norm": 0.6174611382972434,
      "learning_rate": 2.6028264577416418e-06,
      "loss": 1.2526,
      "step": 1301
    },
    {
      "epoch": 2.0832,
      "grad_norm": 0.6551821952051397,
      "learning_rate": 2.594659335161008e-06,
      "loss": 1.3311,
      "step": 1302
    },
    {
      "epoch": 2.0848,
      "grad_norm": 0.6320146821031815,
      "learning_rate": 2.586500554116757e-06,
      "loss": 1.2573,
      "step": 1303
    },
    {
      "epoch": 2.0864,
      "grad_norm": 0.6130674445784801,
      "learning_rate": 2.578350142902909e-06,
      "loss": 1.3125,
      "step": 1304
    },
    {
      "epoch": 2.088,
      "grad_norm": 0.6021325025362738,
      "learning_rate": 2.570208129784466e-06,
      "loss": 1.2301,
      "step": 1305
    },
    {
      "epoch": 2.0896,
      "grad_norm": 0.6368625759917463,
      "learning_rate": 2.562074542997305e-06,
      "loss": 1.3242,
      "step": 1306
    },
    {
      "epoch": 2.0912,
      "grad_norm": 0.6344969349008258,
      "learning_rate": 2.5539494107480746e-06,
      "loss": 1.3236,
      "step": 1307
    },
    {
      "epoch": 2.0928,
      "grad_norm": 0.6480427171275477,
      "learning_rate": 2.545832761214112e-06,
      "loss": 1.299,
      "step": 1308
    },
    {
      "epoch": 2.0944,
      "grad_norm": 0.6385551596056415,
      "learning_rate": 2.5377246225433306e-06,
      "loss": 1.3199,
      "step": 1309
    },
    {
      "epoch": 2.096,
      "grad_norm": 0.6198101270538039,
      "learning_rate": 2.529625022854133e-06,
      "loss": 1.3351,
      "step": 1310
    },
    {
      "epoch": 2.0976,
      "grad_norm": 0.6192980240156636,
      "learning_rate": 2.5215339902353097e-06,
      "loss": 1.3033,
      "step": 1311
    },
    {
      "epoch": 2.0992,
      "grad_norm": 0.616683229418312,
      "learning_rate": 2.513451552745936e-06,
      "loss": 1.3215,
      "step": 1312
    },
    {
      "epoch": 2.1008,
      "grad_norm": 0.6314102487666337,
      "learning_rate": 2.505377738415286e-06,
      "loss": 1.3539,
      "step": 1313
    },
    {
      "epoch": 2.1024,
      "grad_norm": 0.6472580497366945,
      "learning_rate": 2.4973125752427243e-06,
      "loss": 1.2911,
      "step": 1314
    },
    {
      "epoch": 2.104,
      "grad_norm": 0.6257561182479189,
      "learning_rate": 2.4892560911976167e-06,
      "loss": 1.3331,
      "step": 1315
    },
    {
      "epoch": 2.1056,
      "grad_norm": 0.6528203215418753,
      "learning_rate": 2.481208314219233e-06,
      "loss": 1.2758,
      "step": 1316
    },
    {
      "epoch": 2.1072,
      "grad_norm": 0.637710579663159,
      "learning_rate": 2.4731692722166408e-06,
      "loss": 1.2906,
      "step": 1317
    },
    {
      "epoch": 2.1088,
      "grad_norm": 0.6191279416472434,
      "learning_rate": 2.4651389930686226e-06,
      "loss": 1.2348,
      "step": 1318
    },
    {
      "epoch": 2.1104,
      "grad_norm": 0.6240859723383856,
      "learning_rate": 2.457117504623565e-06,
      "loss": 1.2406,
      "step": 1319
    },
    {
      "epoch": 2.112,
      "grad_norm": 0.6601335204850137,
      "learning_rate": 2.4491048346993756e-06,
      "loss": 1.2998,
      "step": 1320
    },
    {
      "epoch": 2.1136,
      "grad_norm": 0.6525489226866292,
      "learning_rate": 2.4411010110833783e-06,
      "loss": 1.2469,
      "step": 1321
    },
    {
      "epoch": 2.1152,
      "grad_norm": 0.6074446053535525,
      "learning_rate": 2.433106061532214e-06,
      "loss": 1.2365,
      "step": 1322
    },
    {
      "epoch": 2.1168,
      "grad_norm": 0.5857086554816848,
      "learning_rate": 2.4251200137717545e-06,
      "loss": 1.2494,
      "step": 1323
    },
    {
      "epoch": 2.1184,
      "grad_norm": 0.6317834016584685,
      "learning_rate": 2.4171428954969995e-06,
      "loss": 1.2642,
      "step": 1324
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.624426135735182,
      "learning_rate": 2.4091747343719828e-06,
      "loss": 1.2672,
      "step": 1325
    },
    {
      "epoch": 2.1216,
      "grad_norm": 0.6416763164251693,
      "learning_rate": 2.401215558029671e-06,
      "loss": 1.2652,
      "step": 1326
    },
    {
      "epoch": 2.1232,
      "grad_norm": 0.691988465179322,
      "learning_rate": 2.3932653940718784e-06,
      "loss": 1.3469,
      "step": 1327
    },
    {
      "epoch": 2.1248,
      "grad_norm": 0.6459638623743178,
      "learning_rate": 2.3853242700691594e-06,
      "loss": 1.3023,
      "step": 1328
    },
    {
      "epoch": 2.1264,
      "grad_norm": 0.6379561317808978,
      "learning_rate": 2.3773922135607217e-06,
      "loss": 1.2711,
      "step": 1329
    },
    {
      "epoch": 2.128,
      "grad_norm": 0.6349573657153336,
      "learning_rate": 2.3694692520543293e-06,
      "loss": 1.2938,
      "step": 1330
    },
    {
      "epoch": 2.1296,
      "grad_norm": 0.6656669628806555,
      "learning_rate": 2.3615554130262003e-06,
      "loss": 1.3208,
      "step": 1331
    },
    {
      "epoch": 2.1312,
      "grad_norm": 0.6318585687192223,
      "learning_rate": 2.3536507239209223e-06,
      "loss": 1.2158,
      "step": 1332
    },
    {
      "epoch": 2.1328,
      "grad_norm": 0.6170760187514637,
      "learning_rate": 2.3457552121513455e-06,
      "loss": 1.3387,
      "step": 1333
    },
    {
      "epoch": 2.1344,
      "grad_norm": 0.6445426776528553,
      "learning_rate": 2.337868905098499e-06,
      "loss": 1.2899,
      "step": 1334
    },
    {
      "epoch": 2.136,
      "grad_norm": 0.6295002002539738,
      "learning_rate": 2.329991830111492e-06,
      "loss": 1.3071,
      "step": 1335
    },
    {
      "epoch": 2.1376,
      "grad_norm": 0.6392979911932264,
      "learning_rate": 2.32212401450741e-06,
      "loss": 1.2341,
      "step": 1336
    },
    {
      "epoch": 2.1391999999999998,
      "grad_norm": 0.6202267906582957,
      "learning_rate": 2.3142654855712353e-06,
      "loss": 1.2518,
      "step": 1337
    },
    {
      "epoch": 2.1408,
      "grad_norm": 0.635481221164335,
      "learning_rate": 2.3064162705557387e-06,
      "loss": 1.2337,
      "step": 1338
    },
    {
      "epoch": 2.1424,
      "grad_norm": 0.626395675382894,
      "learning_rate": 2.2985763966813963e-06,
      "loss": 1.3312,
      "step": 1339
    },
    {
      "epoch": 2.144,
      "grad_norm": 0.6291614656725875,
      "learning_rate": 2.2907458911362885e-06,
      "loss": 1.2543,
      "step": 1340
    },
    {
      "epoch": 2.1456,
      "grad_norm": 0.6368620981675532,
      "learning_rate": 2.2829247810760023e-06,
      "loss": 1.2403,
      "step": 1341
    },
    {
      "epoch": 2.1471999999999998,
      "grad_norm": 0.6013455476024556,
      "learning_rate": 2.275113093623551e-06,
      "loss": 1.2861,
      "step": 1342
    },
    {
      "epoch": 2.1488,
      "grad_norm": 0.6810513001618865,
      "learning_rate": 2.2673108558692603e-06,
      "loss": 1.2629,
      "step": 1343
    },
    {
      "epoch": 2.1504,
      "grad_norm": 0.6502051061862656,
      "learning_rate": 2.259518094870693e-06,
      "loss": 1.2493,
      "step": 1344
    },
    {
      "epoch": 2.152,
      "grad_norm": 0.6361695641397105,
      "learning_rate": 2.251734837652547e-06,
      "loss": 1.3069,
      "step": 1345
    },
    {
      "epoch": 2.1536,
      "grad_norm": 0.6231071357608637,
      "learning_rate": 2.243961111206555e-06,
      "loss": 1.2693,
      "step": 1346
    },
    {
      "epoch": 2.1552,
      "grad_norm": 0.6927019162782431,
      "learning_rate": 2.236196942491407e-06,
      "loss": 1.286,
      "step": 1347
    },
    {
      "epoch": 2.1568,
      "grad_norm": 0.6444253209491777,
      "learning_rate": 2.228442358432638e-06,
      "loss": 1.2687,
      "step": 1348
    },
    {
      "epoch": 2.1584,
      "grad_norm": 0.6167235406951848,
      "learning_rate": 2.2206973859225518e-06,
      "loss": 1.2754,
      "step": 1349
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.6635461615483259,
      "learning_rate": 2.2129620518201184e-06,
      "loss": 1.3239,
      "step": 1350
    },
    {
      "epoch": 2.1616,
      "grad_norm": 0.6778007023050733,
      "learning_rate": 2.2052363829508776e-06,
      "loss": 1.2417,
      "step": 1351
    },
    {
      "epoch": 2.1632,
      "grad_norm": 0.656404126008289,
      "learning_rate": 2.1975204061068594e-06,
      "loss": 1.3794,
      "step": 1352
    },
    {
      "epoch": 2.1648,
      "grad_norm": 0.6265379818127204,
      "learning_rate": 2.189814148046473e-06,
      "loss": 1.2953,
      "step": 1353
    },
    {
      "epoch": 2.1664,
      "grad_norm": 0.641873806831476,
      "learning_rate": 2.182117635494431e-06,
      "loss": 1.2571,
      "step": 1354
    },
    {
      "epoch": 2.168,
      "grad_norm": 0.6948685329648459,
      "learning_rate": 2.1744308951416483e-06,
      "loss": 1.362,
      "step": 1355
    },
    {
      "epoch": 2.1696,
      "grad_norm": 0.6233317078942836,
      "learning_rate": 2.1667539536451455e-06,
      "loss": 1.2489,
      "step": 1356
    },
    {
      "epoch": 2.1712,
      "grad_norm": 0.6645679298019618,
      "learning_rate": 2.1590868376279693e-06,
      "loss": 1.2574,
      "step": 1357
    },
    {
      "epoch": 2.1728,
      "grad_norm": 0.6720444170239465,
      "learning_rate": 2.151429573679084e-06,
      "loss": 1.2463,
      "step": 1358
    },
    {
      "epoch": 2.1744,
      "grad_norm": 0.6159375524332018,
      "learning_rate": 2.1437821883532956e-06,
      "loss": 1.2637,
      "step": 1359
    },
    {
      "epoch": 2.176,
      "grad_norm": 0.6306067441040555,
      "learning_rate": 2.136144708171145e-06,
      "loss": 1.3237,
      "step": 1360
    },
    {
      "epoch": 2.1776,
      "grad_norm": 0.6015331789516242,
      "learning_rate": 2.128517159618827e-06,
      "loss": 1.3194,
      "step": 1361
    },
    {
      "epoch": 2.1792,
      "grad_norm": 0.6669351385212483,
      "learning_rate": 2.1208995691480947e-06,
      "loss": 1.31,
      "step": 1362
    },
    {
      "epoch": 2.1808,
      "grad_norm": 0.5978682406393184,
      "learning_rate": 2.1132919631761637e-06,
      "loss": 1.2836,
      "step": 1363
    },
    {
      "epoch": 2.1824,
      "grad_norm": 0.6035770656154795,
      "learning_rate": 2.1056943680856286e-06,
      "loss": 1.2651,
      "step": 1364
    },
    {
      "epoch": 2.184,
      "grad_norm": 0.651139108543202,
      "learning_rate": 2.098106810224362e-06,
      "loss": 1.2841,
      "step": 1365
    },
    {
      "epoch": 2.1856,
      "grad_norm": 0.6217560376628499,
      "learning_rate": 2.0905293159054315e-06,
      "loss": 1.2337,
      "step": 1366
    },
    {
      "epoch": 2.1872,
      "grad_norm": 0.6071394704441454,
      "learning_rate": 2.0829619114070068e-06,
      "loss": 1.2969,
      "step": 1367
    },
    {
      "epoch": 2.1888,
      "grad_norm": 0.6072488148216679,
      "learning_rate": 2.075404622972261e-06,
      "loss": 1.3288,
      "step": 1368
    },
    {
      "epoch": 2.1904,
      "grad_norm": 0.6185109780199584,
      "learning_rate": 2.0678574768092926e-06,
      "loss": 1.3542,
      "step": 1369
    },
    {
      "epoch": 2.192,
      "grad_norm": 0.6336534591676438,
      "learning_rate": 2.0603204990910195e-06,
      "loss": 1.3141,
      "step": 1370
    },
    {
      "epoch": 2.1936,
      "grad_norm": 0.5987968411276762,
      "learning_rate": 2.0527937159551044e-06,
      "loss": 1.2595,
      "step": 1371
    },
    {
      "epoch": 2.1952,
      "grad_norm": 0.6255481469797565,
      "learning_rate": 2.0452771535038518e-06,
      "loss": 1.2826,
      "step": 1372
    },
    {
      "epoch": 2.1968,
      "grad_norm": 0.609294306325578,
