{
  "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": 7.911514407844048,
      "learning_rate": 0.0,
      "loss": 1.9118,
      "step": 1
    },
    {
      "epoch": 0.0032,
      "grad_norm": 7.589583718996209,
      "learning_rate": 5.319148936170213e-08,
      "loss": 1.9163,
      "step": 2
    },
    {
      "epoch": 0.0048,
      "grad_norm": 7.85131340849695,
      "learning_rate": 1.0638297872340426e-07,
      "loss": 1.9329,
      "step": 3
    },
    {
      "epoch": 0.0064,
      "grad_norm": 7.974722306395065,
      "learning_rate": 1.5957446808510638e-07,
      "loss": 1.9427,
      "step": 4
    },
    {
      "epoch": 0.008,
      "grad_norm": 7.615721008421588,
      "learning_rate": 2.1276595744680852e-07,
      "loss": 1.8895,
      "step": 5
    },
    {
      "epoch": 0.0096,
      "grad_norm": 7.610504547331789,
      "learning_rate": 2.6595744680851066e-07,
      "loss": 2.0018,
      "step": 6
    },
    {
      "epoch": 0.0112,
      "grad_norm": 7.628494657434167,
      "learning_rate": 3.1914893617021275e-07,
      "loss": 1.8836,
      "step": 7
    },
    {
      "epoch": 0.0128,
      "grad_norm": 7.219778302781298,
      "learning_rate": 3.723404255319149e-07,
      "loss": 1.9085,
      "step": 8
    },
    {
      "epoch": 0.0144,
      "grad_norm": 7.638897763817933,
      "learning_rate": 4.2553191489361704e-07,
      "loss": 1.9835,
      "step": 9
    },
    {
      "epoch": 0.016,
      "grad_norm": 6.726597205367486,
      "learning_rate": 4.787234042553192e-07,
      "loss": 1.9029,
      "step": 10
    },
    {
      "epoch": 0.0176,
      "grad_norm": 5.576768940586463,
      "learning_rate": 5.319148936170213e-07,
      "loss": 1.8727,
      "step": 11
    },
    {
      "epoch": 0.0192,
      "grad_norm": 5.371673726127225,
      "learning_rate": 5.851063829787235e-07,
      "loss": 1.894,
      "step": 12
    },
    {
      "epoch": 0.0208,
      "grad_norm": 5.33315954769635,
      "learning_rate": 6.382978723404255e-07,
      "loss": 1.8427,
      "step": 13
    },
    {
      "epoch": 0.0224,
      "grad_norm": 5.603246782127843,
      "learning_rate": 6.914893617021278e-07,
      "loss": 1.8511,
      "step": 14
    },
    {
      "epoch": 0.024,
      "grad_norm": 5.223993453822097,
      "learning_rate": 7.446808510638298e-07,
      "loss": 1.8775,
      "step": 15
    },
    {
      "epoch": 0.0256,
      "grad_norm": 5.329260698466068,
      "learning_rate": 7.97872340425532e-07,
      "loss": 1.8893,
      "step": 16
    },
    {
      "epoch": 0.0272,
      "grad_norm": 4.4526317630513095,
      "learning_rate": 8.510638297872341e-07,
      "loss": 1.8057,
      "step": 17
    },
    {
      "epoch": 0.0288,
      "grad_norm": 4.145130819952591,
      "learning_rate": 9.042553191489363e-07,
      "loss": 1.8592,
      "step": 18
    },
    {
      "epoch": 0.0304,
      "grad_norm": 4.5092095420656655,
      "learning_rate": 9.574468085106384e-07,
      "loss": 1.7782,
      "step": 19
    },
    {
      "epoch": 0.032,
      "grad_norm": 3.492215446039818,
      "learning_rate": 1.0106382978723404e-06,
      "loss": 1.7685,
      "step": 20
    },
    {
      "epoch": 0.0336,
      "grad_norm": 3.018542834755133,
      "learning_rate": 1.0638297872340427e-06,
      "loss": 1.8039,
      "step": 21
    },
    {
      "epoch": 0.0352,
      "grad_norm": 2.827625302017259,
      "learning_rate": 1.1170212765957447e-06,
      "loss": 1.7327,
      "step": 22
    },
    {
      "epoch": 0.0368,
      "grad_norm": 2.8892720008022446,
      "learning_rate": 1.170212765957447e-06,
      "loss": 1.7636,
      "step": 23
    },
    {
      "epoch": 0.0384,
      "grad_norm": 2.47354154460494,
      "learning_rate": 1.223404255319149e-06,
      "loss": 1.7682,
      "step": 24
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.011511693128511,
      "learning_rate": 1.276595744680851e-06,
      "loss": 1.7588,
      "step": 25
    },
    {
      "epoch": 0.0416,
      "grad_norm": 2.223429444553426,
      "learning_rate": 1.3297872340425533e-06,
      "loss": 1.7619,
      "step": 26
    },
    {
      "epoch": 0.0432,
      "grad_norm": 2.9936836015109156,
      "learning_rate": 1.3829787234042555e-06,
      "loss": 1.7028,
      "step": 27
    },
    {
      "epoch": 0.0448,
      "grad_norm": 2.626119203506752,
      "learning_rate": 1.4361702127659578e-06,
      "loss": 1.7164,
      "step": 28
    },
    {
      "epoch": 0.0464,
      "grad_norm": 1.7423531619711112,
      "learning_rate": 1.4893617021276596e-06,
      "loss": 1.6989,
      "step": 29
    },
    {
      "epoch": 0.048,
      "grad_norm": 1.9707378777775941,
      "learning_rate": 1.5425531914893618e-06,
      "loss": 1.6833,
      "step": 30
    },
    {
      "epoch": 0.0496,
      "grad_norm": 1.9609745810910064,
      "learning_rate": 1.595744680851064e-06,
      "loss": 1.7348,
      "step": 31
    },
    {
      "epoch": 0.0512,
      "grad_norm": 1.681448457097492,
      "learning_rate": 1.648936170212766e-06,
      "loss": 1.7088,
      "step": 32
    },
    {
      "epoch": 0.0528,
      "grad_norm": 1.3140483899527649,
      "learning_rate": 1.7021276595744682e-06,
      "loss": 1.7143,
      "step": 33
    },
    {
      "epoch": 0.0544,
      "grad_norm": 1.6329486647621383,
      "learning_rate": 1.7553191489361704e-06,
      "loss": 1.633,
      "step": 34
    },
    {
      "epoch": 0.056,
      "grad_norm": 1.534184975262561,
      "learning_rate": 1.8085106382978727e-06,
      "loss": 1.6234,
      "step": 35
    },
    {
      "epoch": 0.0576,
      "grad_norm": 1.21999730521497,
      "learning_rate": 1.8617021276595745e-06,
      "loss": 1.7035,
      "step": 36
    },
    {
      "epoch": 0.0592,
      "grad_norm": 1.2305687103932375,
      "learning_rate": 1.9148936170212767e-06,
      "loss": 1.6548,
      "step": 37
    },
    {
      "epoch": 0.0608,
      "grad_norm": 1.3953146489304726,
      "learning_rate": 1.968085106382979e-06,
      "loss": 1.5632,
      "step": 38
    },
    {
      "epoch": 0.0624,
      "grad_norm": 1.2793935871154647,
      "learning_rate": 2.021276595744681e-06,
      "loss": 1.6938,
      "step": 39
    },
    {
      "epoch": 0.064,
      "grad_norm": 1.1393630949447648,
      "learning_rate": 2.074468085106383e-06,
      "loss": 1.6367,
      "step": 40
    },
    {
      "epoch": 0.0656,
      "grad_norm": 1.287087998108404,
      "learning_rate": 2.1276595744680853e-06,
      "loss": 1.701,
      "step": 41
    },
    {
      "epoch": 0.0672,
      "grad_norm": 1.1173046005183394,
      "learning_rate": 2.1808510638297876e-06,
      "loss": 1.6264,
      "step": 42
    },
    {
      "epoch": 0.0688,
      "grad_norm": 1.1075611439344315,
      "learning_rate": 2.2340425531914894e-06,
      "loss": 1.6226,
      "step": 43
    },
    {
      "epoch": 0.0704,
      "grad_norm": 1.0190229274905553,
      "learning_rate": 2.2872340425531916e-06,
      "loss": 1.5601,
      "step": 44
    },
    {
      "epoch": 0.072,
      "grad_norm": 1.0982246752713258,
      "learning_rate": 2.340425531914894e-06,
      "loss": 1.5292,
      "step": 45
    },
    {
      "epoch": 0.0736,
      "grad_norm": 1.0466468438415701,
      "learning_rate": 2.393617021276596e-06,
      "loss": 1.6309,
      "step": 46
    },
    {
      "epoch": 0.0752,
      "grad_norm": 1.0439745277735268,
      "learning_rate": 2.446808510638298e-06,
      "loss": 1.6077,
      "step": 47
    },
    {
      "epoch": 0.0768,
      "grad_norm": 0.9132651437230528,
      "learning_rate": 2.5e-06,
      "loss": 1.5766,
      "step": 48
    },
    {
      "epoch": 0.0784,
      "grad_norm": 0.99048628130422,
      "learning_rate": 2.553191489361702e-06,
      "loss": 1.6093,
      "step": 49
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9728303885350286,
      "learning_rate": 2.6063829787234047e-06,
      "loss": 1.6118,
      "step": 50
    },
    {
      "epoch": 0.0816,
      "grad_norm": 0.8910266501185391,
      "learning_rate": 2.6595744680851065e-06,
      "loss": 1.555,
      "step": 51
    },
    {
      "epoch": 0.0832,
      "grad_norm": 0.9325860036931476,
      "learning_rate": 2.7127659574468084e-06,
      "loss": 1.5921,
      "step": 52
    },
    {
      "epoch": 0.0848,
      "grad_norm": 0.899023371841317,
      "learning_rate": 2.765957446808511e-06,
      "loss": 1.6308,
      "step": 53
    },
    {
      "epoch": 0.0864,
      "grad_norm": 0.846376631597996,
      "learning_rate": 2.819148936170213e-06,
      "loss": 1.6288,
      "step": 54
    },
    {
      "epoch": 0.088,
      "grad_norm": 0.8680076365925677,
      "learning_rate": 2.8723404255319155e-06,
      "loss": 1.5026,
      "step": 55
    },
    {
      "epoch": 0.0896,
      "grad_norm": 0.9201796814089657,
      "learning_rate": 2.9255319148936174e-06,
      "loss": 1.5631,
      "step": 56
    },
    {
      "epoch": 0.0912,
      "grad_norm": 0.8961856850392521,
      "learning_rate": 2.978723404255319e-06,
      "loss": 1.6145,
      "step": 57
    },
    {
      "epoch": 0.0928,
      "grad_norm": 0.8627737833069469,
      "learning_rate": 3.031914893617022e-06,
      "loss": 1.5932,
      "step": 58
    },
    {
      "epoch": 0.0944,
      "grad_norm": 0.8113728927378456,
      "learning_rate": 3.0851063829787237e-06,
      "loss": 1.5727,
      "step": 59
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.8443843958662036,
      "learning_rate": 3.1382978723404255e-06,
      "loss": 1.5259,
      "step": 60
    },
    {
      "epoch": 0.0976,
      "grad_norm": 0.7705126296717155,
      "learning_rate": 3.191489361702128e-06,
      "loss": 1.4812,
      "step": 61
    },
    {
      "epoch": 0.0992,
      "grad_norm": 0.8816427225958179,
      "learning_rate": 3.24468085106383e-06,
      "loss": 1.5475,
      "step": 62
    },
    {
      "epoch": 0.1008,
      "grad_norm": 0.8772760723476437,
      "learning_rate": 3.297872340425532e-06,
      "loss": 1.5184,
      "step": 63
    },
    {
      "epoch": 0.1024,
      "grad_norm": 0.9079785435210731,
      "learning_rate": 3.3510638297872345e-06,
      "loss": 1.6039,
      "step": 64
    },
    {
      "epoch": 0.104,
      "grad_norm": 0.8440873841999057,
      "learning_rate": 3.4042553191489363e-06,
      "loss": 1.5163,
      "step": 65
    },
    {
      "epoch": 0.1056,
      "grad_norm": 0.8946169352253525,
      "learning_rate": 3.457446808510639e-06,
      "loss": 1.475,
      "step": 66
    },
    {
      "epoch": 0.1072,
      "grad_norm": 0.8463296941110126,
      "learning_rate": 3.510638297872341e-06,
      "loss": 1.5314,
      "step": 67
    },
    {
      "epoch": 0.1088,
      "grad_norm": 0.9049233882588135,
      "learning_rate": 3.5638297872340426e-06,
      "loss": 1.5287,
      "step": 68
    },
    {
      "epoch": 0.1104,
      "grad_norm": 0.8452664416692913,
      "learning_rate": 3.6170212765957453e-06,
      "loss": 1.514,
      "step": 69
    },
    {
      "epoch": 0.112,
      "grad_norm": 0.9337403721130088,
      "learning_rate": 3.670212765957447e-06,
      "loss": 1.5285,
      "step": 70
    },
    {
      "epoch": 0.1136,
      "grad_norm": 0.9107753807260315,
      "learning_rate": 3.723404255319149e-06,
      "loss": 1.4944,
      "step": 71
    },
    {
      "epoch": 0.1152,
      "grad_norm": 0.9130500052598824,
      "learning_rate": 3.7765957446808516e-06,