      "learning_rate": 2.03777083780412e-06,
      "loss": 1.2602,
      "step": 1373
    },
    {
      "epoch": 2.1984,
      "grad_norm": 0.6171153705148522,
      "learning_rate": 2.030274794887237e-06,
      "loss": 1.3571,
      "step": 1374
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.6065661266881975,
      "learning_rate": 2.0227890507488993e-06,
      "loss": 1.3009,
      "step": 1375
    },
    {
      "epoch": 2.2016,
      "grad_norm": 0.6129516125309937,
      "learning_rate": 2.0153136313490945e-06,
      "loss": 1.3342,
      "step": 1376
    },
    {
      "epoch": 2.2032,
      "grad_norm": 0.6267058156443035,
      "learning_rate": 2.0078485626120015e-06,
      "loss": 1.3113,
      "step": 1377
    },
    {
      "epoch": 2.2048,
      "grad_norm": 0.6155085797678609,
      "learning_rate": 2.000393870425904e-06,
      "loss": 1.3055,
      "step": 1378
    },
    {
      "epoch": 2.2064,
      "grad_norm": 0.610539181933134,
      "learning_rate": 1.9929495806431024e-06,
      "loss": 1.2859,
      "step": 1379
    },
    {
      "epoch": 2.208,
      "grad_norm": 0.6059507441369459,
      "learning_rate": 1.985515719079819e-06,
      "loss": 1.2438,
      "step": 1380
    },
    {
      "epoch": 2.2096,
      "grad_norm": 0.6019636870440168,
      "learning_rate": 1.978092311516116e-06,
      "loss": 1.3007,
      "step": 1381
    },
    {
      "epoch": 2.2112,
      "grad_norm": 0.6025102869191316,
      "learning_rate": 1.9706793836957964e-06,
      "loss": 1.2534,
      "step": 1382
    },
    {
      "epoch": 2.2128,
      "grad_norm": 0.6178249168651243,
      "learning_rate": 1.963276961326326e-06,
      "loss": 1.2548,
      "step": 1383
    },
    {
      "epoch": 2.2144,
      "grad_norm": 0.618593736223826,
      "learning_rate": 1.955885070078737e-06,
      "loss": 1.3284,
      "step": 1384
    },
    {
      "epoch": 2.216,
      "grad_norm": 0.6165368076998654,
      "learning_rate": 1.948503735587537e-06,
      "loss": 1.3493,
      "step": 1385
    },
    {
      "epoch": 2.2176,
      "grad_norm": 0.613917733176555,
      "learning_rate": 1.9411329834506286e-06,
      "loss": 1.3544,
      "step": 1386
    },
    {
      "epoch": 2.2192,
      "grad_norm": 0.6087954321368628,
      "learning_rate": 1.9337728392292104e-06,
      "loss": 1.3158,
      "step": 1387
    },
    {
      "epoch": 2.2208,
      "grad_norm": 0.6100353176363834,
      "learning_rate": 1.926423328447698e-06,
      "loss": 1.2577,
      "step": 1388
    },
    {
      "epoch": 2.2224,
      "grad_norm": 0.6141694492780153,
      "learning_rate": 1.919084476593631e-06,
      "loss": 1.344,
      "step": 1389
    },
    {
      "epoch": 2.224,
      "grad_norm": 0.6413882958330819,
      "learning_rate": 1.9117563091175795e-06,
      "loss": 1.3555,
      "step": 1390
    },
    {
      "epoch": 2.2256,
      "grad_norm": 0.5878227169670541,
      "learning_rate": 1.904438851433068e-06,
      "loss": 1.2795,
      "step": 1391
    },
    {
      "epoch": 2.2272,
      "grad_norm": 0.6014640657321636,
      "learning_rate": 1.897132128916474e-06,
      "loss": 1.3381,
      "step": 1392
    },
    {
      "epoch": 2.2288,
      "grad_norm": 0.6007057049094029,
      "learning_rate": 1.8898361669069497e-06,
      "loss": 1.3158,
      "step": 1393
    },
    {
      "epoch": 2.2304,
      "grad_norm": 0.6498825803157443,
      "learning_rate": 1.8825509907063328e-06,
      "loss": 1.2706,
      "step": 1394
    },
    {
      "epoch": 2.232,
      "grad_norm": 0.5943563501347847,
      "learning_rate": 1.87527662557905e-06,
      "loss": 1.3065,
      "step": 1395
    },
    {
      "epoch": 2.2336,
      "grad_norm": 0.6094840298456197,
      "learning_rate": 1.8680130967520433e-06,
      "loss": 1.3164,
      "step": 1396
    },
    {
      "epoch": 2.2352,
      "grad_norm": 0.5977379275779473,
      "learning_rate": 1.8607604294146685e-06,
      "loss": 1.2714,
      "step": 1397
    },
    {
      "epoch": 2.2368,
      "grad_norm": 1.6944107907509582,
      "learning_rate": 1.8535186487186213e-06,
      "loss": 1.3059,
      "step": 1398
    },
    {
      "epoch": 2.2384,
      "grad_norm": 0.6098356897636243,
      "learning_rate": 1.8462877797778367e-06,
      "loss": 1.3304,
      "step": 1399
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.587632515687048,
      "learning_rate": 1.8390678476684143e-06,
      "loss": 1.3018,
      "step": 1400
    },
    {
      "epoch": 2.2416,
      "grad_norm": 0.5885948646709068,
      "learning_rate": 1.831858877428524e-06,
      "loss": 1.262,
      "step": 1401
    },
    {
      "epoch": 2.2432,
      "grad_norm": 0.624914958750849,
      "learning_rate": 1.8246608940583166e-06,
      "loss": 1.2782,
      "step": 1402
    },
    {
      "epoch": 2.2448,
      "grad_norm": 0.6097478796382055,
      "learning_rate": 1.8174739225198485e-06,
      "loss": 1.3172,
      "step": 1403
    },
    {
      "epoch": 2.2464,
      "grad_norm": 0.5926000486369948,
      "learning_rate": 1.8102979877369808e-06,
      "loss": 1.2971,
      "step": 1404
    },
    {
      "epoch": 2.248,
      "grad_norm": 0.6013001585764688,
      "learning_rate": 1.8031331145953047e-06,
      "loss": 1.3508,
      "step": 1405
    },
    {
      "epoch": 2.2496,
      "grad_norm": 0.6234618305724615,
      "learning_rate": 1.7959793279420507e-06,
      "loss": 1.3255,
      "step": 1406
    },
    {
      "epoch": 2.2512,
      "grad_norm": 0.5978395725790436,
      "learning_rate": 1.7888366525859968e-06,
      "loss": 1.2695,
      "step": 1407
    },
    {
      "epoch": 2.2528,
      "grad_norm": 0.6117318815998245,
      "learning_rate": 1.781705113297396e-06,
      "loss": 1.2809,
      "step": 1408
    },
    {
      "epoch": 2.2544,
      "grad_norm": 0.6282629169727392,
      "learning_rate": 1.7745847348078742e-06,
      "loss": 1.2975,
      "step": 1409
    },
    {
      "epoch": 2.2560000000000002,
      "grad_norm": 0.6216494817457986,
      "learning_rate": 1.7674755418103578e-06,
      "loss": 1.2709,
      "step": 1410
    },
    {
      "epoch": 2.2576,
      "grad_norm": 0.5952783364484842,
      "learning_rate": 1.7603775589589821e-06,
      "loss": 1.2574,
      "step": 1411
    },
    {
      "epoch": 2.2592,
      "grad_norm": 0.5802008141439423,
      "learning_rate": 1.7532908108690038e-06,
      "loss": 1.2915,
      "step": 1412
    },
    {
      "epoch": 2.2608,
      "grad_norm": 0.6034139771779741,
      "learning_rate": 1.7462153221167222e-06,
      "loss": 1.3343,
      "step": 1413
    },
    {
      "epoch": 2.2624,
      "grad_norm": 0.6159402285439073,
      "learning_rate": 1.7391511172393849e-06,
      "loss": 1.2858,
      "step": 1414
    },
    {
      "epoch": 2.2640000000000002,
      "grad_norm": 0.596424777079429,
      "learning_rate": 1.7320982207351128e-06,
      "loss": 1.2733,
      "step": 1415
    },
    {
      "epoch": 2.2656,
      "grad_norm": 0.588447181248288,
      "learning_rate": 1.7250566570628103e-06,
      "loss": 1.2946,
      "step": 1416
    },
    {
      "epoch": 2.2672,
      "grad_norm": 0.6080330187694047,
      "learning_rate": 1.7180264506420746e-06,
      "loss": 1.2922,
      "step": 1417
    },
    {
      "epoch": 2.2688,
      "grad_norm": 0.6185067096948921,
      "learning_rate": 1.7110076258531244e-06,
      "loss": 1.297,
      "step": 1418
    },
    {
      "epoch": 2.2704,
      "grad_norm": 0.600065396917257,
      "learning_rate": 1.7040002070367006e-06,
      "loss": 1.2519,
      "step": 1419
    },
    {
      "epoch": 2.2720000000000002,
      "grad_norm": 0.61462881792995,
      "learning_rate": 1.6970042184939943e-06,
      "loss": 1.2453,
      "step": 1420
    },
    {
      "epoch": 2.2736,
      "grad_norm": 0.6261389224189919,
      "learning_rate": 1.6900196844865575e-06,
      "loss": 1.3373,
      "step": 1421
    },
    {
      "epoch": 2.2752,
      "grad_norm": 0.6206668959538808,
      "learning_rate": 1.683046629236213e-06,
      "loss": 1.2231,
      "step": 1422
    },
    {
      "epoch": 2.2768,
      "grad_norm": 0.6810820225649328,
      "learning_rate": 1.6760850769249837e-06,
      "loss": 1.3562,
      "step": 1423
    },
    {
      "epoch": 2.2784,
      "grad_norm": 0.6012592015237929,
      "learning_rate": 1.669135051694994e-06,
      "loss": 1.2711,
      "step": 1424
    },
    {
      "epoch": 2.2800000000000002,
      "grad_norm": 0.6295351005253844,
      "learning_rate": 1.662196577648398e-06,
      "loss": 1.3467,
      "step": 1425
    },
    {
      "epoch": 2.2816,
      "grad_norm": 0.623779225818277,
      "learning_rate": 1.6552696788472921e-06,
      "loss": 1.2932,
      "step": 1426
    },
    {
      "epoch": 2.2832,
      "grad_norm": 0.6010244232532852,
      "learning_rate": 1.6483543793136247e-06,
      "loss": 1.2264,
      "step": 1427
    },
    {
      "epoch": 2.2848,
      "grad_norm": 0.6128214732801966,
      "learning_rate": 1.6414507030291249e-06,
      "loss": 1.3578,
      "step": 1428
    },
    {
      "epoch": 2.2864,
      "grad_norm": 0.5991055133536752,
      "learning_rate": 1.6345586739352105e-06,
      "loss": 1.2555,
      "step": 1429
    },
    {
      "epoch": 2.288,
      "grad_norm": 0.6381062867876003,
      "learning_rate": 1.6276783159329095e-06,
      "loss": 1.2959,
      "step": 1430
    },
    {
      "epoch": 2.2896,
      "grad_norm": 0.598337500503876,
      "learning_rate": 1.6208096528827717e-06,
      "loss": 1.3146,
      "step": 1431
    },
    {
      "epoch": 2.2912,
      "grad_norm": 0.5750275463638208,
      "learning_rate": 1.6139527086047929e-06,
      "loss": 1.228,
      "step": 1432
    },
    {
      "epoch": 2.2928,
      "grad_norm": 0.6098140369609517,
      "learning_rate": 1.6071075068783303e-06,
      "loss": 1.2749,
      "step": 1433
    },
    {
      "epoch": 2.2944,
      "grad_norm": 0.5999996446996162,
      "learning_rate": 1.600274071442014e-06,
      "loss": 1.2661,
      "step": 1434
    },
    {
      "epoch": 2.296,
      "grad_norm": 0.5922967851254388,
      "learning_rate": 1.5934524259936757e-06,
      "loss": 1.2138,
      "step": 1435
    },
    {
      "epoch": 2.2976,
      "grad_norm": 0.6060569687228017,
      "learning_rate": 1.5866425941902524e-06,
      "loss": 1.2977,
      "step": 1436
    },
    {
      "epoch": 2.2992,
      "grad_norm": 0.5928970959852446,
      "learning_rate": 1.5798445996477219e-06,
      "loss": 1.2185,
      "step": 1437
    },
    {
      "epoch": 2.3008,
      "grad_norm": 0.5760666979032258,
      "learning_rate": 1.573058465941002e-06,
      "loss": 1.2482,
      "step": 1438
    },
    {
      "epoch": 2.3024,
      "grad_norm": 0.6148319867532431,
      "learning_rate": 1.5662842166038844e-06,
      "loss": 1.2677,
      "step": 1439
    },
    {
      "epoch": 2.304,
      "grad_norm": 0.6000742514860248,
      "learning_rate": 1.5595218751289465e-06,
      "loss": 1.2713,
      "step": 1440
    },
    {
      "epoch": 2.3056,
      "grad_norm": 0.5960381217770893,
      "learning_rate": 1.5527714649674641e-06,
      "loss": 1.3011,
      "step": 1441
    },
    {
      "epoch": 2.3072,
      "grad_norm": 0.6177117319073827,
      "learning_rate": 1.5460330095293447e-06,
      "loss": 1.2666,
      "step": 1442
    },
    {
      "epoch": 2.3088,
      "grad_norm": 0.6104440849028048,
      "learning_rate": 1.5393065321830292e-06,
      "loss": 1.2594,
      "step": 1443
    },
    {
      "epoch": 2.3104,
      "grad_norm": 0.5989643507883935,
      "learning_rate": 1.5325920562554259e-06,
      "loss": 1.2483,
      "step": 1444
    },
    {
      "epoch": 2.312,
      "grad_norm": 0.5917045355599979,
      "learning_rate": 1.5258896050318217e-06,
      "loss": 1.254,
      "step": 1445
    },
    {
      "epoch": 2.3136,
      "grad_norm": 0.6169151561593608,
      "learning_rate": 1.5191992017557994e-06,
      "loss": 1.2906,
      "step": 1446
    },
    {
      "epoch": 2.3152,
      "grad_norm": 0.651410164105159,
      "learning_rate": 1.512520869629165e-06,
      "loss": 1.2585,
      "step": 1447
    },
    {
      "epoch": 2.3168,
      "grad_norm": 0.5990542294978096,
      "learning_rate": 1.5058546318118583e-06,
      "loss": 1.256,
      "step": 1448
    },
    {
      "epoch": 2.3184,
      "grad_norm": 0.6010839308492502,
      "learning_rate": 1.4992005114218805e-06,
      "loss": 1.2557,
      "step": 1449
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.5969222022117955,
      "learning_rate": 1.4925585315352108e-06,
      "loss": 1.2815,
      "step": 1450
    },
    {
      "epoch": 2.3216,
      "grad_norm": 0.6074988305022176,
      "learning_rate": 1.485928715185721e-06,
      "loss": 1.2559,
      "step": 1451
    },
    {
      "epoch": 2.3232,
      "grad_norm": 0.5998329854761683,
      "learning_rate": 1.4793110853651077e-06,
      "loss": 1.232,
      "step": 1452
    },
    {
      "epoch": 2.3247999999999998,
      "grad_norm": 0.5874620568522732,
      "learning_rate": 1.472705665022799e-06,
      "loss": 1.2923,
      "step": 1453
    },
    {
      "epoch": 2.3264,
      "grad_norm": 0.61427457883151,
      "learning_rate": 1.4661124770658857e-06,
      "loss": 1.3292,
      "step": 1454
    },
    {
      "epoch": 2.328,
      "grad_norm": 0.6277162493921575,
      "learning_rate": 1.459531544359038e-06,
      "loss": 1.3068,
      "step": 1455
    },
    {
      "epoch": 2.3296,
      "grad_norm": 0.649400000638185,
      "learning_rate": 1.4529628897244214e-06,
      "loss": 1.2451,
      "step": 1456
    },
    {
      "epoch": 2.3312,
      "grad_norm": 0.5926292755981164,
      "learning_rate": 1.4464065359416274e-06,
      "loss": 1.2434,
      "step": 1457
    },
    {
      "epoch": 2.3327999999999998,
      "grad_norm": 0.6105018633380502,
      "learning_rate": 1.4398625057475845e-06,
      "loss": 1.3472,
      "step": 1458
    },
    {
      "epoch": 2.3344,
      "grad_norm": 0.5970410303549498,
      "learning_rate": 1.4333308218364861e-06,
      "loss": 1.2593,
      "step": 1459
    },
    {
      "epoch": 2.336,
      "grad_norm": 0.6019596883404339,
      "learning_rate": 1.4268115068597122e-06,
      "loss": 1.2895,
      "step": 1460
    },
    {
      "epoch": 2.3376,
      "grad_norm": 0.6005638558549871,
      "learning_rate": 1.4203045834257418e-06,
      "loss": 1.3243,
      "step": 1461
    },
    {
      "epoch": 2.3392,
      "grad_norm": 0.604611602957513,
      "learning_rate": 1.4138100741000888e-06,
      "loss": 1.3037,
      "step": 1462
    },
    {
      "epoch": 2.3407999999999998,
      "grad_norm": 0.6168135402669284,
      "learning_rate": 1.4073280014052077e-06,
      "loss": 1.3585,
      "step": 1463
    },
    {
      "epoch": 2.3424,
      "grad_norm": 0.5884805634808513,
      "learning_rate": 1.4008583878204297e-06,
      "loss": 1.2717,
      "step": 1464
    },
    {
      "epoch": 2.344,
      "grad_norm": 0.5947397268023633,
      "learning_rate": 1.3944012557818793e-06,
      "loss": 1.2833,
      "step": 1465
    },
    {
      "epoch": 2.3456,
      "grad_norm": 0.5847879957513972,
      "learning_rate": 1.3879566276823896e-06,
      "loss": 1.3102,
      "step": 1466
    },
    {
      "epoch": 2.3472,
      "grad_norm": 0.6032316172899839,
      "learning_rate": 1.3815245258714393e-06,
      "loss": 1.2354,
      "step": 1467
    },
    {
      "epoch": 2.3487999999999998,
      "grad_norm": 0.595228129211271,
      "learning_rate": 1.3751049726550587e-06,
      "loss": 1.2735,
      "step": 1468
    },
    {
      "epoch": 2.3504,
      "grad_norm": 0.5984473898856323,
      "learning_rate": 1.368697990295766e-06,
      "loss": 1.2603,
      "step": 1469
    },
    {
      "epoch": 2.352,
      "grad_norm": 0.6246950216932992,
      "learning_rate": 1.3623036010124845e-06,
      "loss": 1.2942,
      "step": 1470
    },
    {
      "epoch": 2.3536,
      "grad_norm": 0.6018758989095848,
      "learning_rate": 1.3559218269804624e-06,
      "loss": 1.2308,
      "step": 1471
    },
    {
      "epoch": 2.3552,
      "grad_norm": 0.6266837262167,
      "learning_rate": 1.3495526903312029e-06,
      "loss": 1.3203,
      "step": 1472
    },
    {
      "epoch": 2.3568,
      "grad_norm": 0.606954849740846,
      "learning_rate": 1.3431962131523796e-06,
      "loss": 1.3152,
      "step": 1473
    },
    {
      "epoch": 2.3584,
      "grad_norm": 0.5975912364483327,
      "learning_rate": 1.3368524174877679e-06,
      "loss": 1.2753,
      "step": 1474
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.5844542538271746,
      "learning_rate": 1.330521325337164e-06,
      "loss": 1.3295,
      "step": 1475
    },
    {
      "epoch": 2.3616,
      "grad_norm": 0.6252212648215212,
      "learning_rate": 1.3242029586563054e-06,
      "loss": 1.3129,
      "step": 1476
    },
    {
      "epoch": 2.3632,
      "grad_norm": 0.6075127573180501,
      "learning_rate": 1.3178973393568055e-06,
      "loss": 1.3424,
      "step": 1477
    },
    {
      "epoch": 2.3648,
      "grad_norm": 0.6097782476116073,
      "learning_rate": 1.3116044893060637e-06,
      "loss": 1.3276,
      "step": 1478
    },
    {
      "epoch": 2.3664,
      "grad_norm": 0.5987622118550332,
      "learning_rate": 1.3053244303272022e-06,
      "loss": 1.2746,
      "step": 1479
    },
    {
      "epoch": 2.368,
      "grad_norm": 0.5921212410489747,
      "learning_rate": 1.2990571841989796e-06,
      "loss": 1.2365,
      "step": 1480
    },
    {
      "epoch": 2.3696,
      "grad_norm": 0.6179598807204953,
      "learning_rate": 1.2928027726557257e-06,
      "loss": 1.3107,
      "step": 1481
    },
    {
      "epoch": 2.3712,
      "grad_norm": 0.597056054138814,
      "learning_rate": 1.2865612173872577e-06,
      "loss": 1.2933,
      "step": 1482
    },
    {
      "epoch": 2.3728,
      "grad_norm": 0.5868916042021523,
      "learning_rate": 1.2803325400388095e-06,
      "loss": 1.2916,
      "step": 1483
    },
    {
      "epoch": 2.3744,
      "grad_norm": 0.5878939871644239,
      "learning_rate": 1.2741167622109557e-06,
      "loss": 1.2882,
      "step": 1484
    },
    {
      "epoch": 2.376,
      "grad_norm": 0.6171155776423266,
      "learning_rate": 1.2679139054595335e-06,
      "loss": 1.3109,
      "step": 1485
    },
    {
      "epoch": 2.3776,
      "grad_norm": 0.5934775494341088,
      "learning_rate": 1.261723991295576e-06,
      "loss": 1.2736,
      "step": 1486
    },
    {
      "epoch": 2.3792,
      "grad_norm": 0.6305747607969252,
      "learning_rate": 1.2555470411852262e-06,
      "loss": 1.2917,
      "step": 1487
    },
    {
      "epoch": 2.3808,
      "grad_norm": 0.5988862123701658,
      "learning_rate": 1.2493830765496724e-06,
      "loss": 1.3071,
      "step": 1488
    },
    {
      "epoch": 2.3824,
      "grad_norm": 0.6017582012638375,
      "learning_rate": 1.2432321187650726e-06,
      "loss": 1.2583,
      "step": 1489
    },
    {
      "epoch": 2.384,
      "grad_norm": 0.5942841766709781,
      "learning_rate": 1.237094189162471e-06,
      "loss": 1.264,
      "step": 1490
    },
    {
      "epoch": 2.3856,
      "grad_norm": 0.6089322733899283,
      "learning_rate": 1.2309693090277392e-06,
      "loss": 1.2472,
      "step": 1491
    },
    {
      "epoch": 2.3872,
      "grad_norm": 0.5804113976809748,
      "learning_rate": 1.2248574996014872e-06,
      "loss": 1.3267,
      "step": 1492
    },
    {
      "epoch": 2.3888,
      "grad_norm": 0.6000542117449651,
      "learning_rate": 1.218758782079001e-06,
      "loss": 1.305,
      "step": 1493
    },
    {
      "epoch": 2.3904,
      "grad_norm": 0.5833991216200014,
      "learning_rate": 1.2126731776101657e-06,
      "loss": 1.272,
      "step": 1494
    },
    {
      "epoch": 2.392,
      "grad_norm": 0.6474424078998765,
      "learning_rate": 1.2066007072993856e-06,
      "loss": 1.3168,
      "step": 1495
    },
    {
      "epoch": 2.3936,
      "grad_norm": 0.5842362903803608,
      "learning_rate": 1.2005413922055247e-06,
      "loss": 1.2588,
      "step": 1496
    },
    {
      "epoch": 2.3952,
      "grad_norm": 0.5955193687457158,
      "learning_rate": 1.194495253341818e-06,
      "loss": 1.2662,
      "step": 1497
    },
    {
      "epoch": 2.3968,
      "grad_norm": 0.5719623363764389,
      "learning_rate": 1.1884623116758121e-06,
      "loss": 1.3364,
      "step": 1498
    },
    {
      "epoch": 2.3984,
      "grad_norm": 0.5937078219601443,
      "learning_rate": 1.1824425881292846e-06,
      "loss": 1.2902,
      "step": 1499
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.6104497091533339,
      "learning_rate": 1.1764361035781718e-06,
      "loss": 1.2631,
      "step": 1500
    },
    {
      "epoch": 2.4016,
      "grad_norm": 0.5927549123604863,
      "learning_rate": 1.170442878852503e-06,
      "loss": 1.2799,
      "step": 1501
    },
    {
      "epoch": 2.4032,
      "grad_norm": 0.5877383312101597,
      "learning_rate": 1.1644629347363173e-06,
      "loss": 1.2512,
      "step": 1502
    },
    {
      "epoch": 2.4048,
      "grad_norm": 0.5994787704655653,
      "learning_rate": 1.1584962919676024e-06,
      "loss": 1.2728,
      "step": 1503
    },
    {
      "epoch": 2.4064,
      "grad_norm": 0.5836302373000427,
      "learning_rate": 1.1525429712382175e-06,
      "loss": 1.1966,
      "step": 1504
    },
    {
      "epoch": 2.408,
      "grad_norm": 0.5927597943109549,
      "learning_rate": 1.1466029931938182e-06,
      "loss": 1.3407,
      "step": 1505
    },
    {
      "epoch": 2.4096,
      "grad_norm": 0.5948340301741616,
      "learning_rate": 1.1406763784337948e-06,
      "loss": 1.258,
      "step": 1506
    },
    {
      "epoch": 2.4112,
      "grad_norm": 0.6024523330551553,
      "learning_rate": 1.1347631475111882e-06,
      "loss": 1.2616,
      "step": 1507
    },
    {
      "epoch": 2.4128,
      "grad_norm": 0.5907612214898045,
      "learning_rate": 1.1288633209326288e-06,
      "loss": 1.2903,
      "step": 1508
    },
    {
      "epoch": 2.4144,
      "grad_norm": 0.6234937242142051,
      "learning_rate": 1.122976919158264e-06,
      "loss": 1.2774,
      "step": 1509
    },
    {
      "epoch": 2.416,
      "grad_norm": 0.5917633940557848,
      "learning_rate": 1.1171039626016789e-06,
      "loss": 1.2792,
      "step": 1510
    },
    {
      "epoch": 2.4176,
      "grad_norm": 0.6158536561054258,
      "learning_rate": 1.1112444716298381e-06,
      "loss": 1.3258,
      "step": 1511
    },
    {
      "epoch": 2.4192,
      "grad_norm": 0.5983440524414874,
      "learning_rate": 1.1053984665630025e-06,
      "loss": 1.3139,
      "step": 1512
    },
    {
      "epoch": 2.4208,
      "grad_norm": 0.580113347148789,
      "learning_rate": 1.0995659676746706e-06,
      "loss": 1.2407,
      "step": 1513
    },
    {
      "epoch": 2.4224,
      "grad_norm": 0.6110015089646917,
      "learning_rate": 1.093746995191497e-06,
      "loss": 1.2601,
      "step": 1514
    },
    {
      "epoch": 2.424,
      "grad_norm": 0.568182648556211,
      "learning_rate": 1.0879415692932328e-06,
      "loss": 1.2892,
      "step": 1515
    },
    {
      "epoch": 2.4256,
      "grad_norm": 0.6008564543046635,
      "learning_rate": 1.0821497101126487e-06,
      "loss": 1.3458,
      "step": 1516
    },
    {
      "epoch": 2.4272,
      "grad_norm": 0.5750114224542097,