      "loss": 1.48,
      "step": 72
    },
    {
      "epoch": 0.1168,
      "grad_norm": 0.8900080809779581,
      "learning_rate": 3.8297872340425535e-06,
      "loss": 1.5645,
      "step": 73
    },
    {
      "epoch": 0.1184,
      "grad_norm": 0.9228038000262854,
      "learning_rate": 3.882978723404256e-06,
      "loss": 1.5321,
      "step": 74
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0581876269964883,
      "learning_rate": 3.936170212765958e-06,
      "loss": 1.5378,
      "step": 75
    },
    {
      "epoch": 0.1216,
      "grad_norm": 0.8863473886122691,
      "learning_rate": 3.98936170212766e-06,
      "loss": 1.4955,
      "step": 76
    },
    {
      "epoch": 0.1232,
      "grad_norm": 1.0844541182761351,
      "learning_rate": 4.042553191489362e-06,
      "loss": 1.5423,
      "step": 77
    },
    {
      "epoch": 0.1248,
      "grad_norm": 0.8987422815477396,
      "learning_rate": 4.095744680851064e-06,
      "loss": 1.4643,
      "step": 78
    },
    {
      "epoch": 0.1264,
      "grad_norm": 0.9836127204194123,
      "learning_rate": 4.148936170212766e-06,
      "loss": 1.4839,
      "step": 79
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.9353829768294015,
      "learning_rate": 4.202127659574468e-06,
      "loss": 1.4851,
      "step": 80
    },
    {
      "epoch": 0.1296,
      "grad_norm": 0.8549112487091828,
      "learning_rate": 4.255319148936171e-06,
      "loss": 1.4726,
      "step": 81
    },
    {
      "epoch": 0.1312,
      "grad_norm": 0.9986401478212867,
      "learning_rate": 4.308510638297873e-06,
      "loss": 1.4723,
      "step": 82
    },
    {
      "epoch": 0.1328,
      "grad_norm": 0.7977069671818596,
      "learning_rate": 4.361702127659575e-06,
      "loss": 1.5104,
      "step": 83
    },
    {
      "epoch": 0.1344,
      "grad_norm": 0.9675556491708618,
      "learning_rate": 4.414893617021277e-06,
      "loss": 1.5086,
      "step": 84
    },
    {
      "epoch": 0.136,
      "grad_norm": 0.8918617525119605,
      "learning_rate": 4.468085106382979e-06,
      "loss": 1.5006,
      "step": 85
    },
    {
      "epoch": 0.1376,
      "grad_norm": 1.0268666893679863,
      "learning_rate": 4.521276595744681e-06,
      "loss": 1.5104,
      "step": 86
    },
    {
      "epoch": 0.1392,
      "grad_norm": 0.9732824580913708,
      "learning_rate": 4.574468085106383e-06,
      "loss": 1.5591,
      "step": 87
    },
    {
      "epoch": 0.1408,
      "grad_norm": 0.9579268558272503,
      "learning_rate": 4.6276595744680855e-06,
      "loss": 1.4836,
      "step": 88
    },
    {
      "epoch": 0.1424,
      "grad_norm": 0.9854549528225242,
      "learning_rate": 4.680851063829788e-06,
      "loss": 1.4729,
      "step": 89
    },
    {
      "epoch": 0.144,
      "grad_norm": 0.86656270346725,
      "learning_rate": 4.73404255319149e-06,
      "loss": 1.5018,
      "step": 90
    },
    {
      "epoch": 0.1456,
      "grad_norm": 0.9848615100772067,
      "learning_rate": 4.787234042553192e-06,
      "loss": 1.4919,
      "step": 91
    },
    {
      "epoch": 0.1472,
      "grad_norm": 0.9145758933611112,
      "learning_rate": 4.840425531914894e-06,
      "loss": 1.5288,
      "step": 92
    },
    {
      "epoch": 0.1488,
      "grad_norm": 0.9740071653242416,
      "learning_rate": 4.893617021276596e-06,
      "loss": 1.5187,
      "step": 93
    },
    {
      "epoch": 0.1504,
      "grad_norm": 0.9447909138720109,
      "learning_rate": 4.946808510638298e-06,
      "loss": 1.4909,
      "step": 94
    },
    {
      "epoch": 0.152,
      "grad_norm": 1.0075757672709473,
      "learning_rate": 5e-06,
      "loss": 1.5144,
      "step": 95
    },
    {
      "epoch": 0.1536,
      "grad_norm": 0.8966470913143026,
      "learning_rate": 5.053191489361703e-06,
      "loss": 1.494,
      "step": 96
    },
    {
      "epoch": 0.1552,
      "grad_norm": 0.990610489142169,
      "learning_rate": 5.106382978723404e-06,
      "loss": 1.4866,
      "step": 97
    },
    {
      "epoch": 0.1568,
      "grad_norm": 0.9847809750843722,
      "learning_rate": 5.159574468085107e-06,
      "loss": 1.5555,
      "step": 98
    },
    {
      "epoch": 0.1584,
      "grad_norm": 1.0885385507897434,
      "learning_rate": 5.212765957446809e-06,
      "loss": 1.4878,
      "step": 99
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9794423274786066,
      "learning_rate": 5.265957446808511e-06,
      "loss": 1.5031,
      "step": 100
    },
    {
      "epoch": 0.1616,
      "grad_norm": 0.9921618995172273,
      "learning_rate": 5.319148936170213e-06,
      "loss": 1.5086,
      "step": 101
    },
    {
      "epoch": 0.1632,
      "grad_norm": 1.029656536943759,
      "learning_rate": 5.372340425531915e-06,
      "loss": 1.4299,
      "step": 102
    },
    {
      "epoch": 0.1648,
      "grad_norm": 1.016648889879388,
      "learning_rate": 5.425531914893617e-06,
      "loss": 1.5909,
      "step": 103
    },
    {
      "epoch": 0.1664,
      "grad_norm": 1.1508197819224875,
      "learning_rate": 5.47872340425532e-06,
      "loss": 1.4851,
      "step": 104
    },
    {
      "epoch": 0.168,
      "grad_norm": 0.997970225033964,
      "learning_rate": 5.531914893617022e-06,
      "loss": 1.4993,
      "step": 105
    },
    {
      "epoch": 0.1696,
      "grad_norm": 1.0612664697716985,
      "learning_rate": 5.5851063829787235e-06,
      "loss": 1.4492,
      "step": 106
    },
    {
      "epoch": 0.1712,
      "grad_norm": 1.1575134233546978,
      "learning_rate": 5.638297872340426e-06,
      "loss": 1.457,
      "step": 107
    },
    {
      "epoch": 0.1728,
      "grad_norm": 0.895530989240117,
      "learning_rate": 5.691489361702128e-06,
      "loss": 1.4426,
      "step": 108
    },
    {
      "epoch": 0.1744,
      "grad_norm": 1.2571227573652028,
      "learning_rate": 5.744680851063831e-06,
      "loss": 1.4857,
      "step": 109
    },
    {
      "epoch": 0.176,
      "grad_norm": 1.0312484593643312,
      "learning_rate": 5.7978723404255325e-06,
      "loss": 1.4693,
      "step": 110
    },
    {
      "epoch": 0.1776,
      "grad_norm": 1.2009040650417955,
      "learning_rate": 5.851063829787235e-06,
      "loss": 1.4555,
      "step": 111
    },
    {
      "epoch": 0.1792,
      "grad_norm": 1.0887566354869873,
      "learning_rate": 5.904255319148937e-06,
      "loss": 1.4437,
      "step": 112
    },
    {
      "epoch": 0.1808,
      "grad_norm": 1.1158744304372625,
      "learning_rate": 5.957446808510638e-06,
      "loss": 1.4409,
      "step": 113
    },
    {
      "epoch": 0.1824,
      "grad_norm": 1.1596626401665169,
      "learning_rate": 6.010638297872341e-06,
      "loss": 1.4493,
      "step": 114
    },
    {
      "epoch": 0.184,
      "grad_norm": 1.0514450455816209,
      "learning_rate": 6.063829787234044e-06,
      "loss": 1.4082,
      "step": 115
    },
    {
      "epoch": 0.1856,
      "grad_norm": 1.9321416998181091,
      "learning_rate": 6.117021276595745e-06,
      "loss": 1.4784,
      "step": 116
    },
    {
      "epoch": 0.1872,
      "grad_norm": 1.2043187206885884,
      "learning_rate": 6.170212765957447e-06,
      "loss": 1.4259,
      "step": 117
    },
    {
      "epoch": 0.1888,
      "grad_norm": 1.514027109366682,
      "learning_rate": 6.22340425531915e-06,
      "loss": 1.4995,
      "step": 118
    },
    {
      "epoch": 0.1904,
      "grad_norm": 1.2938264656453617,
      "learning_rate": 6.276595744680851e-06,
      "loss": 1.3756,
      "step": 119
    },
    {
      "epoch": 0.192,
      "grad_norm": 1.779435174077124,
      "learning_rate": 6.329787234042554e-06,
      "loss": 1.4704,
      "step": 120
    },
    {
      "epoch": 0.1936,
      "grad_norm": 1.4925408922451693,
      "learning_rate": 6.382978723404256e-06,
      "loss": 1.4722,
      "step": 121
    },
    {
      "epoch": 0.1952,
      "grad_norm": 1.6965445164828363,
      "learning_rate": 6.436170212765958e-06,
      "loss": 1.5145,
      "step": 122
    },
    {
      "epoch": 0.1968,
      "grad_norm": 1.335605068841665,
      "learning_rate": 6.48936170212766e-06,
      "loss": 1.4632,
      "step": 123
    },
    {
      "epoch": 0.1984,
      "grad_norm": 1.4664716735580978,
      "learning_rate": 6.542553191489362e-06,
      "loss": 1.4528,
      "step": 124
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2481712104508442,
      "learning_rate": 6.595744680851064e-06,
      "loss": 1.4591,
      "step": 125
    },
    {
      "epoch": 0.2016,
      "grad_norm": 1.539645617154428,
      "learning_rate": 6.648936170212767e-06,
      "loss": 1.4422,
      "step": 126
    },
    {
      "epoch": 0.2032,
      "grad_norm": 1.3520744216801055,
      "learning_rate": 6.702127659574469e-06,
      "loss": 1.4732,
      "step": 127
    },
    {
      "epoch": 0.2048,
      "grad_norm": 1.9097870304895597,
      "learning_rate": 6.75531914893617e-06,
      "loss": 1.4489,
      "step": 128
    },
    {
      "epoch": 0.2064,
      "grad_norm": 1.599123881056501,
      "learning_rate": 6.808510638297873e-06,
      "loss": 1.4883,
      "step": 129
    },
    {
      "epoch": 0.208,
      "grad_norm": 1.6645161041968701,
      "learning_rate": 6.861702127659575e-06,
      "loss": 1.4149,
      "step": 130
    },
    {
      "epoch": 0.2096,
      "grad_norm": 1.6836082540020754,
      "learning_rate": 6.914893617021278e-06,
      "loss": 1.4627,
      "step": 131
    },
    {
      "epoch": 0.2112,
      "grad_norm": 1.2094020108667551,
      "learning_rate": 6.968085106382979e-06,
      "loss": 1.4196,
      "step": 132
    },
    {
      "epoch": 0.2128,
      "grad_norm": 1.3098677527941176,
      "learning_rate": 7.021276595744682e-06,
      "loss": 1.4973,
      "step": 133
    },
    {
      "epoch": 0.2144,
      "grad_norm": 1.2501878236202202,
      "learning_rate": 7.074468085106384e-06,
      "loss": 1.4764,
      "step": 134
    },
    {
      "epoch": 0.216,
      "grad_norm": 1.1907368132796006,
      "learning_rate": 7.127659574468085e-06,
      "loss": 1.5067,
      "step": 135
    },
    {
      "epoch": 0.2176,
      "grad_norm": 1.5742605767703846,
      "learning_rate": 7.1808510638297875e-06,
      "loss": 1.3998,
      "step": 136
    },
    {
      "epoch": 0.2192,
      "grad_norm": 1.1730112401404056,
      "learning_rate": 7.234042553191491e-06,
      "loss": 1.4562,
      "step": 137
    },
    {
      "epoch": 0.2208,
      "grad_norm": 1.5492809521000088,
      "learning_rate": 7.287234042553192e-06,
      "loss": 1.3998,
      "step": 138
    },
    {
      "epoch": 0.2224,
      "grad_norm": 1.4032174784354057,
      "learning_rate": 7.340425531914894e-06,
      "loss": 1.4272,
      "step": 139
    },
    {
      "epoch": 0.224,
      "grad_norm": 1.4827279264355246,
      "learning_rate": 7.3936170212765965e-06,
      "loss": 1.4518,
      "step": 140
    },
    {
      "epoch": 0.2256,
      "grad_norm": 1.3445427583918403,
      "learning_rate": 7.446808510638298e-06,
      "loss": 1.4409,
      "step": 141
    },
    {
      "epoch": 0.2272,
      "grad_norm": 1.3393021754377064,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.467,
      "step": 142
    },
    {
      "epoch": 0.2288,
      "grad_norm": 1.2486684868919777,
      "learning_rate": 7.553191489361703e-06,
      "loss": 1.4131,
      "step": 143
    },