      "learning_rate": 1.076371437735465e-06,
      "loss": 1.2317,
      "step": 1517
    },
    {
      "epoch": 2.4288,
      "grad_norm": 0.6077691616186256,
      "learning_rate": 1.0706067722002877e-06,
      "loss": 1.2418,
      "step": 1518
    },
    {
      "epoch": 2.4304,
      "grad_norm": 0.6063813029076257,
      "learning_rate": 1.064855733498531e-06,
      "loss": 1.3144,
      "step": 1519
    },
    {
      "epoch": 2.432,
      "grad_norm": 0.5810517217997063,
      "learning_rate": 1.0591183415743562e-06,
      "loss": 1.3005,
      "step": 1520
    },
    {
      "epoch": 2.4336,
      "grad_norm": 0.5976502558541316,
      "learning_rate": 1.0533946163245984e-06,
      "loss": 1.3032,
      "step": 1521
    },
    {
      "epoch": 2.4352,
      "grad_norm": 0.5715443037145796,
      "learning_rate": 1.047684577598694e-06,
      "loss": 1.311,
      "step": 1522
    },
    {
      "epoch": 2.4368,
      "grad_norm": 0.5973643067465598,
      "learning_rate": 1.0419882451986197e-06,
      "loss": 1.2674,
      "step": 1523
    },
    {
      "epoch": 2.4384,
      "grad_norm": 0.5756916569368717,
      "learning_rate": 1.0363056388788162e-06,
      "loss": 1.2419,
      "step": 1524
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.5707720770664902,
      "learning_rate": 1.0306367783461258e-06,
      "loss": 1.1951,
      "step": 1525
    },
    {
      "epoch": 2.4416,
      "grad_norm": 0.5841110387661754,
      "learning_rate": 1.024981683259723e-06,
      "loss": 1.2509,
      "step": 1526
    },
    {
      "epoch": 2.4432,
      "grad_norm": 0.5803273949811752,
      "learning_rate": 1.0193403732310392e-06,
      "loss": 1.2335,
      "step": 1527
    },
    {
      "epoch": 2.4448,
      "grad_norm": 0.5711394175838481,
      "learning_rate": 1.0137128678237062e-06,
      "loss": 1.3072,
      "step": 1528
    },
    {
      "epoch": 2.4464,
      "grad_norm": 0.5798644737251993,
      "learning_rate": 1.0080991865534773e-06,
      "loss": 1.2988,
      "step": 1529
    },
    {
      "epoch": 2.448,
      "grad_norm": 0.5915242787345593,
      "learning_rate": 1.002499348888169e-06,
      "loss": 1.2541,
      "step": 1530
    },
    {
      "epoch": 2.4496,
      "grad_norm": 0.574614069131946,
      "learning_rate": 9.969133742475883e-07,
      "loss": 1.313,
      "step": 1531
    },
    {
      "epoch": 2.4512,
      "grad_norm": 0.57508463585911,
      "learning_rate": 9.913412820034629e-07,
      "loss": 1.2756,
      "step": 1532
    },
    {
      "epoch": 2.4528,
      "grad_norm": 0.5938993412681152,
      "learning_rate": 9.857830914793827e-07,
      "loss": 1.1967,
      "step": 1533
    },
    {
      "epoch": 2.4544,
      "grad_norm": 0.5954018053166497,
      "learning_rate": 9.802388219507215e-07,
      "loss": 1.2795,
      "step": 1534
    },
    {
      "epoch": 2.456,
      "grad_norm": 0.5778113110618582,
      "learning_rate": 9.747084926445839e-07,
      "loss": 1.2435,
      "step": 1535
    },
    {
      "epoch": 2.4576000000000002,
      "grad_norm": 0.6084417149213327,
      "learning_rate": 9.691921227397227e-07,
      "loss": 1.264,
      "step": 1536
    },
    {
      "epoch": 2.4592,
      "grad_norm": 0.595341503549805,
      "learning_rate": 9.63689731366486e-07,
      "loss": 1.2614,
      "step": 1537
    },
    {
      "epoch": 2.4608,
      "grad_norm": 0.5860087651083125,
      "learning_rate": 9.58201337606745e-07,
      "loss": 1.3271,
      "step": 1538
    },
    {
      "epoch": 2.4624,
      "grad_norm": 0.5790929970103487,
      "learning_rate": 9.527269604938249e-07,
      "loss": 1.2732,
      "step": 1539
    },
    {
      "epoch": 2.464,
      "grad_norm": 0.6181942650799149,
      "learning_rate": 9.472666190124457e-07,
      "loss": 1.3116,
      "step": 1540
    },
    {
      "epoch": 2.4656000000000002,
      "grad_norm": 0.5788041634305043,
      "learning_rate": 9.418203320986502e-07,
      "loss": 1.2558,
      "step": 1541
    },
    {
      "epoch": 2.4672,
      "grad_norm": 1.8650898996190104,
      "learning_rate": 9.363881186397434e-07,
      "loss": 1.2652,
      "step": 1542
    },
    {
      "epoch": 2.4688,
      "grad_norm": 0.5752417288444968,
      "learning_rate": 9.309699974742243e-07,
      "loss": 1.3096,
      "step": 1543
    },
    {
      "epoch": 2.4704,
      "grad_norm": 0.5844090402661063,
      "learning_rate": 9.255659873917183e-07,
      "loss": 1.2269,
      "step": 1544
    },
    {
      "epoch": 2.472,
      "grad_norm": 0.5815436610125875,
      "learning_rate": 9.201761071329196e-07,
      "loss": 1.3464,
      "step": 1545
    },
    {
      "epoch": 2.4736000000000002,
      "grad_norm": 0.5898799657585355,
      "learning_rate": 9.148003753895146e-07,
      "loss": 1.3116,
      "step": 1546
    },
    {
      "epoch": 2.4752,
      "grad_norm": 0.5818578414803242,
      "learning_rate": 9.094388108041302e-07,
      "loss": 1.3309,
      "step": 1547
    },
    {
      "epoch": 2.4768,
      "grad_norm": 0.5788702074654718,
      "learning_rate": 9.040914319702598e-07,
      "loss": 1.3495,
      "step": 1548
    },
    {
      "epoch": 2.4784,
      "grad_norm": 0.5651084700680766,
      "learning_rate": 8.987582574321996e-07,
      "loss": 1.2848,
      "step": 1549
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.5863501777468383,
      "learning_rate": 8.934393056849921e-07,
      "loss": 1.325,
      "step": 1550
    },
    {
      "epoch": 2.4816,
      "grad_norm": 0.5842690382077194,
      "learning_rate": 8.881345951743486e-07,
      "loss": 1.3161,
      "step": 1551
    },
    {
      "epoch": 2.4832,
      "grad_norm": 0.5869156916152269,
      "learning_rate": 8.828441442966013e-07,
      "loss": 1.229,
      "step": 1552
    },
    {
      "epoch": 2.4848,
      "grad_norm": 0.5848945792805206,
      "learning_rate": 8.775679713986235e-07,
      "loss": 1.3072,
      "step": 1553
    },
    {
      "epoch": 2.4864,
      "grad_norm": 0.5870647156286243,
      "learning_rate": 8.723060947777778e-07,
      "loss": 1.2872,
      "step": 1554
    },
    {
      "epoch": 2.488,
      "grad_norm": 0.5898436024931942,
      "learning_rate": 8.670585326818493e-07,
      "loss": 1.2231,
      "step": 1555
    },
    {
      "epoch": 2.4896,
      "grad_norm": 0.5958166284265798,
      "learning_rate": 8.618253033089768e-07,
      "loss": 1.2627,
      "step": 1556
    },
    {
      "epoch": 2.4912,
      "grad_norm": 0.5701789436742846,
      "learning_rate": 8.566064248076001e-07,
      "loss": 1.2746,
      "step": 1557
    },
    {
      "epoch": 2.4928,
      "grad_norm": 0.6082537296502547,
      "learning_rate": 8.514019152763852e-07,
      "loss": 1.3061,
      "step": 1558
    },
    {
      "epoch": 2.4944,
      "grad_norm": 0.5777305842749555,
      "learning_rate": 8.462117927641733e-07,
      "loss": 1.3392,
      "step": 1559
    },
    {
      "epoch": 2.496,
      "grad_norm": 0.5750230965707499,
      "learning_rate": 8.410360752699099e-07,
      "loss": 1.2231,
      "step": 1560
    },
    {
      "epoch": 2.4976,
      "grad_norm": 0.5755198958708807,
      "learning_rate": 8.358747807425827e-07,
      "loss": 1.3299,
      "step": 1561
    },
    {
      "epoch": 2.4992,
      "grad_norm": 0.5807021454271123,
      "learning_rate": 8.307279270811675e-07,
      "loss": 1.2852,
      "step": 1562
    },
    {
      "epoch": 2.5008,
      "grad_norm": 0.5793570568595283,
      "learning_rate": 8.255955321345533e-07,
      "loss": 1.2907,
      "step": 1563
    },
    {
      "epoch": 2.5023999999999997,
      "grad_norm": 0.5640755646079151,
      "learning_rate": 8.20477613701493e-07,
      "loss": 1.3274,
      "step": 1564
    },
    {
      "epoch": 2.504,
      "grad_norm": 0.5664363229588703,
      "learning_rate": 8.153741895305351e-07,
      "loss": 1.2728,
      "step": 1565
    },
    {
      "epoch": 2.5056000000000003,
      "grad_norm": 0.581000705760727,
      "learning_rate": 8.102852773199588e-07,
      "loss": 1.2647,
      "step": 1566
    },
    {
      "epoch": 2.5072,
      "grad_norm": 0.6038111682716425,
      "learning_rate": 8.052108947177234e-07,
      "loss": 1.2821,
      "step": 1567
    },
    {
      "epoch": 2.5088,
      "grad_norm": 0.5883333244076648,
      "learning_rate": 8.001510593213946e-07,
      "loss": 1.2581,
      "step": 1568
    },
    {
      "epoch": 2.5103999999999997,
      "grad_norm": 0.6028475693119455,
      "learning_rate": 7.951057886780939e-07,
      "loss": 1.3452,
      "step": 1569
    },
    {
      "epoch": 2.512,
      "grad_norm": 1.0712923552330136,
      "learning_rate": 7.900751002844326e-07,
      "loss": 1.3255,
      "step": 1570
    },
    {
      "epoch": 2.5136,
      "grad_norm": 0.599428198701695,
      "learning_rate": 7.850590115864481e-07,
      "loss": 1.295,
      "step": 1571
    },
    {
      "epoch": 2.5152,
      "grad_norm": 0.6158263652928382,
      "learning_rate": 7.80057539979554e-07,
      "loss": 1.2231,
      "step": 1572
    },
    {
      "epoch": 2.5168,
      "grad_norm": 0.5948981961253839,
      "learning_rate": 7.750707028084653e-07,
      "loss": 1.3237,
      "step": 1573
    },
    {
      "epoch": 2.5183999999999997,
      "grad_norm": 0.5678862997036135,
      "learning_rate": 7.70098517367151e-07,
      "loss": 1.3147,
      "step": 1574
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.582721588580987,
      "learning_rate": 7.651410008987698e-07,
      "loss": 1.3131,
      "step": 1575
    },
    {
      "epoch": 2.5216,
      "grad_norm": 0.5907524970655942,
      "learning_rate": 7.601981705956041e-07,
      "loss": 1.2553,
      "step": 1576
    },
    {
      "epoch": 2.5232,
      "grad_norm": 0.5615361734800919,
      "learning_rate": 7.552700435990123e-07,
      "loss": 1.2462,
      "step": 1577
    },
    {
      "epoch": 2.5248,
      "grad_norm": 0.570770097533728,
      "learning_rate": 7.503566369993564e-07,
      "loss": 1.2218,
      "step": 1578
    },
    {
      "epoch": 2.5263999999999998,
      "grad_norm": 0.6011152585516141,
      "learning_rate": 7.454579678359547e-07,
      "loss": 1.3078,
      "step": 1579
    },
    {
      "epoch": 2.528,
      "grad_norm": 0.5836743367936337,
      "learning_rate": 7.405740530970157e-07,
      "loss": 1.3052,
      "step": 1580
    },
    {
      "epoch": 2.5296,
      "grad_norm": 0.5680654763051711,
      "learning_rate": 7.357049097195773e-07,
      "loss": 1.2293,
      "step": 1581
    },
    {
      "epoch": 2.5312,
      "grad_norm": 0.580738417104457,
      "learning_rate": 7.308505545894567e-07,
      "loss": 1.3286,
      "step": 1582
    },
    {
      "epoch": 2.5328,
      "grad_norm": 0.5935052366950028,
      "learning_rate": 7.260110045411816e-07,
      "loss": 1.3238,
      "step": 1583
    },
    {
      "epoch": 2.5343999999999998,
      "grad_norm": 0.5865128414250734,
      "learning_rate": 7.211862763579414e-07,
      "loss": 1.2511,
      "step": 1584
    },
    {
      "epoch": 2.536,
      "grad_norm": 0.565047470081287,
      "learning_rate": 7.163763867715218e-07,
      "loss": 1.2533,
      "step": 1585
    },
    {
      "epoch": 2.5376,
      "grad_norm": 0.5646551145164256,
      "learning_rate": 7.115813524622489e-07,
      "loss": 1.2375,
      "step": 1586
    },
    {
      "epoch": 2.5392,
      "grad_norm": 0.5746993034889007,
      "learning_rate": 7.068011900589333e-07,
      "loss": 1.2768,
      "step": 1587
    },
    {
      "epoch": 2.5408,
      "grad_norm": 0.5904272023865584,
      "learning_rate": 7.020359161388108e-07,
      "loss": 1.2689,
      "step": 1588
    },
    {
      "epoch": 2.5423999999999998,
      "grad_norm": 0.5662592689928295,
      "learning_rate": 6.972855472274853e-07,
      "loss": 1.3079,
      "step": 1589
    },
    {
      "epoch": 2.544,
      "grad_norm": 0.58050459523481,
      "learning_rate": 6.925500997988694e-07,
      "loss": 1.2906,
      "step": 1590
    },
    {
      "epoch": 2.5456,
      "grad_norm": 0.5910518143311957,
      "learning_rate": 6.87829590275132e-07,
      "loss": 1.3063,
      "step": 1591
    },
    {
      "epoch": 2.5472,
      "grad_norm": 0.5827154373821485,
      "learning_rate": 6.83124035026635e-07,
      "loss": 1.2796,
      "step": 1592
    },
    {
      "epoch": 2.5488,
      "grad_norm": 0.5900259299797407,
      "learning_rate": 6.784334503718826e-07,
      "loss": 1.2902,
      "step": 1593
    },
    {
      "epoch": 2.5504,
      "grad_norm": 0.5679137988727508,
      "learning_rate": 6.737578525774636e-07,
      "loss": 1.1976,
      "step": 1594
    },
    {
      "epoch": 2.552,
      "grad_norm": 0.60830223490288,
      "learning_rate": 6.690972578579886e-07,
      "loss": 1.2876,
      "step": 1595
    },
    {
      "epoch": 2.5536,
      "grad_norm": 0.5910505434344956,
      "learning_rate": 6.644516823760439e-07,
      "loss": 1.3165,
      "step": 1596
    },
    {
      "epoch": 2.5552,
      "grad_norm": 0.6052014218526193,
      "learning_rate": 6.598211422421258e-07,
      "loss": 1.3106,
      "step": 1597
    },
    {
      "epoch": 2.5568,
      "grad_norm": 0.5958926799461013,
      "learning_rate": 6.552056535145917e-07,
      "loss": 1.2886,
      "step": 1598
    },
    {
      "epoch": 2.5584,
      "grad_norm": 0.6062113348727988,
      "learning_rate": 6.506052321996037e-07,
      "loss": 1.3,
      "step": 1599
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.594078287704702,
      "learning_rate": 6.46019894251066e-07,
      "loss": 1.2419,
      "step": 1600
    },
    {
      "epoch": 2.5616,
      "grad_norm": 0.5794452254084113,
      "learning_rate": 6.414496555705802e-07,
      "loss": 1.237,
      "step": 1601
    },
    {
      "epoch": 2.5632,
      "grad_norm": 0.6052682608856761,
      "learning_rate": 6.368945320073799e-07,
      "loss": 1.3283,
      "step": 1602
    },
    {
      "epoch": 2.5648,
      "grad_norm": 0.5694567243226236,
      "learning_rate": 6.323545393582847e-07,
      "loss": 1.2565,
      "step": 1603
    },
    {
      "epoch": 2.5664,
      "grad_norm": 0.5849459654617165,
      "learning_rate": 6.278296933676414e-07,
      "loss": 1.3218,
      "step": 1604
    },
    {
      "epoch": 2.568,
      "grad_norm": 0.5821728955619804,
      "learning_rate": 6.233200097272646e-07,
      "loss": 1.2655,
      "step": 1605
    },
    {
      "epoch": 2.5696,
      "grad_norm": 0.593931373057073,
      "learning_rate": 6.188255040763929e-07,
      "loss": 1.2243,
      "step": 1606
    },
    {
      "epoch": 2.5712,
      "grad_norm": 0.5638370594450967,
      "learning_rate": 6.143461920016247e-07,
      "loss": 1.2609,
      "step": 1607
    },
    {
      "epoch": 2.5728,
      "grad_norm": 0.5679734901837457,
      "learning_rate": 6.098820890368696e-07,
      "loss": 1.3002,
      "step": 1608
    },
    {
      "epoch": 2.5744,
      "grad_norm": 0.6025586501038099,
      "learning_rate": 6.054332106632943e-07,
      "loss": 1.2328,
      "step": 1609
    },
    {
      "epoch": 2.576,
      "grad_norm": 0.588186059643339,
      "learning_rate": 6.009995723092655e-07,
      "loss": 1.2852,
      "step": 1610
    },
    {
      "epoch": 2.5776,
      "grad_norm": 0.5557719332053224,
      "learning_rate": 5.965811893503015e-07,
      "loss": 1.2398,
      "step": 1611
    },
    {
      "epoch": 2.5792,
      "grad_norm": 0.5981780811014763,
      "learning_rate": 5.921780771090124e-07,
      "loss": 1.3219,
      "step": 1612
    },
    {
      "epoch": 2.5808,
      "grad_norm": 0.5737468665082083,
      "learning_rate": 5.877902508550542e-07,
      "loss": 1.2809,
      "step": 1613
    },
    {
      "epoch": 2.5824,
      "grad_norm": 0.5570332536963906,
      "learning_rate": 5.834177258050711e-07,
      "loss": 1.2523,
      "step": 1614
    },
    {
      "epoch": 2.584,
      "grad_norm": 0.5606436775835361,
      "learning_rate": 5.790605171226421e-07,
      "loss": 1.2454,
      "step": 1615
    },
    {
      "epoch": 2.5856,
      "grad_norm": 0.5614864385839112,
      "learning_rate": 5.747186399182336e-07,
      "loss": 1.2508,
      "step": 1616
    },
    {
      "epoch": 2.5872,
      "grad_norm": 28.031611183525854,
      "learning_rate": 5.703921092491393e-07,
      "loss": 1.4736,
      "step": 1617
    },
    {
      "epoch": 2.5888,
      "grad_norm": 0.5673226923821233,
      "learning_rate": 5.660809401194362e-07,
      "loss": 1.2661,
      "step": 1618
    },
    {
      "epoch": 2.5904,
      "grad_norm": 0.5935506068485887,
      "learning_rate": 5.617851474799285e-07,
      "loss": 1.2717,
      "step": 1619
    },
    {
      "epoch": 2.592,
      "grad_norm": 0.5748727234437686,
      "learning_rate": 5.575047462280919e-07,
      "loss": 1.3195,
      "step": 1620
    },
    {
      "epoch": 2.5936,
      "grad_norm": 0.5861351079721301,
      "learning_rate": 5.532397512080306e-07,
      "loss": 1.29,
      "step": 1621
    },
    {
      "epoch": 2.5952,
      "grad_norm": 0.5649033443972413,
      "learning_rate": 5.489901772104178e-07,
      "loss": 1.3012,
      "step": 1622
    },
    {
      "epoch": 2.5968,
      "grad_norm": 0.5871986111186086,
      "learning_rate": 5.447560389724499e-07,
      "loss": 1.2666,
      "step": 1623
    },
    {
      "epoch": 2.5984,
      "grad_norm": 0.5990091493336653,
      "learning_rate": 5.405373511777939e-07,
      "loss": 1.2968,
      "step": 1624
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.5879996424377435,
      "learning_rate": 5.363341284565316e-07,
      "loss": 1.3097,
      "step": 1625
    },
    {
      "epoch": 2.6016,
      "grad_norm": 0.6057404474162924,
      "learning_rate": 5.321463853851189e-07,
      "loss": 1.2631,
      "step": 1626
    },
    {
      "epoch": 2.6032,
      "grad_norm": 0.5985274618138063,
      "learning_rate": 5.279741364863244e-07,
      "loss": 1.2672,
      "step": 1627
    },
    {
      "epoch": 2.6048,
      "grad_norm": 0.5885727935018763,
      "learning_rate": 5.238173962291881e-07,
      "loss": 1.2863,
      "step": 1628
    },
    {
      "epoch": 2.6064,
      "grad_norm": 0.5812436146458022,
      "learning_rate": 5.196761790289639e-07,
      "loss": 1.2703,
      "step": 1629
    },
    {
      "epoch": 2.608,
      "grad_norm": 0.5972910615770379,
      "learning_rate": 5.155504992470751e-07,
      "loss": 1.252,
      "step": 1630
    },
    {
      "epoch": 2.6096,
      "grad_norm": 0.5736912111495462,
      "learning_rate": 5.114403711910631e-07,
      "loss": 1.2777,
      "step": 1631
    },
    {
      "epoch": 2.6112,
      "grad_norm": 0.5967311750606411,
      "learning_rate": 5.073458091145328e-07,
      "loss": 1.2516,
      "step": 1632
    },
    {
      "epoch": 2.6128,
      "grad_norm": 0.5739959053879414,
      "learning_rate": 5.032668272171138e-07,
      "loss": 1.2672,
      "step": 1633
    },
    {
      "epoch": 2.6144,
      "grad_norm": 0.5763016848254372,
      "learning_rate": 4.99203439644399e-07,
      "loss": 1.29,
      "step": 1634
    },
    {
      "epoch": 2.616,
      "grad_norm": 0.571445451792373,
      "learning_rate": 4.951556604879049e-07,
      "loss": 1.3069,
      "step": 1635
    },
    {
      "epoch": 2.6176,
      "grad_norm": 0.5690452678840582,
      "learning_rate": 4.911235037850187e-07,
      "loss": 1.285,
      "step": 1636
    },
    {
      "epoch": 2.6192,
      "grad_norm": 0.5853335359673296,
      "learning_rate": 4.871069835189485e-07,
      "loss": 1.3111,
      "step": 1637
    },
    {
      "epoch": 2.6208,
      "grad_norm": 0.5732697986553369,
      "learning_rate": 4.831061136186787e-07,
      "loss": 1.1852,
      "step": 1638
    },
    {
      "epoch": 2.6224,
      "grad_norm": 0.5727471368134726,
      "learning_rate": 4.791209079589165e-07,
      "loss": 1.2975,
      "step": 1639
    },
    {
      "epoch": 2.624,
      "grad_norm": 0.5919144870327738,
      "learning_rate": 4.7515138036005157e-07,
      "loss": 1.2862,
      "step": 1640
    },
    {
      "epoch": 2.6256,
      "grad_norm": 0.6005811157576372,
      "learning_rate": 4.7119754458809727e-07,
      "loss": 1.3553,
      "step": 1641
    },
    {
      "epoch": 2.6272,
      "grad_norm": 0.5761988450938259,
      "learning_rate": 4.672594143546538e-07,
      "loss": 1.2402,
      "step": 1642
    },
    {
      "epoch": 2.6288,
      "grad_norm": 0.6062197191538801,
      "learning_rate": 4.6333700331685385e-07,
      "loss": 1.285,
      "step": 1643
    },
    {
      "epoch": 2.6304,
      "grad_norm": 0.5827390911976612,
      "learning_rate": 4.594303250773152e-07,
      "loss": 1.3339,
      "step": 1644
    },
    {
      "epoch": 2.632,
      "grad_norm": 0.5848572136973985,
      "learning_rate": 4.555393931841001e-07,
      "loss": 1.3009,
      "step": 1645
    },
    {
      "epoch": 2.6336,
      "grad_norm": 0.5925591159222636,
      "learning_rate": 4.5166422113065877e-07,
      "loss": 1.2913,
      "step": 1646
    },
    {
      "epoch": 2.6352,
      "grad_norm": 0.5642443568275711,
      "learning_rate": 4.478048223557907e-07,
      "loss": 1.3053,
      "step": 1647
    },
    {
      "epoch": 2.6368,
      "grad_norm": 0.552050094937832,
      "learning_rate": 4.439612102435942e-07,
      "loss": 1.2482,
      "step": 1648
    },
    {
      "epoch": 2.6384,
      "grad_norm": 0.5600715577486747,
      "learning_rate": 4.401333981234196e-07,
      "loss": 1.2976,
      "step": 1649
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.5599575962003553,
      "learning_rate": 4.3632139926982676e-07,
      "loss": 1.2707,
      "step": 1650
    },
    {
      "epoch": 2.6416,
      "grad_norm": 0.5766769979815745,
      "learning_rate": 4.325252269025315e-07,
      "loss": 1.2767,
      "step": 1651
    },
    {
      "epoch": 2.6432,
      "grad_norm": 0.6171718247271476,
      "learning_rate": 4.287448941863692e-07,
      "loss": 1.2886,
      "step": 1652
    },
    {
      "epoch": 2.6448,
      "grad_norm": 0.5976573851774704,
      "learning_rate": 4.249804142312436e-07,
      "loss": 1.3292,
      "step": 1653
    },
    {
      "epoch": 2.6464,
      "grad_norm": 0.5513927254933091,
      "learning_rate": 4.2123180009207956e-07,
      "loss": 1.2871,
      "step": 1654
    },
    {
      "epoch": 2.648,
      "grad_norm": 0.5623250807225485,
      "learning_rate": 4.1749906476878486e-07,
      "loss": 1.2481,
      "step": 1655
    },
    {
      "epoch": 2.6496,
      "grad_norm": 0.5753661567025741,
      "learning_rate": 4.137822212061965e-07,
      "loss": 1.3133,
      "step": 1656
    },
    {
      "epoch": 2.6512000000000002,
      "grad_norm": 0.5620714384609996,
      "learning_rate": 4.100812822940431e-07,
      "loss": 1.266,
      "step": 1657
    },
    {
      "epoch": 2.6528,
      "grad_norm": 0.5578873555538328,
      "learning_rate": 4.063962608668959e-07,
      "loss": 1.291,
      "step": 1658
    },
    {
      "epoch": 2.6544,
      "grad_norm": 0.5889081633441212,
      "learning_rate": 4.0272716970412516e-07,
      "loss": 1.3594,
      "step": 1659
    },
    {
      "epoch": 2.656,
      "grad_norm": 0.6066947452599447,
      "learning_rate": 3.990740215298583e-07,
      "loss": 1.275,
      "step": 1660
    },
    {
      "epoch": 2.6576,
      "grad_norm": 0.5628117936094952,
      "learning_rate": 3.954368290129301e-07,
      "loss": 1.224,
      "step": 1661