    {
      "epoch": 0.2304,
      "grad_norm": 1.324236263394619,
      "learning_rate": 7.606382978723405e-06,
      "loss": 1.4948,
      "step": 144
    },
    {
      "epoch": 0.232,
      "grad_norm": 1.0660998642523025,
      "learning_rate": 7.659574468085107e-06,
      "loss": 1.4552,
      "step": 145
    },
    {
      "epoch": 0.2336,
      "grad_norm": 1.480140397508996,
      "learning_rate": 7.71276595744681e-06,
      "loss": 1.4217,
      "step": 146
    },
    {
      "epoch": 0.2352,
      "grad_norm": 1.19883408640608,
      "learning_rate": 7.765957446808511e-06,
      "loss": 1.3707,
      "step": 147
    },
    {
      "epoch": 0.2368,
      "grad_norm": 1.4993809188848033,
      "learning_rate": 7.819148936170213e-06,
      "loss": 1.4702,
      "step": 148
    },
    {
      "epoch": 0.2384,
      "grad_norm": 1.2922350986196787,
      "learning_rate": 7.872340425531916e-06,
      "loss": 1.4459,
      "step": 149
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1222771177260011,
      "learning_rate": 7.925531914893617e-06,
      "loss": 1.4927,
      "step": 150
    },
    {
      "epoch": 0.2416,
      "grad_norm": 1.0686183587514617,
      "learning_rate": 7.97872340425532e-06,
      "loss": 1.4041,
      "step": 151
    },
    {
      "epoch": 0.2432,
      "grad_norm": 1.2329969811449608,
      "learning_rate": 8.031914893617022e-06,
      "loss": 1.453,
      "step": 152
    },
    {
      "epoch": 0.2448,
      "grad_norm": 0.9909445648558505,
      "learning_rate": 8.085106382978723e-06,
      "loss": 1.3902,
      "step": 153
    },
    {
      "epoch": 0.2464,
      "grad_norm": 1.1203021995145899,
      "learning_rate": 8.138297872340426e-06,
      "loss": 1.4088,
      "step": 154
    },
    {
      "epoch": 0.248,
      "grad_norm": 0.9855573161876993,
      "learning_rate": 8.191489361702128e-06,
      "loss": 1.446,
      "step": 155
    },
    {
      "epoch": 0.2496,
      "grad_norm": 1.9457438747157743,
      "learning_rate": 8.24468085106383e-06,
      "loss": 1.4238,
      "step": 156
    },
    {
      "epoch": 0.2512,
      "grad_norm": 1.1567054361039588,
      "learning_rate": 8.297872340425532e-06,
      "loss": 1.4266,
      "step": 157
    },
    {
      "epoch": 0.2528,
      "grad_norm": 1.1460688939393093,
      "learning_rate": 8.351063829787235e-06,
      "loss": 1.4523,
      "step": 158
    },
    {
      "epoch": 0.2544,
      "grad_norm": 1.1196396276167333,
      "learning_rate": 8.404255319148937e-06,
      "loss": 1.4857,
      "step": 159
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.974697717468726,
      "learning_rate": 8.457446808510638e-06,
      "loss": 1.449,
      "step": 160
    },
    {
      "epoch": 0.2576,
      "grad_norm": 1.2505435558015776,
      "learning_rate": 8.510638297872341e-06,
      "loss": 1.5052,
      "step": 161
    },
    {
      "epoch": 0.2592,
      "grad_norm": 1.2083114851104333,
      "learning_rate": 8.563829787234044e-06,
      "loss": 1.4981,
      "step": 162
    },
    {
      "epoch": 0.2608,
      "grad_norm": 1.072533567559305,
      "learning_rate": 8.617021276595746e-06,
      "loss": 1.4795,
      "step": 163
    },
    {
      "epoch": 0.2624,
      "grad_norm": 1.4102912049504792,
      "learning_rate": 8.670212765957447e-06,
      "loss": 1.4947,
      "step": 164
    },
    {
      "epoch": 0.264,
      "grad_norm": 1.0921390591190427,
      "learning_rate": 8.72340425531915e-06,
      "loss": 1.4762,
      "step": 165
    },
    {
      "epoch": 0.2656,
      "grad_norm": 1.058859216143935,
      "learning_rate": 8.776595744680852e-06,
      "loss": 1.3948,
      "step": 166
    },
    {
      "epoch": 0.2672,
      "grad_norm": 6.207551031374942,
      "learning_rate": 8.829787234042555e-06,
      "loss": 1.4541,
      "step": 167
    },
    {
      "epoch": 0.2688,
      "grad_norm": 1.788593256068761,
      "learning_rate": 8.882978723404256e-06,
      "loss": 1.4277,
      "step": 168
    },
    {
      "epoch": 0.2704,
      "grad_norm": 1.103738599319844,
      "learning_rate": 8.936170212765958e-06,
      "loss": 1.4035,
      "step": 169
    },
    {
      "epoch": 0.272,
      "grad_norm": 1.2662928742542168,
      "learning_rate": 8.98936170212766e-06,
      "loss": 1.4385,
      "step": 170
    },
    {
      "epoch": 0.2736,
      "grad_norm": 1.2729510656111525,
      "learning_rate": 9.042553191489362e-06,
      "loss": 1.3972,
      "step": 171
    },
    {
      "epoch": 0.2752,
      "grad_norm": 1.1545905787813067,
      "learning_rate": 9.095744680851063e-06,
      "loss": 1.4675,
      "step": 172
    },
    {
      "epoch": 0.2768,
      "grad_norm": 1.2642352355577577,
      "learning_rate": 9.148936170212767e-06,
      "loss": 1.4635,
      "step": 173
    },
    {
      "epoch": 0.2784,
      "grad_norm": 1.0786981676042762,
      "learning_rate": 9.20212765957447e-06,
      "loss": 1.5062,
      "step": 174
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.15694838933617,
      "learning_rate": 9.255319148936171e-06,
      "loss": 1.5612,
      "step": 175
    },
    {
      "epoch": 0.2816,
      "grad_norm": 1.0848505852205708,
      "learning_rate": 9.308510638297872e-06,
      "loss": 1.4881,
      "step": 176
    },
    {
      "epoch": 0.2832,
      "grad_norm": 1.1248987355463487,
      "learning_rate": 9.361702127659576e-06,
      "loss": 1.4539,
      "step": 177
    },
    {
      "epoch": 0.2848,
      "grad_norm": 1.1639282438920266,
      "learning_rate": 9.414893617021279e-06,
      "loss": 1.4784,
      "step": 178
    },
    {
      "epoch": 0.2864,
      "grad_norm": 1.0633159276234743,
      "learning_rate": 9.46808510638298e-06,
      "loss": 1.4082,
      "step": 179
    },
    {
      "epoch": 0.288,
      "grad_norm": 1.1548703345183795,
      "learning_rate": 9.521276595744681e-06,
      "loss": 1.4298,
      "step": 180
    },
    {
      "epoch": 0.2896,
      "grad_norm": 1.136949803640767,
      "learning_rate": 9.574468085106385e-06,
      "loss": 1.526,
      "step": 181
    },
    {
      "epoch": 0.2912,
      "grad_norm": 1.2420780934563853,
      "learning_rate": 9.627659574468086e-06,
      "loss": 1.4625,
      "step": 182
    },
    {
      "epoch": 0.2928,
      "grad_norm": 1.0152368990101912,
      "learning_rate": 9.680851063829787e-06,
      "loss": 1.4145,
      "step": 183
    },
    {
      "epoch": 0.2944,
      "grad_norm": 1.2068085364639327,
      "learning_rate": 9.73404255319149e-06,
      "loss": 1.3768,
      "step": 184
    },
    {
      "epoch": 0.296,
      "grad_norm": 1.0093859624798345,
      "learning_rate": 9.787234042553192e-06,
      "loss": 1.4831,
      "step": 185
    },
    {
      "epoch": 0.2976,
      "grad_norm": 1.2996188707880074,
      "learning_rate": 9.840425531914895e-06,
      "loss": 1.4476,
      "step": 186
    },
    {
      "epoch": 0.2992,
      "grad_norm": 1.3009118594518103,
      "learning_rate": 9.893617021276596e-06,
      "loss": 1.5284,
      "step": 187
    },
    {
      "epoch": 0.3008,
      "grad_norm": 1.2886573415401181,
      "learning_rate": 9.946808510638298e-06,
      "loss": 1.5303,
      "step": 188
    },
    {
      "epoch": 0.3024,
      "grad_norm": 1.285661911495839,
      "learning_rate": 1e-05,
      "loss": 1.4178,
      "step": 189
    },
    {
      "epoch": 0.304,
      "grad_norm": 1.0179854563056594,
      "learning_rate": 9.99999133019233e-06,
      "loss": 1.4122,
      "step": 190
    },
    {
      "epoch": 0.3056,
      "grad_norm": 1.1394971141808132,
      "learning_rate": 9.999965320799377e-06,
      "loss": 1.4008,
      "step": 191
    },
    {
      "epoch": 0.3072,
      "grad_norm": 1.075178266812112,
      "learning_rate": 9.999921971911345e-06,
      "loss": 1.4243,
      "step": 192
    },
    {
      "epoch": 0.3088,
      "grad_norm": 1.0339801015960692,
      "learning_rate": 9.999861283678563e-06,
      "loss": 1.4322,
      "step": 193
    },
    {
      "epoch": 0.3104,
      "grad_norm": 1.3241296156337388,
      "learning_rate": 9.999783256311494e-06,
      "loss": 1.4608,
      "step": 194
    },
    {
      "epoch": 0.312,
      "grad_norm": 0.9987436348234228,
      "learning_rate": 9.99968789008073e-06,
      "loss": 1.3802,
      "step": 195
    },
    {
      "epoch": 0.3136,
      "grad_norm": 1.146028217538445,
      "learning_rate": 9.999575185316994e-06,
      "loss": 1.4488,
      "step": 196
    },
    {
      "epoch": 0.3152,
      "grad_norm": 1.016310307324633,
      "learning_rate": 9.999445142411139e-06,
      "loss": 1.4376,
      "step": 197
    },
    {
      "epoch": 0.3168,
      "grad_norm": 1.1404353125905415,
      "learning_rate": 9.99929776181414e-06,
      "loss": 1.5125,
      "step": 198
    },
    {
      "epoch": 0.3184,
      "grad_norm": 1.1422003418899684,
      "learning_rate": 9.999133044037107e-06,
      "loss": 1.4317,
      "step": 199
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0568681936325013,
      "learning_rate": 9.998950989651261e-06,
      "loss": 1.4662,
      "step": 200
    },
    {
      "epoch": 0.3216,
      "grad_norm": 1.3243421906285568,
      "learning_rate": 9.99875159928796e-06,
      "loss": 1.3844,
      "step": 201
    },
    {
      "epoch": 0.3232,
      "grad_norm": 1.2141593433372975,
      "learning_rate": 9.99853487363867e-06,
      "loss": 1.3953,
      "step": 202
    },
    {
      "epoch": 0.3248,
      "grad_norm": 1.148008559577646,
      "learning_rate": 9.998300813454981e-06,
      "loss": 1.5234,
      "step": 203
    },
    {
      "epoch": 0.3264,
      "grad_norm": 1.0674250189273513,
      "learning_rate": 9.998049419548597e-06,
      "loss": 1.3775,
      "step": 204
    },
    {
      "epoch": 0.328,
      "grad_norm": 1.2923261664903005,
      "learning_rate": 9.997780692791329e-06,
      "loss": 1.3961,
      "step": 205
    },
    {
      "epoch": 0.3296,
      "grad_norm": 0.9754502571028354,
      "learning_rate": 9.997494634115101e-06,
      "loss": 1.4033,
      "step": 206
    },
    {
      "epoch": 0.3312,
      "grad_norm": 1.3921048448348645,
      "learning_rate": 9.997191244511947e-06,
      "loss": 1.4438,
      "step": 207
    },
    {
      "epoch": 0.3328,
      "grad_norm": 1.142009378456334,
      "learning_rate": 9.996870525033994e-06,
      "loss": 1.4219,
      "step": 208
    },
    {
      "epoch": 0.3344,
      "grad_norm": 1.4191924663950402,
      "learning_rate": 9.996532476793475e-06,
      "loss": 1.4382,
      "step": 209
    },
    {
      "epoch": 0.336,
      "grad_norm": 1.208879303154965,
      "learning_rate": 9.996177100962714e-06,
      "loss": 1.44,
      "step": 210
    },
    {
      "epoch": 0.3376,
      "grad_norm": 1.2270441847555584,
      "learning_rate": 9.995804398774129e-06,
      "loss": 1.4302,
      "step": 211
    },
    {
      "epoch": 0.3392,
      "grad_norm": 1.1920678670866047,
      "learning_rate": 9.99541437152022e-06,
      "loss": 1.4081,
      "step": 212
    },
    {
      "epoch": 0.3408,
      "grad_norm": 1.21850843892503,
      "learning_rate": 9.995007020553572e-06,
      "loss": 1.4257,
      "step": 213
    },
    {
      "epoch": 0.3424,
      "grad_norm": 1.3081603945854845,
      "learning_rate": 9.994582347286849e-06,
      "loss": 1.5154,
      "step": 214
    },
    {
      "epoch": 0.344,
      "grad_norm": 1.1806364862514427,