    },
    {
      "epoch": 2.6592000000000002,
      "grad_norm": 0.5783934823020681,
      "learning_rate": 3.918156047668453e-07,
      "loss": 1.2935,
      "step": 1662
    },
    {
      "epoch": 2.6608,
      "grad_norm": 0.5734082446254137,
      "learning_rate": 3.882103613497318e-07,
      "loss": 1.349,
      "step": 1663
    },
    {
      "epoch": 2.6624,
      "grad_norm": 0.579699419091109,
      "learning_rate": 3.84621111264295e-07,
      "loss": 1.2846,
      "step": 1664
    },
    {
      "epoch": 2.664,
      "grad_norm": 0.5771020202617754,
      "learning_rate": 3.810478669577794e-07,
      "loss": 1.2993,
      "step": 1665
    },
    {
      "epoch": 2.6656,
      "grad_norm": 0.5774140522518857,
      "learning_rate": 3.7749064082191976e-07,
      "loss": 1.2841,
      "step": 1666
    },
    {
      "epoch": 2.6672000000000002,
      "grad_norm": 0.5684063715585106,
      "learning_rate": 3.739494451929049e-07,
      "loss": 1.2269,
      "step": 1667
    },
    {
      "epoch": 2.6688,
      "grad_norm": 0.5701622474589515,
      "learning_rate": 3.7042429235132625e-07,
      "loss": 1.2481,
      "step": 1668
    },
    {
      "epoch": 2.6704,
      "grad_norm": 0.5496858234779316,
      "learning_rate": 3.6691519452214387e-07,
      "loss": 1.2665,
      "step": 1669
    },
    {
      "epoch": 2.672,
      "grad_norm": 0.5751386508230802,
      "learning_rate": 3.6342216387464047e-07,
      "loss": 1.3362,
      "step": 1670
    },
    {
      "epoch": 2.6736,
      "grad_norm": 0.5762702343514577,
      "learning_rate": 3.5994521252237516e-07,
      "loss": 1.2857,
      "step": 1671
    },
    {
      "epoch": 2.6752000000000002,
      "grad_norm": 0.551678453357494,
      "learning_rate": 3.564843525231498e-07,
      "loss": 1.221,
      "step": 1672
    },
    {
      "epoch": 2.6768,
      "grad_norm": 0.5732721653299723,
      "learning_rate": 3.53039595878959e-07,
      "loss": 1.3437,
      "step": 1673
    },
    {
      "epoch": 2.6784,
      "grad_norm": 0.5693213885959723,
      "learning_rate": 3.496109545359544e-07,
      "loss": 1.2348,
      "step": 1674
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.5601341261342322,
      "learning_rate": 3.461984403844015e-07,
      "loss": 1.2473,
      "step": 1675
    },
    {
      "epoch": 2.6816,
      "grad_norm": 0.5502348498089403,
      "learning_rate": 3.42802065258635e-07,
      "loss": 1.2515,
      "step": 1676
    },
    {
      "epoch": 2.6832000000000003,
      "grad_norm": 0.5769657960980621,
      "learning_rate": 3.394218409370242e-07,
      "loss": 1.2891,
      "step": 1677
    },
    {
      "epoch": 2.6848,
      "grad_norm": 0.5609439544895503,
      "learning_rate": 3.360577791419256e-07,
      "loss": 1.2657,
      "step": 1678
    },
    {
      "epoch": 2.6864,
      "grad_norm": 0.5633770283399552,
      "learning_rate": 3.3270989153964707e-07,
      "loss": 1.321,
      "step": 1679
    },
    {
      "epoch": 2.6879999999999997,
      "grad_norm": 0.5810337826264936,
      "learning_rate": 3.2937818974040637e-07,
      "loss": 1.2533,
      "step": 1680
    },
    {
      "epoch": 2.6896,
      "grad_norm": 1.0986322331214313,
      "learning_rate": 3.260626852982873e-07,
      "loss": 1.2873,
      "step": 1681
    },
    {
      "epoch": 2.6912000000000003,
      "grad_norm": 0.5672317115860638,
      "learning_rate": 3.227633897112059e-07,
      "loss": 1.2743,
      "step": 1682
    },
    {
      "epoch": 2.6928,
      "grad_norm": 0.5619800101961776,
      "learning_rate": 3.194803144208636e-07,
      "loss": 1.2444,
      "step": 1683
    },
    {
      "epoch": 2.6944,
      "grad_norm": 0.5387287816906554,
      "learning_rate": 3.16213470812714e-07,
      "loss": 1.2103,
      "step": 1684
    },
    {
      "epoch": 2.6959999999999997,
      "grad_norm": 0.5699668411185578,
      "learning_rate": 3.129628702159204e-07,
      "loss": 1.2588,
      "step": 1685
    },
    {
      "epoch": 2.6976,
      "grad_norm": 0.5536577934309528,
      "learning_rate": 3.097285239033138e-07,
      "loss": 1.2914,
      "step": 1686
    },
    {
      "epoch": 2.6992000000000003,
      "grad_norm": 0.5811200109651641,
      "learning_rate": 3.0651044309136016e-07,
      "loss": 1.2844,
      "step": 1687
    },
    {
      "epoch": 2.7008,
      "grad_norm": 0.5710875273086526,
      "learning_rate": 3.033086389401141e-07,
      "loss": 1.2867,
      "step": 1688
    },
    {
      "epoch": 2.7024,
      "grad_norm": 0.5740766823606193,
      "learning_rate": 3.0012312255318696e-07,
      "loss": 1.3529,
      "step": 1689
    },
    {
      "epoch": 2.7039999999999997,
      "grad_norm": 0.5533775083551555,
      "learning_rate": 2.9695390497770535e-07,
      "loss": 1.2336,
      "step": 1690
    },
    {
      "epoch": 2.7056,
      "grad_norm": 0.5546683133370243,
      "learning_rate": 2.93800997204271e-07,
      "loss": 1.3281,
      "step": 1691
    },
    {
      "epoch": 2.7072000000000003,
      "grad_norm": 0.5683610260236122,
      "learning_rate": 2.9066441016692594e-07,
      "loss": 1.2725,
      "step": 1692
    },
    {
      "epoch": 2.7088,
      "grad_norm": 0.5627496839179749,
      "learning_rate": 2.8754415474311235e-07,
      "loss": 1.2553,
      "step": 1693
    },
    {
      "epoch": 2.7104,
      "grad_norm": 0.5740518542563428,
      "learning_rate": 2.844402417536374e-07,
      "loss": 1.2971,
      "step": 1694
    },
    {
      "epoch": 2.7119999999999997,
      "grad_norm": 0.5475813945213532,
      "learning_rate": 2.8135268196263055e-07,
      "loss": 1.2984,
      "step": 1695
    },
    {
      "epoch": 2.7136,
      "grad_norm": 0.5830371739024621,
      "learning_rate": 2.782814860775124e-07,
      "loss": 1.3355,
      "step": 1696
    },
    {
      "epoch": 2.7152,
      "grad_norm": 0.5578289700683081,
      "learning_rate": 2.752266647489549e-07,
      "loss": 1.2511,
      "step": 1697
    },
    {
      "epoch": 2.7168,
      "grad_norm": 0.5879636207984887,
      "learning_rate": 2.7218822857084217e-07,
      "loss": 1.2914,
      "step": 1698
    },
    {
      "epoch": 2.7184,
      "grad_norm": 0.5760655967845457,
      "learning_rate": 2.691661880802382e-07,
      "loss": 1.308,
      "step": 1699
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 0.5476052316325335,
      "learning_rate": 2.661605537573453e-07,
      "loss": 1.2695,
      "step": 1700
    },
    {
      "epoch": 2.7216,
      "grad_norm": 0.5501493961143324,
      "learning_rate": 2.631713360254734e-07,
      "loss": 1.2393,
      "step": 1701
    },
    {
      "epoch": 2.7232,
      "grad_norm": 0.558761059986976,
      "learning_rate": 2.6019854525099977e-07,
      "loss": 1.3131,
      "step": 1702
    },
    {
      "epoch": 2.7248,
      "grad_norm": 0.5683802972185101,
      "learning_rate": 2.572421917433332e-07,
      "loss": 1.2855,
      "step": 1703
    },
    {
      "epoch": 2.7264,
      "grad_norm": 0.5713251549551913,
      "learning_rate": 2.5430228575488156e-07,
      "loss": 1.2939,
      "step": 1704
    },
    {
      "epoch": 2.7279999999999998,
      "grad_norm": 0.5539786940885243,
      "learning_rate": 2.513788374810111e-07,
      "loss": 1.3172,
      "step": 1705
    },
    {
      "epoch": 2.7296,
      "grad_norm": 0.568283191536193,
      "learning_rate": 2.4847185706001643e-07,
      "loss": 1.2919,
      "step": 1706
    },
    {
      "epoch": 2.7312,
      "grad_norm": 0.5724476777466965,
      "learning_rate": 2.455813545730812e-07,
      "loss": 1.2631,
      "step": 1707
    },
    {
      "epoch": 2.7328,
      "grad_norm": 0.5602857315561627,
      "learning_rate": 2.4270734004424643e-07,
      "loss": 1.3004,
      "step": 1708
    },
    {
      "epoch": 2.7344,
      "grad_norm": 0.5805830443010401,
      "learning_rate": 2.39849823440374e-07,
      "loss": 1.2563,
      "step": 1709
    },
    {
      "epoch": 2.7359999999999998,
      "grad_norm": 0.5746651608733866,
      "learning_rate": 2.3700881467111025e-07,
      "loss": 1.2826,
      "step": 1710
    },
    {
      "epoch": 2.7376,
      "grad_norm": 0.5732369434784824,
      "learning_rate": 2.3418432358885633e-07,
      "loss": 1.272,
      "step": 1711
    },
    {
      "epoch": 2.7392,
      "grad_norm": 0.6128333621185169,
      "learning_rate": 2.3137635998872808e-07,
      "loss": 1.3245,
      "step": 1712
    },
    {
      "epoch": 2.7408,
      "grad_norm": 0.5577360368155155,
      "learning_rate": 2.285849336085294e-07,
      "loss": 1.3166,
      "step": 1713
    },
    {
      "epoch": 2.7424,
      "grad_norm": 0.5834057552999696,
      "learning_rate": 2.258100541287117e-07,
      "loss": 1.2921,
      "step": 1714
    },
    {
      "epoch": 2.7439999999999998,
      "grad_norm": 0.5795042116625362,
      "learning_rate": 2.2305173117234236e-07,
      "loss": 1.2871,
      "step": 1715
    },
    {
      "epoch": 2.7456,
      "grad_norm": 0.5625678427798254,
      "learning_rate": 2.2030997430507462e-07,
      "loss": 1.2719,
      "step": 1716
    },
    {
      "epoch": 2.7472,
      "grad_norm": 0.5571211583685395,
      "learning_rate": 2.1758479303510937e-07,
      "loss": 1.2794,
      "step": 1717
    },
    {
      "epoch": 2.7488,
      "grad_norm": 0.5509390062519293,
      "learning_rate": 2.148761968131663e-07,
      "loss": 1.3698,
      "step": 1718
    },
    {
      "epoch": 2.7504,
      "grad_norm": 0.5576861508452913,
      "learning_rate": 2.121841950324488e-07,
      "loss": 1.226,
      "step": 1719
    },
    {
      "epoch": 2.752,
      "grad_norm": 0.5535772121761925,
      "learning_rate": 2.0950879702861082e-07,
      "loss": 1.2583,
      "step": 1720
    },
    {
      "epoch": 2.7536,
      "grad_norm": 0.5746330325355924,
      "learning_rate": 2.0685001207972843e-07,
      "loss": 1.2866,
      "step": 1721
    },
    {
      "epoch": 2.7552,
      "grad_norm": 0.5441800998813338,
      "learning_rate": 2.042078494062616e-07,
      "loss": 1.2149,
      "step": 1722
    },
    {
      "epoch": 2.7568,
      "grad_norm": 0.568078259210911,
      "learning_rate": 2.0158231817102858e-07,
      "loss": 1.3024,
      "step": 1723
    },
    {
      "epoch": 2.7584,
      "grad_norm": 0.5794666774201223,
      "learning_rate": 1.9897342747916938e-07,
      "loss": 1.3321,
      "step": 1724
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.5725342722077056,
      "learning_rate": 1.9638118637811564e-07,
      "loss": 1.203,
      "step": 1725
    },
    {
      "epoch": 2.7616,
      "grad_norm": 0.5562548096235562,
      "learning_rate": 1.9380560385756088e-07,
      "loss": 1.2492,
      "step": 1726
    },
    {
      "epoch": 2.7632,
      "grad_norm": 0.5542222482899214,
      "learning_rate": 1.9124668884942632e-07,
      "loss": 1.2876,
      "step": 1727
    },
    {
      "epoch": 2.7648,
      "grad_norm": 0.5653204470729977,
      "learning_rate": 1.8870445022783234e-07,
      "loss": 1.3023,
      "step": 1728
    },
    {
      "epoch": 2.7664,
      "grad_norm": 0.5828871962448525,
      "learning_rate": 1.861788968090683e-07,
      "loss": 1.3015,
      "step": 1729
    },
    {
      "epoch": 2.768,
      "grad_norm": 0.5412710518218398,
      "learning_rate": 1.8367003735155764e-07,
      "loss": 1.2485,
      "step": 1730
    },
    {
      "epoch": 2.7696,
      "grad_norm": 0.5553840446624939,
      "learning_rate": 1.8117788055583286e-07,
      "loss": 1.2478,
      "step": 1731
    },
    {
      "epoch": 2.7712,
      "grad_norm": 0.5622052949506016,
      "learning_rate": 1.7870243506450113e-07,
      "loss": 1.2996,
      "step": 1732
    },
    {
      "epoch": 2.7728,
      "grad_norm": 0.555423566375322,
      "learning_rate": 1.762437094622177e-07,
      "loss": 1.2678,
      "step": 1733
    },
    {
      "epoch": 2.7744,
      "grad_norm": 0.5677423187219658,
      "learning_rate": 1.738017122756541e-07,
      "loss": 1.289,
      "step": 1734
    },
    {
      "epoch": 2.776,
      "grad_norm": 0.5573862376832547,
      "learning_rate": 1.713764519734673e-07,
      "loss": 1.2801,
      "step": 1735
    },
    {
      "epoch": 2.7776,
      "grad_norm": 0.5557632302011095,
      "learning_rate": 1.68967936966275e-07,
      "loss": 1.2699,
      "step": 1736
    },
    {
      "epoch": 2.7792,
      "grad_norm": 0.5778018573184343,
      "learning_rate": 1.6657617560662088e-07,
      "loss": 1.3042,
      "step": 1737
    },
    {
      "epoch": 2.7808,
      "grad_norm": 0.5732861442283814,
      "learning_rate": 1.6420117618895003e-07,
      "loss": 1.3161,
      "step": 1738
    },
    {
      "epoch": 2.7824,
      "grad_norm": 0.5566974290902864,
      "learning_rate": 1.6184294694957747e-07,
      "loss": 1.2192,
      "step": 1739
    },
    {
      "epoch": 2.784,
      "grad_norm": 0.5667622941434295,
      "learning_rate": 1.5950149606666077e-07,
      "loss": 1.317,
      "step": 1740
    },
    {
      "epoch": 2.7856,
      "grad_norm": 0.5737478731560492,
      "learning_rate": 1.5717683166017184e-07,
      "loss": 1.3423,
      "step": 1741
    },
    {
      "epoch": 2.7872,
      "grad_norm": 0.5554914118127725,
      "learning_rate": 1.5486896179186693e-07,
      "loss": 1.2688,
      "step": 1742
    },
    {
      "epoch": 2.7888,
      "grad_norm": 0.565717003208711,
      "learning_rate": 1.5257789446526172e-07,
      "loss": 1.2326,
      "step": 1743
    },
    {
      "epoch": 2.7904,
      "grad_norm": 0.5616284595491774,
      "learning_rate": 1.5030363762560228e-07,
      "loss": 1.2687,
      "step": 1744
    },
    {
      "epoch": 2.792,
      "grad_norm": 0.586962313159236,
      "learning_rate": 1.480461991598353e-07,
      "loss": 1.2807,
      "step": 1745
    },
    {
      "epoch": 2.7936,
      "grad_norm": 0.5674137370367577,
      "learning_rate": 1.458055868965841e-07,
      "loss": 1.2574,
      "step": 1746
    },
    {
      "epoch": 2.7952,
      "grad_norm": 0.5742119208527271,
      "learning_rate": 1.4358180860611913e-07,
      "loss": 1.2795,
      "step": 1747
    },
    {
      "epoch": 2.7968,
      "grad_norm": 0.5880441698297544,
      "learning_rate": 1.4137487200033383e-07,
      "loss": 1.2245,
      "step": 1748
    },
    {
      "epoch": 2.7984,
      "grad_norm": 0.5608500833757297,
      "learning_rate": 1.3918478473271325e-07,
      "loss": 1.3545,
      "step": 1749
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.5815096323954879,
      "learning_rate": 1.3701155439831249e-07,
      "loss": 1.2877,
      "step": 1750
    },
    {
      "epoch": 2.8016,
      "grad_norm": 0.5714319051572612,
      "learning_rate": 1.3485518853372625e-07,
      "loss": 1.3157,
      "step": 1751
    },
    {
      "epoch": 2.8032,
      "grad_norm": 0.5626878091502198,
      "learning_rate": 1.3271569461706547e-07,
      "loss": 1.2701,
      "step": 1752
    },
    {
      "epoch": 2.8048,
      "grad_norm": 0.5852073424169997,
      "learning_rate": 1.305930800679317e-07,
      "loss": 1.2833,
      "step": 1753
    },
    {
      "epoch": 2.8064,
      "grad_norm": 0.5697520027057954,
      "learning_rate": 1.2848735224738729e-07,
      "loss": 1.2719,
      "step": 1754
    },
    {
      "epoch": 2.808,
      "grad_norm": 0.5628407010027053,
      "learning_rate": 1.2639851845793583e-07,
      "loss": 1.3005,
      "step": 1755
    },
    {
      "epoch": 2.8096,
      "grad_norm": 0.560425393251839,
      "learning_rate": 1.2432658594349113e-07,
      "loss": 1.2961,
      "step": 1756
    },
    {
      "epoch": 2.8112,
      "grad_norm": 0.5568969594389682,
      "learning_rate": 1.2227156188935552e-07,
      "loss": 1.2785,
      "step": 1757
    },
    {
      "epoch": 2.8128,
      "grad_norm": 0.5545317759699009,
      "learning_rate": 1.202334534221955e-07,
      "loss": 1.2611,
      "step": 1758
    },
    {
      "epoch": 2.8144,
      "grad_norm": 0.5719580494964629,
      "learning_rate": 1.1821226761001391e-07,
      "loss": 1.3211,
      "step": 1759
    },
    {
      "epoch": 2.816,
      "grad_norm": 0.5672206847046435,
      "learning_rate": 1.1620801146212723e-07,
      "loss": 1.2717,
      "step": 1760
    },
    {
      "epoch": 2.8176,
      "grad_norm": 0.5589192881602912,
      "learning_rate": 1.1422069192914221e-07,
      "loss": 1.3331,
      "step": 1761
    },
    {
      "epoch": 2.8192,
      "grad_norm": 0.5672143514850739,
      "learning_rate": 1.1225031590292923e-07,
      "loss": 1.2678,
      "step": 1762
    },
    {
      "epoch": 2.8208,
      "grad_norm": 0.5711127583971003,
      "learning_rate": 1.1029689021660183e-07,
      "loss": 1.3863,
      "step": 1763
    },
    {
      "epoch": 2.8224,
      "grad_norm": 0.5684526305930276,
      "learning_rate": 1.0836042164448945e-07,
      "loss": 1.231,
      "step": 1764
    },
    {
      "epoch": 2.824,
      "grad_norm": 0.5946230133084417,
      "learning_rate": 1.0644091690211633e-07,
      "loss": 1.2618,
      "step": 1765
    },
    {
      "epoch": 2.8256,
      "grad_norm": 0.5519508915632919,
      "learning_rate": 1.0453838264617711e-07,
      "loss": 1.3305,
      "step": 1766
    },
    {
      "epoch": 2.8272,
      "grad_norm": 0.5701454398204907,
      "learning_rate": 1.0265282547451405e-07,
      "loss": 1.2986,
      "step": 1767
    },
    {
      "epoch": 2.8288,
      "grad_norm": 0.5768042813537503,
      "learning_rate": 1.0078425192609487e-07,
      "loss": 1.3484,
      "step": 1768
    },
    {
      "epoch": 2.8304,
      "grad_norm": 0.57181150162004,
      "learning_rate": 9.893266848098826e-08,
      "loss": 1.3113,
      "step": 1769
    },
    {
      "epoch": 2.832,
      "grad_norm": 0.5722006544454465,
      "learning_rate": 9.709808156034394e-08,
      "loss": 1.2364,
      "step": 1770
    },
    {
      "epoch": 2.8336,
      "grad_norm": 0.5496225231592293,
      "learning_rate": 9.528049752636714e-08,
      "loss": 1.2538,
      "step": 1771
    },
    {
      "epoch": 2.8352,
      "grad_norm": 0.5786938505217666,
      "learning_rate": 9.347992268230022e-08,
      "loss": 1.3421,
      "step": 1772
    },
    {
      "epoch": 2.8368,
      "grad_norm": 0.5502313208297342,
      "learning_rate": 9.169636327239883e-08,
      "loss": 1.2799,
      "step": 1773
    },
    {
      "epoch": 2.8384,
      "grad_norm": 0.5674231475812329,
      "learning_rate": 8.992982548190809e-08,
      "loss": 1.2982,
      "step": 1774
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.5653925881754057,
      "learning_rate": 8.818031543704641e-08,
      "loss": 1.2016,
      "step": 1775
    },
    {
      "epoch": 2.8416,
      "grad_norm": 0.5634078886369406,
      "learning_rate": 8.644783920498001e-08,
      "loss": 1.278,
      "step": 1776
    },
    {
      "epoch": 2.8432,
      "grad_norm": 0.5701044492226841,
      "learning_rate": 8.473240279380235e-08,
      "loss": 1.2655,
      "step": 1777
    },
    {
      "epoch": 2.8448,
      "grad_norm": 0.5721073707411589,
      "learning_rate": 8.303401215251583e-08,
      "loss": 1.2294,
      "step": 1778
    },
    {
      "epoch": 2.8464,
      "grad_norm": 0.5721270338942107,
      "learning_rate": 8.135267317100792e-08,
      "loss": 1.3198,
      "step": 1779
    },
    {
      "epoch": 2.848,
      "grad_norm": 0.5637674858989231,
      "learning_rate": 7.968839168003395e-08,
      "loss": 1.2734,
      "step": 1780
    },
    {
      "epoch": 2.8496,
      "grad_norm": 6.366281037675526,
      "learning_rate": 7.804117345119266e-08,
      "loss": 1.3402,
      "step": 1781
    },
    {
      "epoch": 2.8512,
      "grad_norm": 0.5496011659804143,
      "learning_rate": 7.64110241969107e-08,
      "loss": 1.2798,
      "step": 1782
    },
    {
      "epoch": 2.8528000000000002,
      "grad_norm": 0.5568680384305391,
      "learning_rate": 7.479794957042041e-08,
      "loss": 1.2535,
      "step": 1783
    },
    {
      "epoch": 2.8544,
      "grad_norm": 0.5602222548451022,
      "learning_rate": 7.320195516574036e-08,
      "loss": 1.2926,
      "step": 1784
    },
    {
      "epoch": 2.856,
      "grad_norm": 0.5536509984807575,
      "learning_rate": 7.16230465176565e-08,
      "loss": 1.2401,
      "step": 1785
    },
    {
      "epoch": 2.8576,
      "grad_norm": 0.5610121397771486,
      "learning_rate": 7.00612291017022e-08,
      "loss": 1.3029,
      "step": 1786
    },
    {
      "epoch": 2.8592,
      "grad_norm": 0.552918487471406,
      "learning_rate": 6.851650833414103e-08,
      "loss": 1.3144,
      "step": 1787
    },
    {
      "epoch": 2.8608000000000002,
      "grad_norm": 0.5854399867981849,
      "learning_rate": 6.698888957194505e-08,
      "loss": 1.2908,
      "step": 1788
    },
    {
      "epoch": 2.8624,
      "grad_norm": 0.5554656360953278,
      "learning_rate": 6.547837811277824e-08,
      "loss": 1.2567,
      "step": 1789
    },
    {
      "epoch": 2.864,
      "grad_norm": 0.5554586192030543,
      "learning_rate": 6.39849791949787e-08,
      "loss": 1.2637,
      "step": 1790
    },
    {
      "epoch": 2.8656,
      "grad_norm": 0.5621640054310905,
      "learning_rate": 6.250869799753866e-08,
      "loss": 1.248,
      "step": 1791
    },
    {
      "epoch": 2.8672,
      "grad_norm": 0.5654919404635569,
      "learning_rate": 6.104953964008897e-08,
      "loss": 1.3362,
      "step": 1792
    },
    {
      "epoch": 2.8688000000000002,
      "grad_norm": 0.5309418264820442,
      "learning_rate": 5.960750918287627e-08,
      "loss": 1.246,
      "step": 1793
    },
    {
      "epoch": 2.8704,
      "grad_norm": 0.5701070902431137,
      "learning_rate": 5.818261162675309e-08,
      "loss": 1.3248,
      "step": 1794
    },
    {
      "epoch": 2.872,
      "grad_norm": 0.5661486678320851,
      "learning_rate": 5.677485191315391e-08,
      "loss": 1.2279,
      "step": 1795
    },
    {
      "epoch": 2.8736,
      "grad_norm": 0.5598007769326387,
      "learning_rate": 5.538423492408129e-08,
      "loss": 1.2444,
      "step": 1796
    },
    {
      "epoch": 2.8752,
      "grad_norm": 0.5606043216059553,
      "learning_rate": 5.401076548208761e-08,
      "loss": 1.3009,
      "step": 1797
    },
    {
      "epoch": 2.8768000000000002,
      "grad_norm": 0.5753304774048688,
      "learning_rate": 5.265444835025946e-08,
      "loss": 1.3657,
      "step": 1798
    },
    {
      "epoch": 2.8784,
      "grad_norm": 0.5709954495994486,
      "learning_rate": 5.1315288232201e-08,
      "loss": 1.349,
      "step": 1799
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.5619437209865865,
      "learning_rate": 4.9993289772015116e-08,
      "loss": 1.3135,
      "step": 1800
    },
    {
      "epoch": 2.8816,
      "grad_norm": 0.5531158796461306,
      "learning_rate": 4.8688457554291746e-08,
      "loss": 1.241,
      "step": 1801
    },
    {
      "epoch": 2.8832,
      "grad_norm": 0.5835503980379451,
      "learning_rate": 4.7400796104088434e-08,
      "loss": 1.3257,
      "step": 1802
    },
    {
      "epoch": 2.8848000000000003,
      "grad_norm": 0.568508785117014,
      "learning_rate": 4.613030988691536e-08,
      "loss": 1.3472,
      "step": 1803
    },
    {
      "epoch": 2.8864,
      "grad_norm": 0.5568288660080997,
      "learning_rate": 4.4877003308722575e-08,
      "loss": 1.2067,
      "step": 1804
    },
    {
      "epoch": 2.888,
      "grad_norm": 0.5540164328282703,