      "learning_rate": 9.994140353192782e-06,
      "loss": 1.4859,
      "step": 215
    },
    {
      "epoch": 0.3456,
      "grad_norm": 1.0139149260153242,
      "learning_rate": 9.993681039804176e-06,
      "loss": 1.4443,
      "step": 216
    },
    {
      "epoch": 0.3472,
      "grad_norm": 1.3295975299239842,
      "learning_rate": 9.99320440871389e-06,
      "loss": 1.4379,
      "step": 217
    },
    {
      "epoch": 0.3488,
      "grad_norm": 1.0372533875603938,
      "learning_rate": 9.99271046157485e-06,
      "loss": 1.422,
      "step": 218
    },
    {
      "epoch": 0.3504,
      "grad_norm": 1.4126879727275392,
      "learning_rate": 9.99219920010002e-06,
      "loss": 1.48,
      "step": 219
    },
    {
      "epoch": 0.352,
      "grad_norm": 1.1924505021654157,
      "learning_rate": 9.991670626062422e-06,
      "loss": 1.5117,
      "step": 220
    },
    {
      "epoch": 0.3536,
      "grad_norm": 1.0284058525062192,
      "learning_rate": 9.991124741295106e-06,
      "loss": 1.4417,
      "step": 221
    },
    {
      "epoch": 0.3552,
      "grad_norm": 1.2611181538371046,
      "learning_rate": 9.990561547691159e-06,
      "loss": 1.4509,
      "step": 222
    },
    {
      "epoch": 0.3568,
      "grad_norm": 0.8701797684656124,
      "learning_rate": 9.989981047203693e-06,
      "loss": 1.4417,
      "step": 223
    },
    {
      "epoch": 0.3584,
      "grad_norm": 1.1941855140118884,
      "learning_rate": 9.98938324184584e-06,
      "loss": 1.3961,
      "step": 224
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.3066742899930284,
      "learning_rate": 9.988768133690741e-06,
      "loss": 1.4349,
      "step": 225
    },
    {
      "epoch": 0.3616,
      "grad_norm": 1.3892772630552792,
      "learning_rate": 9.988135724871546e-06,
      "loss": 1.4037,
      "step": 226
    },
    {
      "epoch": 0.3632,
      "grad_norm": 0.9929435747273122,
      "learning_rate": 9.987486017581401e-06,
      "loss": 1.4323,
      "step": 227
    },
    {
      "epoch": 0.3648,
      "grad_norm": 1.0480592002588243,
      "learning_rate": 9.986819014073436e-06,
      "loss": 1.4782,
      "step": 228
    },
    {
      "epoch": 0.3664,
      "grad_norm": 1.2350519067141474,
      "learning_rate": 9.986134716660774e-06,
      "loss": 1.5092,
      "step": 229
    },
    {
      "epoch": 0.368,
      "grad_norm": 1.0257676189374298,
      "learning_rate": 9.9854331277165e-06,
      "loss": 1.4658,
      "step": 230
    },
    {
      "epoch": 0.3696,
      "grad_norm": 0.8575836188724145,
      "learning_rate": 9.984714249673676e-06,
      "loss": 1.4611,
      "step": 231
    },
    {
      "epoch": 0.3712,
      "grad_norm": 1.1771815569069763,
      "learning_rate": 9.98397808502531e-06,
      "loss": 1.4591,
      "step": 232
    },
    {
      "epoch": 0.3728,
      "grad_norm": 1.006056033893931,
      "learning_rate": 9.983224636324369e-06,
      "loss": 1.4444,
      "step": 233
    },
    {
      "epoch": 0.3744,
      "grad_norm": 1.1156091079604689,
      "learning_rate": 9.982453906183754e-06,
      "loss": 1.433,
      "step": 234
    },
    {
      "epoch": 0.376,
      "grad_norm": 1.0727489694157373,
      "learning_rate": 9.981665897276298e-06,
      "loss": 1.4563,
      "step": 235
    },
    {
      "epoch": 0.3776,
      "grad_norm": 1.1394293882303717,
      "learning_rate": 9.980860612334753e-06,
      "loss": 1.4633,
      "step": 236
    },
    {
      "epoch": 0.3792,
      "grad_norm": 0.9988098538551692,
      "learning_rate": 9.980038054151789e-06,
      "loss": 1.4624,
      "step": 237
    },
    {
      "epoch": 0.3808,
      "grad_norm": 0.998366296631071,
      "learning_rate": 9.979198225579968e-06,
      "loss": 1.4352,
      "step": 238
    },
    {
      "epoch": 0.3824,
      "grad_norm": 0.9356538419454717,
      "learning_rate": 9.97834112953176e-06,
      "loss": 1.4624,
      "step": 239
    },
    {
      "epoch": 0.384,
      "grad_norm": 1.0596843640118303,
      "learning_rate": 9.9774667689795e-06,
      "loss": 1.4356,
      "step": 240
    },
    {
      "epoch": 0.3856,
      "grad_norm": 1.0139488248026194,
      "learning_rate": 9.976575146955409e-06,
      "loss": 1.4423,
      "step": 241
    },
    {
      "epoch": 0.3872,
      "grad_norm": 1.0567704384564636,
      "learning_rate": 9.97566626655156e-06,
      "loss": 1.4104,
      "step": 242
    },
    {
      "epoch": 0.3888,
      "grad_norm": 1.0070703190784271,
      "learning_rate": 9.974740130919883e-06,
      "loss": 1.3561,
      "step": 243
    },
    {
      "epoch": 0.3904,
      "grad_norm": 1.0629806544184184,
      "learning_rate": 9.973796743272141e-06,
      "loss": 1.4378,
      "step": 244
    },
    {
      "epoch": 0.392,
      "grad_norm": 0.8663606871725029,
      "learning_rate": 9.972836106879936e-06,
      "loss": 1.4033,
      "step": 245
    },
    {
      "epoch": 0.3936,
      "grad_norm": 1.0506827186713372,
      "learning_rate": 9.971858225074672e-06,
      "loss": 1.4263,
      "step": 246
    },
    {
      "epoch": 0.3952,
      "grad_norm": 0.9751619767449494,
      "learning_rate": 9.970863101247578e-06,
      "loss": 1.3904,
      "step": 247
    },
    {
      "epoch": 0.3968,
      "grad_norm": 1.0671059246073533,
      "learning_rate": 9.96985073884966e-06,
      "loss": 1.5067,
      "step": 248
    },
    {
      "epoch": 0.3984,
      "grad_norm": 0.9117431368763176,
      "learning_rate": 9.968821141391716e-06,
      "loss": 1.4955,
      "step": 249
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0627960437448065,
      "learning_rate": 9.96777431244431e-06,
      "loss": 1.4487,
      "step": 250
    },
    {
      "epoch": 0.4016,
      "grad_norm": 1.1024916153392768,
      "learning_rate": 9.966710255637764e-06,
      "loss": 1.401,
      "step": 251
    },
    {
      "epoch": 0.4032,
      "grad_norm": 1.0183402494825537,
      "learning_rate": 9.965628974662145e-06,
      "loss": 1.4005,
      "step": 252
    },
    {
      "epoch": 0.4048,
      "grad_norm": 1.2160200830595236,
      "learning_rate": 9.964530473267253e-06,
      "loss": 1.4716,
      "step": 253
    },
    {
      "epoch": 0.4064,
      "grad_norm": 0.8580982196474778,
      "learning_rate": 9.963414755262606e-06,
      "loss": 1.4693,
      "step": 254
    },
    {
      "epoch": 0.408,
      "grad_norm": 1.3294537280273169,
      "learning_rate": 9.962281824517427e-06,
      "loss": 1.5233,
      "step": 255
    },
    {
      "epoch": 0.4096,
      "grad_norm": 0.9631823767618344,
      "learning_rate": 9.961131684960635e-06,
      "loss": 1.495,
      "step": 256
    },
    {
      "epoch": 0.4112,
      "grad_norm": 1.0915266400799861,
      "learning_rate": 9.959964340580823e-06,
      "loss": 1.4221,
      "step": 257
    },
    {
      "epoch": 0.4128,
      "grad_norm": 1.2996727337765348,
      "learning_rate": 9.958779795426253e-06,
      "loss": 1.4862,
      "step": 258
    },
    {
      "epoch": 0.4144,
      "grad_norm": 1.0499780609643798,
      "learning_rate": 9.957578053604837e-06,
      "loss": 1.4263,
      "step": 259
    },
    {
      "epoch": 0.416,
      "grad_norm": 1.2101028711408262,
      "learning_rate": 9.956359119284123e-06,
      "loss": 1.4102,
      "step": 260
    },
    {
      "epoch": 0.4176,
      "grad_norm": 0.9407476757118552,
      "learning_rate": 9.955122996691278e-06,
      "loss": 1.4224,
      "step": 261
    },
    {
      "epoch": 0.4192,
      "grad_norm": 1.149970532376527,
      "learning_rate": 9.953869690113085e-06,
      "loss": 1.5073,
      "step": 262
    },
    {
      "epoch": 0.4208,
      "grad_norm": 1.1569381304422945,
      "learning_rate": 9.952599203895912e-06,
      "loss": 1.5272,
      "step": 263
    },
    {
      "epoch": 0.4224,
      "grad_norm": 10.193055402393679,
      "learning_rate": 9.95131154244571e-06,
      "loss": 1.4674,
      "step": 264
    },
    {
      "epoch": 0.424,
      "grad_norm": 1.3332013883365008,
      "learning_rate": 9.950006710227986e-06,
      "loss": 1.4453,
      "step": 265
    },
    {
      "epoch": 0.4256,
      "grad_norm": 1.0611249517553396,
      "learning_rate": 9.9486847117678e-06,
      "loss": 1.4595,
      "step": 266
    },
    {
      "epoch": 0.4272,
      "grad_norm": 1.1230565109915367,
      "learning_rate": 9.947345551649741e-06,
      "loss": 1.4161,
      "step": 267
    },
    {
      "epoch": 0.4288,
      "grad_norm": 0.9912452550233025,
      "learning_rate": 9.945989234517913e-06,
      "loss": 1.4283,
      "step": 268
    },
    {
      "epoch": 0.4304,
      "grad_norm": 1.2650314820306618,
      "learning_rate": 9.94461576507592e-06,
      "loss": 1.4973,
      "step": 269
    },
    {
      "epoch": 0.432,
      "grad_norm": 1.0363528927803416,
      "learning_rate": 9.943225148086846e-06,
      "loss": 1.4474,
      "step": 270
    },
    {
      "epoch": 0.4336,
      "grad_norm": 1.161932569729354,
      "learning_rate": 9.941817388373248e-06,
      "loss": 1.4044,
      "step": 271
    },
    {
      "epoch": 0.4352,
      "grad_norm": 1.192769107680309,
      "learning_rate": 9.940392490817124e-06,
      "loss": 1.4437,
      "step": 272
    },
    {
      "epoch": 0.4368,
      "grad_norm": 1.0180719676628722,
      "learning_rate": 9.938950460359912e-06,
      "loss": 1.4594,
      "step": 273
    },
    {
      "epoch": 0.4384,
      "grad_norm": 1.32336444485259,
      "learning_rate": 9.937491302002462e-06,
      "loss": 1.397,
      "step": 274
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.941758227441272,
      "learning_rate": 9.936015020805022e-06,
      "loss": 1.4154,
      "step": 275
    },
    {
      "epoch": 0.4416,
      "grad_norm": 1.0636754483140736,
      "learning_rate": 9.934521621887223e-06,
      "loss": 1.3815,
      "step": 276
    },
    {
      "epoch": 0.4432,
      "grad_norm": 1.1160913999526447,
      "learning_rate": 9.933011110428058e-06,
      "loss": 1.4231,
      "step": 277
    },
    {
      "epoch": 0.4448,
      "grad_norm": 0.9285406054969918,
      "learning_rate": 9.93148349166586e-06,
      "loss": 1.5062,
      "step": 278
    },
    {
      "epoch": 0.4464,
      "grad_norm": 1.2127035989655095,
      "learning_rate": 9.929938770898299e-06,
      "loss": 1.4377,
      "step": 279
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.9867642164803131,
      "learning_rate": 9.928376953482343e-06,
      "loss": 1.3943,
      "step": 280
    },
    {
      "epoch": 0.4496,
      "grad_norm": 0.960509543356191,
      "learning_rate": 9.926798044834261e-06,
      "loss": 1.4543,
      "step": 281
    },
    {
      "epoch": 0.4512,
      "grad_norm": 1.49967217369327,
      "learning_rate": 9.92520205042958e-06,
      "loss": 1.49,
      "step": 282
    },
    {
      "epoch": 0.4528,
      "grad_norm": 1.0517424297126032,
      "learning_rate": 9.92358897580309e-06,
      "loss": 1.3678,
      "step": 283
    },
    {
      "epoch": 0.4544,
      "grad_norm": 1.6582813824212093,
      "learning_rate": 9.921958826548808e-06,
      "loss": 1.39,
      "step": 284
    },
    {
      "epoch": 0.456,
      "grad_norm": 1.226854818742527,
      "learning_rate": 9.920311608319968e-06,
      "loss": 1.3902,
      "step": 285
    },
    {
      "epoch": 0.4576,
      "grad_norm": 1.8070074310844861,
      "learning_rate": 9.918647326828993e-06,