      "learning_rate": 4.364088071587891e-08,
      "loss": 1.2277,
      "step": 1805
    },
    {
      "epoch": 2.8895999999999997,
      "grad_norm": 0.5577958319822962,
      "learning_rate": 4.2421946395164174e-08,
      "loss": 1.2974,
      "step": 1806
    },
    {
      "epoch": 2.8912,
      "grad_norm": 0.5623634884160116,
      "learning_rate": 4.1220204573747534e-08,
      "loss": 1.309,
      "step": 1807
    },
    {
      "epoch": 2.8928000000000003,
      "grad_norm": 0.5614708645188787,
      "learning_rate": 4.0035659419178086e-08,
      "loss": 1.2888,
      "step": 1808
    },
    {
      "epoch": 2.8944,
      "grad_norm": 0.5674167782133157,
      "learning_rate": 3.88683150393665e-08,
      "loss": 1.2682,
      "step": 1809
    },
    {
      "epoch": 2.896,
      "grad_norm": 0.5383311281064183,
      "learning_rate": 3.771817548257395e-08,
      "loss": 1.2653,
      "step": 1810
    },
    {
      "epoch": 2.8975999999999997,
      "grad_norm": 0.5441773463601839,
      "learning_rate": 3.658524473739544e-08,
      "loss": 1.256,
      "step": 1811
    },
    {
      "epoch": 2.8992,
      "grad_norm": 0.5638768556744691,
      "learning_rate": 3.546952673274817e-08,
      "loss": 1.2126,
      "step": 1812
    },
    {
      "epoch": 2.9008000000000003,
      "grad_norm": 0.6101605467743239,
      "learning_rate": 3.437102533785541e-08,
      "loss": 1.2521,
      "step": 1813
    },
    {
      "epoch": 2.9024,
      "grad_norm": 0.5733990722198719,
      "learning_rate": 3.328974436223709e-08,
      "loss": 1.3394,
      "step": 1814
    },
    {
      "epoch": 2.904,
      "grad_norm": 0.5648310913843545,
      "learning_rate": 3.2225687555690886e-08,
      "loss": 1.258,
      "step": 1815
    },
    {
      "epoch": 2.9055999999999997,
      "grad_norm": 0.5715192301222051,
      "learning_rate": 3.117885860828396e-08,
      "loss": 1.3024,
      "step": 1816
    },
    {
      "epoch": 2.9072,
      "grad_norm": 0.5421697599843172,
      "learning_rate": 3.014926115034012e-08,
      "loss": 1.2518,
      "step": 1817
    },
    {
      "epoch": 2.9088000000000003,
      "grad_norm": 0.6030524907953939,
      "learning_rate": 2.9136898752422648e-08,
      "loss": 1.2601,
      "step": 1818
    },
    {
      "epoch": 2.9104,
      "grad_norm": 0.5650524350504813,
      "learning_rate": 2.8141774925327103e-08,
      "loss": 1.2847,
      "step": 1819
    },
    {
      "epoch": 2.912,
      "grad_norm": 0.5617141349030577,
      "learning_rate": 2.7163893120066288e-08,
      "loss": 1.2802,
      "step": 1820
    },
    {
      "epoch": 2.9135999999999997,
      "grad_norm": 0.5646698917303378,
      "learning_rate": 2.6203256727859172e-08,
      "loss": 1.2823,
      "step": 1821
    },
    {
      "epoch": 2.9152,
      "grad_norm": 0.5904302023208247,
      "learning_rate": 2.5259869080118127e-08,
      "loss": 1.2973,
      "step": 1822
    },
    {
      "epoch": 2.9168,
      "grad_norm": 0.5438195694288955,
      "learning_rate": 2.4333733448440033e-08,
      "loss": 1.1975,
      "step": 1823
    },
    {
      "epoch": 2.9184,
      "grad_norm": 0.5620680029264215,
      "learning_rate": 2.34248530445913e-08,
      "loss": 1.2398,
      "step": 1824
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.5676068315943834,
      "learning_rate": 2.2533231020499536e-08,
      "loss": 1.2578,
      "step": 1825
    },
    {
      "epoch": 2.9215999999999998,
      "grad_norm": 0.5537138711886406,
      "learning_rate": 2.1658870468241332e-08,
      "loss": 1.2717,
      "step": 1826
    },
    {
      "epoch": 2.9232,
      "grad_norm": 0.5549504004969718,
      "learning_rate": 2.0801774420031172e-08,
      "loss": 1.2809,
      "step": 1827
    },
    {
      "epoch": 2.9248,
      "grad_norm": 0.5847037512304446,
      "learning_rate": 1.9961945848213092e-08,
      "loss": 1.3137,
      "step": 1828
    },
    {
      "epoch": 2.9264,
      "grad_norm": 0.5568171869110831,
      "learning_rate": 1.9139387665247922e-08,
      "loss": 1.3417,
      "step": 1829
    },
    {
      "epoch": 2.928,
      "grad_norm": 0.5656450426225103,
      "learning_rate": 1.8334102723703286e-08,
      "loss": 1.3694,
      "step": 1830
    },
    {
      "epoch": 2.9295999999999998,
      "grad_norm": 0.5497568620828591,
      "learning_rate": 1.754609381624639e-08,
      "loss": 1.2526,
      "step": 1831
    },
    {
      "epoch": 2.9312,
      "grad_norm": 0.5651704113088953,
      "learning_rate": 1.677536367563126e-08,
      "loss": 1.3486,
      "step": 1832
    },
    {
      "epoch": 2.9328,
      "grad_norm": 0.5456453637320016,
      "learning_rate": 1.6021914974690413e-08,
      "loss": 1.2594,
      "step": 1833
    },
    {
      "epoch": 2.9344,
      "grad_norm": 0.5635851957646197,
      "learning_rate": 1.5285750326325953e-08,
      "loss": 1.287,
      "step": 1834
    },
    {
      "epoch": 2.936,
      "grad_norm": 0.5809949240062731,
      "learning_rate": 1.4566872283500733e-08,
      "loss": 1.2919,
      "step": 1835
    },
    {
      "epoch": 2.9375999999999998,
      "grad_norm": 0.5592087158569868,
      "learning_rate": 1.3865283339228319e-08,
      "loss": 1.2734,
      "step": 1836
    },
    {
      "epoch": 2.9392,
      "grad_norm": 0.5640233964667584,
      "learning_rate": 1.3180985926564693e-08,
      "loss": 1.2973,
      "step": 1837
    },
    {
      "epoch": 2.9408,
      "grad_norm": 0.5555328182384419,
      "learning_rate": 1.2513982418601024e-08,
      "loss": 1.2582,
      "step": 1838
    },
    {
      "epoch": 2.9424,
      "grad_norm": 0.5790138804370322,
      "learning_rate": 1.1864275128454783e-08,
      "loss": 1.3042,
      "step": 1839
    },
    {
      "epoch": 2.944,
      "grad_norm": 0.5431784501088861,
      "learning_rate": 1.1231866309259764e-08,
      "loss": 1.2904,
      "step": 1840
    },
    {
      "epoch": 2.9455999999999998,
      "grad_norm": 0.5504903423916736,
      "learning_rate": 1.0616758154161633e-08,
      "loss": 1.2581,
      "step": 1841
    },
    {
      "epoch": 2.9472,
      "grad_norm": 0.5809388281956364,
      "learning_rate": 1.0018952796307934e-08,
      "loss": 1.3023,
      "step": 1842
    },
    {
      "epoch": 2.9488,
      "grad_norm": 0.5645677054821269,
      "learning_rate": 9.438452308841995e-09,
      "loss": 1.3208,
      "step": 1843
    },
    {
      "epoch": 2.9504,
      "grad_norm": 0.5325031893334977,
      "learning_rate": 8.87525870489514e-09,
      "loss": 1.2355,
      "step": 1844
    },
    {
      "epoch": 2.952,
      "grad_norm": 0.5571000231733368,
      "learning_rate": 8.329373937578378e-09,
      "loss": 1.3121,
      "step": 1845
    },
    {
      "epoch": 2.9536,
      "grad_norm": 0.5610343981773074,
      "learning_rate": 7.800799899979061e-09,
      "loss": 1.3284,
      "step": 1846
    },
    {
      "epoch": 2.9552,
      "grad_norm": 0.5634135866586591,
      "learning_rate": 7.289538425150899e-09,
      "loss": 1.3221,
      "step": 1847
    },
    {
      "epoch": 2.9568,
      "grad_norm": 0.5814188717158965,
      "learning_rate": 6.7955912861095155e-09,
      "loss": 1.2573,
      "step": 1848
    },
    {
      "epoch": 2.9584,
      "grad_norm": 0.549713451310637,
      "learning_rate": 6.31896019582523e-09,
      "loss": 1.2918,
      "step": 1849
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.5654088750545045,
      "learning_rate": 5.8596468072180665e-09,
      "loss": 1.1928,
      "step": 1850
    },
    {
      "epoch": 2.9616,
      "grad_norm": 0.5736089748384048,
      "learning_rate": 5.417652713152199e-09,
      "loss": 1.2595,
      "step": 1851
    },
    {
      "epoch": 2.9632,
      "grad_norm": 0.5767876440656464,
      "learning_rate": 4.992979446428736e-09,
      "loss": 1.2654,
      "step": 1852
    },
    {
      "epoch": 2.9648,
      "grad_norm": 0.5578324163588724,
      "learning_rate": 4.585628479781279e-09,
      "loss": 1.2738,
      "step": 1853
    },
    {
      "epoch": 2.9664,
      "grad_norm": 0.5489877360044414,
      "learning_rate": 4.195601225872592e-09,
      "loss": 1.2624,
      "step": 1854
    },
    {
      "epoch": 2.968,
      "grad_norm": 0.565530190957416,
      "learning_rate": 3.822899037286276e-09,
      "loss": 1.3164,
      "step": 1855
    },
    {
      "epoch": 2.9696,
      "grad_norm": 0.5896407244996497,
      "learning_rate": 3.4675232065256583e-09,
      "loss": 1.2201,
      "step": 1856
    },
    {
      "epoch": 2.9712,
      "grad_norm": 0.5413621768743335,
      "learning_rate": 3.129474966006574e-09,
      "loss": 1.2589,
      "step": 1857
    },
    {
      "epoch": 2.9728,
      "grad_norm": 0.56093934590888,
      "learning_rate": 2.808755488054038e-09,
      "loss": 1.3523,
      "step": 1858
    },
    {
      "epoch": 2.9744,
      "grad_norm": 0.5812028747372194,
      "learning_rate": 2.5053658848989137e-09,
      "loss": 1.3282,
      "step": 1859
    },
    {
      "epoch": 2.976,
      "grad_norm": 0.5854664286839666,
      "learning_rate": 2.219307208672361e-09,
      "loss": 1.2751,
      "step": 1860
    },
    {
      "epoch": 2.9776,
      "grad_norm": 0.5490658045366089,
      "learning_rate": 1.9505804514047266e-09,
      "loss": 1.2188,
      "step": 1861
    },
    {
      "epoch": 2.9792,
      "grad_norm": 0.5697794581791655,
      "learning_rate": 1.6991865450188827e-09,
      "loss": 1.2717,
      "step": 1862
    },
    {
      "epoch": 2.9808,
      "grad_norm": 0.5576813286978362,
      "learning_rate": 1.465126361330227e-09,
      "loss": 1.2799,
      "step": 1863
    },
    {
      "epoch": 2.9824,
      "grad_norm": 0.5561489197225744,
      "learning_rate": 1.2484007120411312e-09,
      "loss": 1.2535,
      "step": 1864
    },
    {
      "epoch": 2.984,
      "grad_norm": 0.5610779612441494,
      "learning_rate": 1.0490103487392766e-09,
      "loss": 1.3033,
      "step": 1865
    },
    {
      "epoch": 2.9856,
      "grad_norm": 0.5763801339272827,
      "learning_rate": 8.669559628954327e-10,
      "loss": 1.2831,
      "step": 1866
    },
    {
      "epoch": 2.9872,
      "grad_norm": 0.6081020090289561,
      "learning_rate": 7.02238185860682e-10,
      "loss": 1.3234,
      "step": 1867
    },
    {
      "epoch": 2.9888,
      "grad_norm": 0.5551611136678277,
      "learning_rate": 5.54857588862534e-10,
      "loss": 1.3324,
      "step": 1868
    },
    {
      "epoch": 2.9904,
      "grad_norm": 0.567761131568866,
      "learning_rate": 4.2481468300603625e-10,
      "loss": 1.2678,
      "step": 1869
    },
    {
      "epoch": 2.992,
      "grad_norm": 0.5729007837194088,
      "learning_rate": 3.1210991927044244e-10,
      "loss": 1.2818,
      "step": 1870
    },
    {
      "epoch": 2.9936,
      "grad_norm": 0.5548046483717011,
      "learning_rate": 2.167436885064378e-10,
      "loss": 1.2796,
      "step": 1871
    },
    {
      "epoch": 2.9952,
      "grad_norm": 0.5455194698153667,
      "learning_rate": 1.387163214372489e-10,
      "loss": 1.3194,
      "step": 1872
    },
    {
      "epoch": 2.9968,
      "grad_norm": 0.5628469758619941,
      "learning_rate": 7.80280886558682e-11,
      "loss": 1.3299,
      "step": 1873
    },
    {
      "epoch": 2.9984,
      "grad_norm": 0.598272969454025,
      "learning_rate": 3.467920062394381e-11,
      "loss": 1.2944,
      "step": 1874
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.5621518596924057,
      "learning_rate": 8.669807672334606e-12,
      "loss": 1.3106,
      "step": 1875
    }
  ],
  "logging_steps": 1,
  "max_steps": 1875,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1439107152936960.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}