      "loss": 1.3359,
      "step": 286
    },
    {
      "epoch": 0.4592,
      "grad_norm": 1.4471492901979368,
      "learning_rate": 9.916965987847485e-06,
      "loss": 1.3846,
      "step": 287
    },
    {
      "epoch": 0.4608,
      "grad_norm": 1.7079800011389257,
      "learning_rate": 9.915267597206198e-06,
      "loss": 1.4116,
      "step": 288
    },
    {
      "epoch": 0.4624,
      "grad_norm": 1.5876672995030492,
      "learning_rate": 9.913552160795022e-06,
      "loss": 1.393,
      "step": 289
    },
    {
      "epoch": 0.464,
      "grad_norm": 1.323460068237611,
      "learning_rate": 9.911819684562954e-06,
      "loss": 1.4056,
      "step": 290
    },
    {
      "epoch": 0.4656,
      "grad_norm": 1.2321240245570981,
      "learning_rate": 9.910070174518093e-06,
      "loss": 1.4416,
      "step": 291
    },
    {
      "epoch": 0.4672,
      "grad_norm": 1.4770316649187911,
      "learning_rate": 9.908303636727604e-06,
      "loss": 1.4322,
      "step": 292
    },
    {
      "epoch": 0.4688,
      "grad_norm": 0.9913225229482509,
      "learning_rate": 9.9065200773177e-06,
      "loss": 1.4129,
      "step": 293
    },
    {
      "epoch": 0.4704,
      "grad_norm": 1.3269877123382519,
      "learning_rate": 9.904719502473635e-06,
      "loss": 1.4054,
      "step": 294
    },
    {
      "epoch": 0.472,
      "grad_norm": 1.0001666292519054,
      "learning_rate": 9.902901918439658e-06,
      "loss": 1.3816,
      "step": 295
    },
    {
      "epoch": 0.4736,
      "grad_norm": 1.410444405799514,
      "learning_rate": 9.901067331519013e-06,
      "loss": 1.4742,
      "step": 296
    },
    {
      "epoch": 0.4752,
      "grad_norm": 1.1597018873494362,
      "learning_rate": 9.899215748073906e-06,
      "loss": 1.4791,
      "step": 297
    },
    {
      "epoch": 0.4768,
      "grad_norm": 1.4373547862983356,
      "learning_rate": 9.897347174525487e-06,
      "loss": 1.4981,
      "step": 298
    },
    {
      "epoch": 0.4784,
      "grad_norm": 1.1213336408421022,
      "learning_rate": 9.895461617353823e-06,
      "loss": 1.4345,
      "step": 299
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5107234916425694,
      "learning_rate": 9.893559083097885e-06,
      "loss": 1.4634,
      "step": 300
    },
    {
      "epoch": 0.4816,
      "grad_norm": 1.329065351807543,
      "learning_rate": 9.891639578355511e-06,
      "loss": 1.3736,
      "step": 301
    },
    {
      "epoch": 0.4832,
      "grad_norm": 1.5180620734385422,
      "learning_rate": 9.8897031097834e-06,
      "loss": 1.4055,
      "step": 302
    },
    {
      "epoch": 0.4848,
      "grad_norm": 1.4447464690119618,
      "learning_rate": 9.887749684097072e-06,
      "loss": 1.4004,
      "step": 303
    },
    {
      "epoch": 0.4864,
      "grad_norm": 1.308094195688055,
      "learning_rate": 9.88577930807086e-06,
      "loss": 1.4257,
      "step": 304
    },
    {
      "epoch": 0.488,
      "grad_norm": 1.2698412499915335,
      "learning_rate": 9.883791988537874e-06,
      "loss": 1.4222,
      "step": 305
    },
    {
      "epoch": 0.4896,
      "grad_norm": 1.2388425973832347,
      "learning_rate": 9.881787732389987e-06,
      "loss": 1.4137,
      "step": 306
    },
    {
      "epoch": 0.4912,
      "grad_norm": 1.3168275121468376,
      "learning_rate": 9.879766546577805e-06,
      "loss": 1.3622,
      "step": 307
    },
    {
      "epoch": 0.4928,
      "grad_norm": 1.0388197484360662,
      "learning_rate": 9.877728438110645e-06,
      "loss": 1.4047,
      "step": 308
    },
    {
      "epoch": 0.4944,
      "grad_norm": 1.3978281078136385,
      "learning_rate": 9.87567341405651e-06,
      "loss": 1.4264,
      "step": 309
    },
    {
      "epoch": 0.496,
      "grad_norm": 1.0555958320961698,
      "learning_rate": 9.873601481542065e-06,
      "loss": 1.4832,
      "step": 310
    },
    {
      "epoch": 0.4976,
      "grad_norm": 1.444905102786209,
      "learning_rate": 9.871512647752612e-06,
      "loss": 1.4609,
      "step": 311
    },
    {
      "epoch": 0.4992,
      "grad_norm": 1.0696657856283853,
      "learning_rate": 9.86940691993207e-06,
      "loss": 1.4209,
      "step": 312
    },
    {
      "epoch": 0.5008,
      "grad_norm": 1.2234961583125523,
      "learning_rate": 9.867284305382936e-06,
      "loss": 1.4014,
      "step": 313
    },
    {
      "epoch": 0.5024,
      "grad_norm": 1.1441628399757677,
      "learning_rate": 9.865144811466275e-06,
      "loss": 1.3982,
      "step": 314
    },
    {
      "epoch": 0.504,
      "grad_norm": 1.8549531939666313,
      "learning_rate": 9.86298844560169e-06,
      "loss": 1.4186,
      "step": 315
    },
    {
      "epoch": 0.5056,
      "grad_norm": 1.4095203652666193,
      "learning_rate": 9.860815215267288e-06,
      "loss": 1.4121,
      "step": 316
    },
    {
      "epoch": 0.5072,
      "grad_norm": 1.0224434920456036,
      "learning_rate": 9.858625127999668e-06,
      "loss": 1.4318,
      "step": 317
    },
    {
      "epoch": 0.5088,
      "grad_norm": 1.1216381954433874,
      "learning_rate": 9.856418191393881e-06,
      "loss": 1.361,
      "step": 318
    },
    {
      "epoch": 0.5104,
      "grad_norm": 1.1121124042683275,
      "learning_rate": 9.854194413103418e-06,
      "loss": 1.5041,
      "step": 319
    },
    {
      "epoch": 0.512,
      "grad_norm": 1.0197686517424895,
      "learning_rate": 9.851953800840166e-06,
      "loss": 1.3988,
      "step": 320
    },
    {
      "epoch": 0.5136,
      "grad_norm": 1.061405473485146,
      "learning_rate": 9.849696362374399e-06,
      "loss": 1.357,
      "step": 321
    },
    {
      "epoch": 0.5152,
      "grad_norm": 1.1015967093286565,
      "learning_rate": 9.847422105534739e-06,
      "loss": 1.386,
      "step": 322
    },
    {
      "epoch": 0.5168,
      "grad_norm": 0.9079988448274122,
      "learning_rate": 9.845131038208135e-06,
      "loss": 1.4365,
      "step": 323
    },
    {
      "epoch": 0.5184,
      "grad_norm": 1.0110100906341442,
      "learning_rate": 9.84282316833983e-06,
      "loss": 1.3872,
      "step": 324
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9798445954226414,
      "learning_rate": 9.84049850393334e-06,
      "loss": 1.3491,
      "step": 325
    },
    {
      "epoch": 0.5216,
      "grad_norm": 0.8534218407972894,
      "learning_rate": 9.838157053050423e-06,
      "loss": 1.3956,
      "step": 326
    },
    {
      "epoch": 0.5232,
      "grad_norm": 0.9264848042214087,
      "learning_rate": 9.83579882381105e-06,
      "loss": 1.4443,
      "step": 327
    },
    {
      "epoch": 0.5248,
      "grad_norm": 0.9307460121587705,
      "learning_rate": 9.83342382439338e-06,
      "loss": 1.4221,
      "step": 328
    },
    {
      "epoch": 0.5264,
      "grad_norm": 0.9267345073414902,
      "learning_rate": 9.831032063033726e-06,
      "loss": 1.4014,
      "step": 329
    },
    {
      "epoch": 0.528,
      "grad_norm": 0.9409938018166801,
      "learning_rate": 9.828623548026533e-06,
      "loss": 1.3966,
      "step": 330
    },
    {
      "epoch": 0.5296,
      "grad_norm": 0.8691475684909293,
      "learning_rate": 9.826198287724346e-06,
      "loss": 1.4191,
      "step": 331
    },
    {
      "epoch": 0.5312,
      "grad_norm": 0.8701050497894401,
      "learning_rate": 9.823756290537783e-06,
      "loss": 1.511,
      "step": 332
    },
    {
      "epoch": 0.5328,
      "grad_norm": 0.9197558560938672,
      "learning_rate": 9.821297564935499e-06,
      "loss": 1.5062,
      "step": 333
    },
    {
      "epoch": 0.5344,
      "grad_norm": 0.8201135092588552,
      "learning_rate": 9.81882211944417e-06,
      "loss": 1.3992,
      "step": 334
    },
    {
      "epoch": 0.536,
      "grad_norm": 0.9158364134220053,
      "learning_rate": 9.816329962648444e-06,
      "loss": 1.4359,
      "step": 335
    },
    {
      "epoch": 0.5376,
      "grad_norm": 0.9027208432008131,
      "learning_rate": 9.813821103190932e-06,
      "loss": 1.3923,
      "step": 336
    },
    {
      "epoch": 0.5392,
      "grad_norm": 0.864710987449274,
      "learning_rate": 9.811295549772169e-06,
      "loss": 1.407,
      "step": 337
    },
    {
      "epoch": 0.5408,
      "grad_norm": 0.9531030602034842,
      "learning_rate": 9.808753311150575e-06,
      "loss": 1.4849,
      "step": 338
    },
    {
      "epoch": 0.5424,
      "grad_norm": 0.8975694513758065,
      "learning_rate": 9.80619439614244e-06,
      "loss": 1.3987,
      "step": 339
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.9226361172738301,
      "learning_rate": 9.803618813621885e-06,
      "loss": 1.3523,
      "step": 340
    },
    {
      "epoch": 0.5456,
      "grad_norm": 0.935735106356502,
      "learning_rate": 9.801026572520832e-06,
      "loss": 1.4338,
      "step": 341
    },
    {
      "epoch": 0.5472,
      "grad_norm": 1.1119375081459137,
      "learning_rate": 9.798417681828972e-06,
      "loss": 1.3934,
      "step": 342
    },
    {
      "epoch": 0.5488,
      "grad_norm": 0.8175393803330344,
      "learning_rate": 9.795792150593739e-06,
      "loss": 1.4327,
      "step": 343
    },
    {
      "epoch": 0.5504,
      "grad_norm": 0.8990950616930327,
      "learning_rate": 9.793149987920273e-06,
      "loss": 1.3767,
      "step": 344
    },
    {
      "epoch": 0.552,
      "grad_norm": 0.9793471745241659,
      "learning_rate": 9.79049120297139e-06,
      "loss": 1.4813,
      "step": 345
    },
    {
      "epoch": 0.5536,
      "grad_norm": 0.8562369137771697,
      "learning_rate": 9.787815804967554e-06,
      "loss": 1.3845,
      "step": 346
    },
    {
      "epoch": 0.5552,
      "grad_norm": 0.9139607579169609,
      "learning_rate": 9.785123803186834e-06,
      "loss": 1.4076,
      "step": 347
    },
    {
      "epoch": 0.5568,
      "grad_norm": 0.8711154833351458,
      "learning_rate": 9.782415206964892e-06,
      "loss": 1.4578,
      "step": 348
    },
    {
      "epoch": 0.5584,
      "grad_norm": 0.930104451574292,
      "learning_rate": 9.779690025694926e-06,
      "loss": 1.4171,
      "step": 349
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0020358794657596,
      "learning_rate": 9.776948268827658e-06,
      "loss": 1.4144,
      "step": 350
    },
    {
      "epoch": 0.5616,
      "grad_norm": 0.9372436270627912,
      "learning_rate": 9.77418994587129e-06,
      "loss": 1.4119,
      "step": 351
    },
    {
      "epoch": 0.5632,
      "grad_norm": 0.9862287588516723,
      "learning_rate": 9.771415066391473e-06,
      "loss": 1.4157,
      "step": 352
    },
    {
      "epoch": 0.5648,
      "grad_norm": 0.8360953660537506,
      "learning_rate": 9.768623640011272e-06,
      "loss": 1.3996,
      "step": 353
    },
    {
      "epoch": 0.5664,
      "grad_norm": 0.922015992077748,
      "learning_rate": 9.765815676411145e-06,
      "loss": 1.4021,
      "step": 354
    },
    {
      "epoch": 0.568,
      "grad_norm": 0.9540299899122587,
      "learning_rate": 9.762991185328891e-06,
      "loss": 1.4625,
      "step": 355
    },
    {
      "epoch": 0.5696,
      "grad_norm": 0.8150915076428624,
      "learning_rate": 9.760150176559627e-06,
      "loss": 1.3542,
      "step": 356
    },
    {
      "epoch": 0.5712,
      "grad_norm": 0.9352551557771939,
      "learning_rate": 9.757292659955755e-06,
      "loss": 1.475,
      "step": 357
    },
    {
      "epoch": 0.5728,
      "grad_norm": 0.9814448125512287,
      "learning_rate": 9.754418645426919e-06,
      "loss": 1.3793,
      "step": 358
    },
    {
      "epoch": 0.5744,
      "grad_norm": 0.9205895186526052,
      "learning_rate": 9.751528142939986e-06,
      "loss": 1.4801,
      "step": 359
    },
    {
      "epoch": 0.576,
      "grad_norm": 1.0341133369581048,
      "learning_rate": 9.74862116251899e-06,
      "loss": 1.4494,
      "step": 360
    },
    {
      "epoch": 0.5776,
      "grad_norm": 0.9571493072954973,
      "learning_rate": 9.74569771424512e-06,
      "loss": 1.4297,
      "step": 361
    },
    {
      "epoch": 0.5792,
      "grad_norm": 0.840076326142813,
      "learning_rate": 9.742757808256667e-06,
      "loss": 1.4816,
      "step": 362
    },
    {
      "epoch": 0.5808,
      "grad_norm": 0.9110763443702862,
      "learning_rate": 9.739801454749e-06,
      "loss": 1.4119,
      "step": 363
    },
    {
      "epoch": 0.5824,
      "grad_norm": 0.9371883466688021,
      "learning_rate": 9.736828663974527e-06,
      "loss": 1.4685,
      "step": 364
    },
    {
      "epoch": 0.584,
      "grad_norm": 0.8786398200577361,
      "learning_rate": 9.733839446242655e-06,
      "loss": 1.4566,
      "step": 365
    },
    {
      "epoch": 0.5856,
      "grad_norm": 1.0133986010412885,
      "learning_rate": 9.730833811919763e-06,
      "loss": 1.4685,
      "step": 366
    },
    {
      "epoch": 0.5872,
      "grad_norm": 0.8178498395531137,
      "learning_rate": 9.727811771429158e-06,
      "loss": 1.4209,
      "step": 367
    },
    {
      "epoch": 0.5888,
      "grad_norm": 0.8362489680723357,
      "learning_rate": 9.724773335251046e-06,
      "loss": 1.4451,
      "step": 368
    },
    {
      "epoch": 0.5904,
      "grad_norm": 0.8218543297004491,
      "learning_rate": 9.721718513922488e-06,
      "loss": 1.3717,
      "step": 369
    },
    {
      "epoch": 0.592,
      "grad_norm": 0.8767797758998097,
      "learning_rate": 9.71864731803737e-06,
      "loss": 1.47,
      "step": 370
    },
    {
      "epoch": 0.5936,
      "grad_norm": 0.8276344534323786,
      "learning_rate": 9.715559758246363e-06,
      "loss": 1.4859,
      "step": 371
    },
    {
      "epoch": 0.5952,
      "grad_norm": 0.9489831977644896,
      "learning_rate": 9.712455845256888e-06,
      "loss": 1.448,
      "step": 372
    },
    {
      "epoch": 0.5968,
      "grad_norm": 0.827971597744399,
      "learning_rate": 9.709335589833076e-06,
      "loss": 1.4082,
      "step": 373
    },
    {
      "epoch": 0.5984,
      "grad_norm": 0.9056881000779577,
      "learning_rate": 9.70619900279573e-06,
      "loss": 1.4484,
      "step": 374
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8474485398845036,
      "learning_rate": 9.703046095022297e-06,
      "loss": 1.4284,
      "step": 375
    },
    {
      "epoch": 0.6016,
      "grad_norm": 0.9203585143081829,
      "learning_rate": 9.699876877446815e-06,
      "loss": 1.3837,
      "step": 376
    },
    {
      "epoch": 0.6032,
      "grad_norm": 0.7766434687066069,
      "learning_rate": 9.696691361059886e-06,
      "loss": 1.416,
      "step": 377
    },
    {
      "epoch": 0.6048,
      "grad_norm": 0.9951693227130275,
      "learning_rate": 9.693489556908641e-06,
      "loss": 1.3927,
      "step": 378
    },
    {
      "epoch": 0.6064,
      "grad_norm": 0.8800047285292374,
      "learning_rate": 9.690271476096686e-06,
      "loss": 1.4438,
      "step": 379
    },
    {
      "epoch": 0.608,
      "grad_norm": 0.9225665334573252,
      "learning_rate": 9.68703712978408e-06,
      "loss": 1.3998,
      "step": 380
    },
    {
      "epoch": 0.6096,
      "grad_norm": 0.9755823059307451,
      "learning_rate": 9.683786529187287e-06,
      "loss": 1.4589,
      "step": 381
    },
    {
      "epoch": 0.6112,
      "grad_norm": 0.8628218112668279,
      "learning_rate": 9.680519685579137e-06,
      "loss": 1.4244,
      "step": 382
    },
    {
      "epoch": 0.6128,
      "grad_norm": 0.9258187167646135,
      "learning_rate": 9.677236610288797e-06,
      "loss": 1.4251,
      "step": 383
    },
    {
      "epoch": 0.6144,
      "grad_norm": 0.8735136649619224,
      "learning_rate": 9.673937314701714e-06,
      "loss": 1.4499,
      "step": 384
    },
    {
      "epoch": 0.616,
      "grad_norm": 0.9606267760302185,
      "learning_rate": 9.670621810259596e-06,
      "loss": 1.3737,
      "step": 385
    },
    {
      "epoch": 0.6176,
      "grad_norm": 0.8255293910333185,
      "learning_rate": 9.667290108460354e-06,
      "loss": 1.4089,
      "step": 386
    },
    {
      "epoch": 0.6192,
      "grad_norm": 0.9020908642811367,
      "learning_rate": 9.663942220858075e-06,
      "loss": 1.4128,
      "step": 387
    },
    {
      "epoch": 0.6208,
      "grad_norm": 0.8265705218744396,
      "learning_rate": 9.660578159062977e-06,
      "loss": 1.4139,
      "step": 388
    },
    {
      "epoch": 0.6224,
      "grad_norm": 0.8979546323010333,
      "learning_rate": 9.657197934741366e-06,
      "loss": 1.4717,
      "step": 389
    },
    {
      "epoch": 0.624,
      "grad_norm": 3.767167042160706,
      "learning_rate": 9.6538015596156e-06,
      "loss": 1.3936,
      "step": 390
    },
    {
      "epoch": 0.6256,
      "grad_norm": 1.0310897763864673,
      "learning_rate": 9.650389045464046e-06,
      "loss": 1.425,
      "step": 391
    },
    {
      "epoch": 0.6272,
      "grad_norm": 0.797479000532521,
      "learning_rate": 9.646960404121042e-06,
      "loss": 1.3942,
      "step": 392
    },
    {
      "epoch": 0.6288,
      "grad_norm": 0.9138992787800942,
      "learning_rate": 9.643515647476851e-06,
      "loss": 1.4173,
      "step": 393
    },
    {
      "epoch": 0.6304,
      "grad_norm": 0.8696333313464794,
      "learning_rate": 9.640054787477626e-06,
      "loss": 1.3645,
      "step": 394
    },
    {
      "epoch": 0.632,
      "grad_norm": 0.9178266930714306,
      "learning_rate": 9.63657783612536e-06,
      "loss": 1.3566,
      "step": 395
    },
    {
      "epoch": 0.6336,
      "grad_norm": 0.951919975547467,
      "learning_rate": 9.633084805477857e-06,
      "loss": 1.4247,
      "step": 396
    },
    {
      "epoch": 0.6352,
      "grad_norm": 0.7943477853823203,
      "learning_rate": 9.629575707648675e-06,
      "loss": 1.4639,
      "step": 397
    },
    {
      "epoch": 0.6368,
      "grad_norm": 0.8980820566462163,
      "learning_rate": 9.626050554807096e-06,
      "loss": 1.4354,
      "step": 398
    },
    {
      "epoch": 0.6384,
      "grad_norm": 0.8560588599116303,
      "learning_rate": 9.62250935917808e-06,
      "loss": 1.421,
      "step": 399
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7815638173148018,
      "learning_rate": 9.618952133042223e-06,
      "loss": 1.3994,
      "step": 400
    },
    {
      "epoch": 0.6416,
      "grad_norm": 0.8621011094542361,
      "learning_rate": 9.615378888735706e-06,
      "loss": 1.4511,
      "step": 401
    },
    {
      "epoch": 0.6432,
      "grad_norm": 0.8102132442449165,
      "learning_rate": 9.611789638650269e-06,
      "loss": 1.4319,
      "step": 402
    },
    {
      "epoch": 0.6448,
      "grad_norm": 0.7465725308310809,
      "learning_rate": 9.608184395233156e-06,
      "loss": 1.4256,
      "step": 403
    },
    {
      "epoch": 0.6464,
      "grad_norm": 0.7897051421741036,
      "learning_rate": 9.604563170987072e-06,
      "loss": 1.3629,
      "step": 404
    },
    {
      "epoch": 0.648,
      "grad_norm": 0.859369833080061,
      "learning_rate": 9.600925978470143e-06,
      "loss": 1.4322,
      "step": 405
    },
    {
      "epoch": 0.6496,
      "grad_norm": 0.7545946441264648,
      "learning_rate": 9.597272830295877e-06,
      "loss": 1.3984,
      "step": 406
    },
    {
      "epoch": 0.6512,
      "grad_norm": 0.9394041179615708,
      "learning_rate": 9.593603739133105e-06,
      "loss": 1.4063,
      "step": 407
    },
    {
      "epoch": 0.6528,
      "grad_norm": 0.802451285552132,
      "learning_rate": 9.589918717705957e-06,
      "loss": 1.3471,
      "step": 408
    },
    {
      "epoch": 0.6544,
      "grad_norm": 0.9248469483326801,
      "learning_rate": 9.586217778793804e-06,
      "loss": 1.4236,
      "step": 409
    },
    {
      "epoch": 0.656,
      "grad_norm": 0.7963208852636423,
      "learning_rate": 9.582500935231215e-06,
      "loss": 1.3847,
      "step": 410
    },
    {
      "epoch": 0.6576,
      "grad_norm": 0.8361212228294335,
      "learning_rate": 9.57876819990792e-06,
      "loss": 1.3799,
      "step": 411
    },
    {
      "epoch": 0.6592,
      "grad_norm": 0.9240936529862012,
      "learning_rate": 9.575019585768758e-06,
      "loss": 1.5016,
      "step": 412
    },
    {
      "epoch": 0.6608,
      "grad_norm": 0.7457519759513939,
      "learning_rate": 9.571255105813632e-06,
      "loss": 1.4432,
      "step": 413
    },
    {
      "epoch": 0.6624,
      "grad_norm": 0.9906229639819034,
      "learning_rate": 9.567474773097469e-06,
      "loss": 1.4332,
      "step": 414
    },
    {
      "epoch": 0.664,
      "grad_norm": 0.7471233867358443,
      "learning_rate": 9.563678600730175e-06,
      "loss": 1.3954,
      "step": 415
    },
    {
      "epoch": 0.6656,
      "grad_norm": 0.9970219240466511,
      "learning_rate": 9.559866601876581e-06,
      "loss": 1.4023,
      "step": 416
    },
    {
      "epoch": 0.6672,
      "grad_norm": 0.8428032735780085,
      "learning_rate": 9.556038789756407e-06,
      "loss": 1.4451,
      "step": 417
    },
    {
      "epoch": 0.6688,
      "grad_norm": 0.8202976136620377,
      "learning_rate": 9.55219517764421e-06,
      "loss": 1.4603,
      "step": 418
    },
    {
      "epoch": 0.6704,
      "grad_norm": 0.7734368738380671,
      "learning_rate": 9.548335778869342e-06,
      "loss": 1.3851,
      "step": 419
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.777475021545702,
      "learning_rate": 9.544460606815901e-06,
      "loss": 1.4173,
      "step": 420
    },
    {
      "epoch": 0.6736,
      "grad_norm": 0.839841562782382,
      "learning_rate": 9.540569674922685e-06,
      "loss": 1.4016,
      "step": 421
    },
    {
      "epoch": 0.6752,
      "grad_norm": 0.7166071073166973,
      "learning_rate": 9.536662996683146e-06,
      "loss": 1.3878,
      "step": 422
    },
    {
      "epoch": 0.6768,
      "grad_norm": 1.7290951609403928,
      "learning_rate": 9.532740585645346e-06,
      "loss": 1.3992,
      "step": 423
    },
    {
      "epoch": 0.6784,
      "grad_norm": 0.7637246776466564,
      "learning_rate": 9.528802455411902e-06,
      "loss": 1.3978,
      "step": 424
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.815967134174933,
      "learning_rate": 9.52484861963995e-06,
      "loss": 1.426,
      "step": 425
    },
    {
      "epoch": 0.6816,
      "grad_norm": 0.7653835655880084,
      "learning_rate": 9.520879092041085e-06,
      "loss": 1.3691,
      "step": 426
    },
    {
      "epoch": 0.6832,
      "grad_norm": 0.833860013742291,
      "learning_rate": 9.516893886381324e-06,
      "loss": 1.4308,
      "step": 427
    },
    {
      "epoch": 0.6848,
      "grad_norm": 0.7975825684611244,
      "learning_rate": 9.512893016481053e-06,
      "loss": 1.3581,
      "step": 428
    },
    {
      "epoch": 0.6864,
      "grad_norm": 0.7852277850368625,
      "learning_rate": 9.508876496214983e-06,
      "loss": 1.3345,
      "step": 429
    },
    {
      "epoch": 0.688,
      "grad_norm": 0.9116755364286001,
      "learning_rate": 9.504844339512096e-06,
      "loss": 1.4623,
      "step": 430
    },
    {
      "epoch": 0.6896,
      "grad_norm": 0.7492761632107944,
      "learning_rate": 9.500796560355603e-06,
      "loss": 1.3587,
      "step": 431
    },
    {
      "epoch": 0.6912,
      "grad_norm": 0.8668228268574004,
      "learning_rate": 9.496733172782889e-06,
      "loss": 1.4516,
      "step": 432
    },
    {
      "epoch": 0.6928,
      "grad_norm": 0.8194548677105827,
      "learning_rate": 9.492654190885469e-06,
      "loss": 1.3867,
      "step": 433
    },
    {
      "epoch": 0.6944,
      "grad_norm": 0.8038784048261554,
      "learning_rate": 9.488559628808939e-06,
      "loss": 1.443,
      "step": 434
    },
    {
      "epoch": 0.696,
      "grad_norm": 0.8476102326635411,
      "learning_rate": 9.484449500752927e-06,
      "loss": 1.3529,
      "step": 435
    },
    {
      "epoch": 0.6976,
      "grad_norm": 0.8168595141045176,
      "learning_rate": 9.480323820971039e-06,
      "loss": 1.4434,
      "step": 436
    },
    {
      "epoch": 0.6992,
      "grad_norm": 0.8435253639326066,
      "learning_rate": 9.476182603770814e-06,
      "loss": 1.4215,
      "step": 437
    },
    {
      "epoch": 0.7008,
      "grad_norm": 0.7737763972536591,
      "learning_rate": 9.472025863513676e-06,
      "loss": 1.442,
      "step": 438
    },
    {
      "epoch": 0.7024,
      "grad_norm": 0.8428085541751823,
      "learning_rate": 9.467853614614883e-06,
      "loss": 1.4239,
      "step": 439
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.8869857916367294,
      "learning_rate": 9.46366587154347e-06,
      "loss": 1.3911,
      "step": 440
    },
    {
      "epoch": 0.7056,
      "grad_norm": 0.8758929688292688,
      "learning_rate": 9.459462648822209e-06,
      "loss": 1.4102,
      "step": 441
    },
    {
      "epoch": 0.7072,
      "grad_norm": 0.8165211116030516,
      "learning_rate": 9.45524396102755e-06,
      "loss": 1.3547,
      "step": 442
    },
    {
      "epoch": 0.7088,
      "grad_norm": 0.82840366117552,
      "learning_rate": 9.451009822789583e-06,
      "loss": 1.4768,
      "step": 443
    },
    {
      "epoch": 0.7104,
      "grad_norm": 2.785844451722044,
      "learning_rate": 9.44676024879197e-06,
      "loss": 1.436,
      "step": 444
    },
    {
      "epoch": 0.712,
      "grad_norm": 1.148653952394221,
      "learning_rate": 9.442495253771909e-06,
      "loss": 1.444,
      "step": 445
    },
    {
      "epoch": 0.7136,
      "grad_norm": 0.8156755983238848,
      "learning_rate": 9.438214852520073e-06,
      "loss": 1.4134,
      "step": 446
    },
    {
      "epoch": 0.7152,
      "grad_norm": 0.9307850006838604,
      "learning_rate": 9.433919059880564e-06,
      "loss": 1.438,
      "step": 447
    },
    {
      "epoch": 0.7168,
      "grad_norm": 0.8320833096214428,
      "learning_rate": 9.429607890750863e-06,
      "loss": 1.4315,
      "step": 448
    },
    {
      "epoch": 0.7184,
      "grad_norm": 0.8895851045566666,
      "learning_rate": 9.425281360081769e-06,
      "loss": 1.4151,
      "step": 449
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.91081971997764,
      "learning_rate": 9.420939482877359e-06,
      "loss": 1.3142,
      "step": 450
    },
    {
      "epoch": 0.7216,
      "grad_norm": 0.8538194731862825,
      "learning_rate": 9.416582274194929e-06,
      "loss": 1.3536,
      "step": 451
    },
    {
      "epoch": 0.7232,
      "grad_norm": 0.9409730889981983,
      "learning_rate": 9.412209749144947e-06,
      "loss": 1.3926,
      "step": 452
    },
    {
      "epoch": 0.7248,
      "grad_norm": 1.0099838889739883,
      "learning_rate": 9.40782192289099e-06,
      "loss": 1.3961,
      "step": 453
    },
    {
      "epoch": 0.7264,
      "grad_norm": 1.3087634457424013,
      "learning_rate": 9.4034188106497e-06,
      "loss": 1.3664,
      "step": 454
    },
    {
      "epoch": 0.728,
      "grad_norm": 1.03862182904616,
      "learning_rate": 9.399000427690736e-06,
      "loss": 1.4342,
      "step": 455
    },
    {
      "epoch": 0.7296,
      "grad_norm": 0.9222921830646378,
      "learning_rate": 9.394566789336707e-06,
      "loss": 1.4523,
      "step": 456
    },
    {
      "epoch": 0.7312,
      "grad_norm": 0.8895463733697926,
      "learning_rate": 9.390117910963132e-06,
      "loss": 1.4349,
      "step": 457
    },
    {
      "epoch": 0.7328,
      "grad_norm": 1.022324993039078,
      "learning_rate": 9.385653807998376e-06,
      "loss": 1.4487,
      "step": 458
    },
    {
      "epoch": 0.7344,
      "grad_norm": 0.7540574375962111,
      "learning_rate": 9.381174495923608e-06,
      "loss": 1.3358,
      "step": 459
    },
    {
      "epoch": 0.736,
      "grad_norm": 1.1535140906136885,
      "learning_rate": 9.376679990272736e-06,
      "loss": 1.393,
      "step": 460
    },
    {
      "epoch": 0.7376,
      "grad_norm": 1.0634024765310957,
      "learning_rate": 9.37217030663236e-06,
      "loss": 1.3971,
      "step": 461
    },
    {
      "epoch": 0.7392,
      "grad_norm": 0.8219108754271596,
      "learning_rate": 9.367645460641716e-06,
      "loss": 1.3677,
      "step": 462
    },
    {
      "epoch": 0.7408,
      "grad_norm": 1.0214330143217178,
      "learning_rate": 9.36310546799262e-06,
      "loss": 1.4194,
      "step": 463
    },
    {
      "epoch": 0.7424,
      "grad_norm": 0.8790276684005608,
      "learning_rate": 9.358550344429421e-06,
      "loss": 1.3891,
      "step": 464
    },
    {
      "epoch": 0.744,
      "grad_norm": 1.0645231837510494,
      "learning_rate": 9.353980105748934e-06,
      "loss": 1.4413,
      "step": 465
    },
    {
      "epoch": 0.7456,
      "grad_norm": 1.0625998109266122,
      "learning_rate": 9.349394767800397e-06,
      "loss": 1.5055,
      "step": 466
    },
    {
      "epoch": 0.7472,
      "grad_norm": 0.7941737733747604,
      "learning_rate": 9.344794346485408e-06,
      "loss": 1.3338,
      "step": 467
    },
    {
      "epoch": 0.7488,
      "grad_norm": 0.9727488310586699,
      "learning_rate": 9.340178857757876e-06,
      "loss": 1.3746,
      "step": 468
    },
    {
      "epoch": 0.7504,
      "grad_norm": 0.881861844678689,
      "learning_rate": 9.335548317623957e-06,
      "loss": 1.3904,
      "step": 469
    },
    {
      "epoch": 0.752,
      "grad_norm": 0.8461114332314847,
      "learning_rate": 9.330902742142013e-06,
      "loss": 1.3818,
      "step": 470
    },
    {
      "epoch": 0.7536,
      "grad_norm": 1.1480147425170497,
      "learning_rate": 9.326242147422538e-06,
      "loss": 1.4491,
      "step": 471
    },
    {
      "epoch": 0.7552,
      "grad_norm": 0.872100329016772,
      "learning_rate": 9.321566549628118e-06,
      "loss": 1.3705,
      "step": 472
    },
    {
      "epoch": 0.7568,
      "grad_norm": 1.21908638464621,
      "learning_rate": 9.316875964973366e-06,
      "loss": 1.4483,
      "step": 473
    },
    {
      "epoch": 0.7584,
      "grad_norm": 0.8471504811579136,
      "learning_rate": 9.31217040972487e-06,
      "loss": 1.3531,
      "step": 474
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.088231702594767,
      "learning_rate": 9.307449900201132e-06,
      "loss": 1.4511,
      "step": 475
    },
    {
      "epoch": 0.7616,
      "grad_norm": 0.8621246038086033,
      "learning_rate": 9.302714452772515e-06,
      "loss": 1.3659,
      "step": 476
    },
    {
      "epoch": 0.7632,
      "grad_norm": 1.134841965102251,
      "learning_rate": 9.29796408386119e-06,
      "loss": 1.3717,
      "step": 477
    },
    {
      "epoch": 0.7648,
      "grad_norm": 0.8852017382177069,
      "learning_rate": 9.293198809941067e-06,
      "loss": 1.3804,
      "step": 478
    },
    {
      "epoch": 0.7664,
      "grad_norm": 0.9560019397337868,
      "learning_rate": 9.288418647537752e-06,
      "loss": 1.4028,
      "step": 479
    },
    {
      "epoch": 0.768,
      "grad_norm": 0.9803701265225006,
      "learning_rate": 9.283623613228479e-06,
      "loss": 1.4659,
      "step": 480
    },
    {
      "epoch": 0.7696,
      "grad_norm": 0.836190379173003,
      "learning_rate": 9.27881372364206e-06,
      "loss": 1.4015,
      "step": 481
    },
    {
      "epoch": 0.7712,
      "grad_norm": 0.8701813725124924,
      "learning_rate": 9.27398899545882e-06,
      "loss": 1.3947,
      "step": 482
    },
    {
      "epoch": 0.7728,
      "grad_norm": 0.8238499211066291,
      "learning_rate": 9.269149445410545e-06,
      "loss": 1.4123,
      "step": 483
    },
    {
      "epoch": 0.7744,
      "grad_norm": 0.9096259568409332,
      "learning_rate": 9.264295090280424e-06,
      "loss": 1.4398,
      "step": 484
    },
    {
      "epoch": 0.776,
      "grad_norm": 0.9386694426741384,
      "learning_rate": 9.259425946902987e-06,
      "loss": 1.4226,
      "step": 485
    },
    {
      "epoch": 0.7776,
      "grad_norm": 0.799185421027831,
      "learning_rate": 9.254542032164047e-06,
      "loss": 1.3927,
      "step": 486
    },
    {
      "epoch": 0.7792,
      "grad_norm": 1.030084562615177,
      "learning_rate": 9.249643363000645e-06,
      "loss": 1.4312,
      "step": 487
    },
    {
      "epoch": 0.7808,
      "grad_norm": 0.773745472971282,
      "learning_rate": 9.24472995640099e-06,
      "loss": 1.3902,
      "step": 488
    },
    {
      "epoch": 0.7824,
      "grad_norm": 1.0824211144994762,
      "learning_rate": 9.239801829404396e-06,
      "loss": 1.4318,
      "step": 489
    },
    {
      "epoch": 0.784,
      "grad_norm": 0.848045121540796,
      "learning_rate": 9.234858999101232e-06,
      "loss": 1.469,
      "step": 490
    },
    {
      "epoch": 0.7856,
      "grad_norm": 0.8076224826808945,
      "learning_rate": 9.22990148263285e-06,
      "loss": 1.4531,
      "step": 491
    },
    {
      "epoch": 0.7872,
      "grad_norm": 0.9564041569658456,
      "learning_rate": 9.224929297191536e-06,
      "loss": 1.3661,
      "step": 492
    },
    {
      "epoch": 0.7888,
      "grad_norm": 0.8200524478586152,
      "learning_rate": 9.219942460020447e-06,
      "loss": 1.3941,
      "step": 493
    },
    {
      "epoch": 0.7904,
      "grad_norm": 0.8362904292925779,
      "learning_rate": 9.214940988413552e-06,
      "loss": 1.3714,
      "step": 494
    },
    {
      "epoch": 0.792,
      "grad_norm": 0.7738077164020026,
      "learning_rate": 9.20992489971557e-06,
      "loss": 1.3797,
      "step": 495
    },
    {
      "epoch": 0.7936,
      "grad_norm": 0.8798564129944805,
      "learning_rate": 9.204894211321906e-06,
      "loss": 1.4132,
      "step": 496
    },
    {
      "epoch": 0.7952,
      "grad_norm": 0.7617262157123474,
      "learning_rate": 9.199848940678607e-06,
      "loss": 1.3599,
      "step": 497
    },
    {
      "epoch": 0.7968,
      "grad_norm": 0.9110266237373182,
      "learning_rate": 9.194789105282277e-06,
      "loss": 1.3675,
      "step": 498
    },
    {
      "epoch": 0.7984,
      "grad_norm": 0.873501834372565,
      "learning_rate": 9.189714722680041e-06,
      "loss": 1.4164,
      "step": 499
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0729896420118386,
      "learning_rate": 9.184625810469468e-06,
      "loss": 1.3555,
      "step": 500
    },
    {
      "epoch": 0.8016,
      "grad_norm": 0.8280469602412659,
      "learning_rate": 9.179522386298508e-06,
      "loss": 1.4239,
      "step": 501
    },
    {
      "epoch": 0.8032,
      "grad_norm": 0.7958203356147793,
      "learning_rate": 9.174404467865447e-06,
      "loss": 1.4214,
      "step": 502
    },
    {
      "epoch": 0.8048,
      "grad_norm": 0.7263890100383209,
      "learning_rate": 9.169272072918834e-06,
      "loss": 1.4172,
      "step": 503
    },
    {
      "epoch": 0.8064,
      "grad_norm": 0.7757808511811949,
      "learning_rate": 9.164125219257419e-06,
      "loss": 1.4474,
      "step": 504
    },
    {
      "epoch": 0.808,
      "grad_norm": 0.7616747734526494,
      "learning_rate": 9.158963924730092e-06,
      "loss": 1.4056,
      "step": 505
    },
    {
      "epoch": 0.8096,
      "grad_norm": 0.817570547891624,
      "learning_rate": 9.153788207235827e-06,
      "loss": 1.3926,
      "step": 506
    },
    {
      "epoch": 0.8112,
      "grad_norm": 0.7774621269976391,
      "learning_rate": 9.148598084723615e-06,
      "loss": 1.3808,
      "step": 507
    },
    {
      "epoch": 0.8128,
      "grad_norm": 0.7777628567385435,
      "learning_rate": 9.143393575192402e-06,
      "loss": 1.3737,
      "step": 508
    },
    {
      "epoch": 0.8144,
      "grad_norm": 0.8526727740381155,
      "learning_rate": 9.138174696691025e-06,
      "loss": 1.3927,
      "step": 509
    },
    {
      "epoch": 0.816,
      "grad_norm": 0.8378834062713753,
      "learning_rate": 9.132941467318152e-06,
      "loss": 1.3762,
      "step": 510
    },
    {
      "epoch": 0.8176,
      "grad_norm": 0.81568551494805,
      "learning_rate": 9.127693905222223e-06,
      "loss": 1.3633,
      "step": 511
    },
    {
      "epoch": 0.8192,
      "grad_norm": 0.7831691799298403,
      "learning_rate": 9.122432028601377e-06,
      "loss": 1.392,
      "step": 512
    },
    {
      "epoch": 0.8208,
      "grad_norm": 0.7808594330844733,
      "learning_rate": 9.1171558557034e-06,
      "loss": 1.4532,
      "step": 513
    },
    {
      "epoch": 0.8224,
      "grad_norm": 0.7944410910028369,
      "learning_rate": 9.111865404825652e-06,
      "loss": 1.451,
      "step": 514
    },
    {
      "epoch": 0.824,
      "grad_norm": 0.7837640223395103,
      "learning_rate": 9.10656069431501e-06,
      "loss": 1.3751,
      "step": 515
    },
    {
      "epoch": 0.8256,
      "grad_norm": 0.8338984635064485,
      "learning_rate": 9.101241742567802e-06,
      "loss": 1.4198,
      "step": 516
    },
    {
      "epoch": 0.8272,
      "grad_norm": 0.755315881074111,
      "learning_rate": 9.095908568029741e-06,
      "loss": 1.4329,
      "step": 517
    },
    {
      "epoch": 0.8288,
      "grad_norm": 0.7807124251393178,
      "learning_rate": 9.09056118919587e-06,
      "loss": 1.4454,
      "step": 518
    },
    {
      "epoch": 0.8304,
      "grad_norm": 0.7899381852736618,
      "learning_rate": 9.085199624610486e-06,
      "loss": 1.3578,
      "step": 519
    },
    {
      "epoch": 0.832,
      "grad_norm": 0.7937230419341623,
      "learning_rate": 9.079823892867083e-06,
      "loss": 1.4483,
      "step": 520
    },
    {
      "epoch": 0.8336,
      "grad_norm": 0.7413560093532062,
      "learning_rate": 9.074434012608282e-06,
      "loss": 1.4303,
      "step": 521
    },
    {
      "epoch": 0.8352,
      "grad_norm": 0.8086252684641532,
      "learning_rate": 9.069030002525777e-06,
      "loss": 1.3174,
      "step": 522
    },
    {
      "epoch": 0.8368,
      "grad_norm": 0.7282390345647293,
      "learning_rate": 9.063611881360258e-06,
      "loss": 1.3577,
      "step": 523
    },
    {
      "epoch": 0.8384,
      "grad_norm": 0.762724117756095,
      "learning_rate": 9.05817966790135e-06,
      "loss": 1.3625,
      "step": 524
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.782272367994204,
      "learning_rate": 9.052733380987555e-06,
      "loss": 1.4952,
      "step": 525
    },
    {
      "epoch": 0.8416,
      "grad_norm": 0.7389684817307696,
      "learning_rate": 9.047273039506174e-06,
      "loss": 1.373,
      "step": 526
    },
    {
      "epoch": 0.8432,
      "grad_norm": 0.7573217326891243,
      "learning_rate": 9.041798662393255e-06,
      "loss": 1.4163,
      "step": 527
    },
    {
      "epoch": 0.8448,
      "grad_norm": 0.7735338710486039,
      "learning_rate": 9.036310268633515e-06,
      "loss": 1.4299,
      "step": 528
    },
    {
      "epoch": 0.8464,
      "grad_norm": 0.8213782602576338,
      "learning_rate": 9.030807877260278e-06,
      "loss": 1.4018,
      "step": 529
    },
    {
      "epoch": 0.848,
      "grad_norm": 0.7175008158731276,
      "learning_rate": 9.025291507355419e-06,
      "loss": 1.277,
      "step": 530
    },
    {
      "epoch": 0.8496,
      "grad_norm": 0.8226147304525623,
      "learning_rate": 9.01976117804928e-06,
      "loss": 1.3544,
      "step": 531
    },
    {
      "epoch": 0.8512,
      "grad_norm": 0.7257494307598222,
      "learning_rate": 9.014216908520619e-06,
      "loss": 1.3934,
      "step": 532
    },
    {
      "epoch": 0.8528,
      "grad_norm": 0.8664256202651811,
      "learning_rate": 9.008658717996538e-06,
      "loss": 1.427,
      "step": 533
    },
    {
      "epoch": 0.8544,
      "grad_norm": 0.6925640048984086,
      "learning_rate": 9.003086625752414e-06,
      "loss": 1.5218,
      "step": 534
    },
    {
      "epoch": 0.856,
      "grad_norm": 0.9936339229024544,
      "learning_rate": 8.997500651111833e-06,
      "loss": 1.4674,
      "step": 535
    },
    {
      "epoch": 0.8576,
      "grad_norm": 1.1216601844872118,
      "learning_rate": 8.991900813446523e-06,
      "loss": 1.3774,
      "step": 536
    },
    {
      "epoch": 0.8592,
      "grad_norm": 0.9552909967309356,
      "learning_rate": 8.986287132176295e-06,
      "loss": 1.4137,
      "step": 537
    },
    {
      "epoch": 0.8608,
      "grad_norm": 1.0924722500846789,
      "learning_rate": 8.980659626768961e-06,
      "loss": 1.4085,
      "step": 538
    },
    {
      "epoch": 0.8624,
      "grad_norm": 0.7853072872408081,
      "learning_rate": 8.975018316740278e-06,
      "loss": 1.3882,
      "step": 539
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.9648404349533861,
      "learning_rate": 8.969363221653875e-06,
      "loss": 1.469,
      "step": 540
    },
    {
      "epoch": 0.8656,
      "grad_norm": 0.8690872786753435,
      "learning_rate": 8.963694361121186e-06,
      "loss": 1.4023,
      "step": 541
    },
    {
      "epoch": 0.8672,
      "grad_norm": 0.7458017159503525,
      "learning_rate": 8.958011754801383e-06,
      "loss": 1.3892,
      "step": 542
    },
    {
      "epoch": 0.8688,
      "grad_norm": 0.8366566897981844,
      "learning_rate": 8.952315422401307e-06,
      "loss": 1.4334,
      "step": 543
    },
    {
      "epoch": 0.8704,
      "grad_norm": 0.75889985108478,
      "learning_rate": 8.946605383675403e-06,
      "loss": 1.389,
      "step": 544
    },
    {
      "epoch": 0.872,
      "grad_norm": 0.8061409131251733,
      "learning_rate": 8.940881658425645e-06,
      "loss": 1.3803,
      "step": 545
    },
    {
      "epoch": 0.8736,
      "grad_norm": 2.401950282048642,
      "learning_rate": 8.93514426650147e-06,
      "loss": 1.4424,
      "step": 546
    },
    {
      "epoch": 0.8752,
      "grad_norm": 0.9135985193419704,
      "learning_rate": 8.929393227799715e-06,
      "loss": 1.4219,
      "step": 547
    },
    {
      "epoch": 0.8768,
      "grad_norm": 0.8117471935975381,
      "learning_rate": 8.923628562264536e-06,
      "loss": 1.4005,
      "step": 548
    },
    {
      "epoch": 0.8784,
      "grad_norm": 0.8710901397013183,
      "learning_rate": 8.917850289887353e-06,
      "loss": 1.3923,
      "step": 549
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9740503145564567,
      "learning_rate": 8.91205843070677e-06,
      "loss": 1.4361,
      "step": 550
    },
    {
      "epoch": 0.8816,
      "grad_norm": 0.8889264237869361,
      "learning_rate": 8.906253004808506e-06,
      "loss": 1.4404,
      "step": 551
    },
    {
      "epoch": 0.8832,
      "grad_norm": 0.8649292267149401,
      "learning_rate": 8.900434032325332e-06,
      "loss": 1.3326,
      "step": 552
    },
    {
      "epoch": 0.8848,
      "grad_norm": 0.9536076292396073,
      "learning_rate": 8.894601533437e-06,
      "loss": 1.4521,
      "step": 553
    },
    {
      "epoch": 0.8864,
      "grad_norm": 0.7683158513670622,
      "learning_rate": 8.888755528370163e-06,
      "loss": 1.3244,
      "step": 554
    },
    {
      "epoch": 0.888,
      "grad_norm": 0.9321337995073972,
      "learning_rate": 8.882896037398322e-06,
      "loss": 1.3416,
      "step": 555
    },
    {
      "epoch": 0.8896,
      "grad_norm": 0.8411046907771801,
      "learning_rate": 8.877023080841739e-06,
      "loss": 1.486,
      "step": 556
    },
    {
      "epoch": 0.8912,
      "grad_norm": 0.9244277728786393,
      "learning_rate": 8.871136679067372e-06,
      "loss": 1.4471,
      "step": 557
    },
    {
      "epoch": 0.8928,
      "grad_norm": 0.8747124716429374,
      "learning_rate": 8.865236852488813e-06,
      "loss": 1.433,
      "step": 558
    },
    {
      "epoch": 0.8944,
      "grad_norm": 0.8738980111689701,
      "learning_rate": 8.859323621566207e-06,
      "loss": 1.3443,
      "step": 559
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.9652473718399738,
      "learning_rate": 8.853397006806183e-06,
      "loss": 1.3743,
      "step": 560
    },
    {
      "epoch": 0.8976,
      "grad_norm": 0.8463146639322702,
      "learning_rate": 8.847457028761783e-06,
      "loss": 1.4216,
      "step": 561
    },
    {
      "epoch": 0.8992,
      "grad_norm": 0.8878924544314835,
      "learning_rate": 8.841503708032398e-06,
      "loss": 1.3696,
      "step": 562
    },
    {
      "epoch": 0.9008,
      "grad_norm": 0.775901377069116,
      "learning_rate": 8.835537065263684e-06,
      "loss": 1.388,
      "step": 563
    },
    {
      "epoch": 0.9024,
      "grad_norm": 0.8732914426341744,
      "learning_rate": 8.829557121147499e-06,
      "loss": 1.3667,
      "step": 564
    },
    {
      "epoch": 0.904,
      "grad_norm": 0.8204729073641053,
      "learning_rate": 8.82356389642183e-06,
      "loss": 1.3617,
      "step": 565
    },
    {
      "epoch": 0.9056,
      "grad_norm": 0.8564776576753247,
      "learning_rate": 8.817557411870717e-06,
      "loss": 1.3723,
      "step": 566
    },
    {
      "epoch": 0.9072,
      "grad_norm": 0.8999167125873342,
      "learning_rate": 8.811537688324187e-06,
      "loss": 1.3843,
      "step": 567
    },
    {
      "epoch": 0.9088,
      "grad_norm": 0.7484649186146651,
      "learning_rate": 8.805504746658183e-06,
      "loss": 1.4586,
      "step": 568
    },
    {
      "epoch": 0.9104,
      "grad_norm": 0.8922090504614343,
      "learning_rate": 8.799458607794476e-06,
      "loss": 1.4197,
      "step": 569
    },
    {
      "epoch": 0.912,
      "grad_norm": 0.7590571867301965,
      "learning_rate": 8.793399292700616e-06,
      "loss": 1.4117,
      "step": 570
    },
    {
      "epoch": 0.9136,
      "grad_norm": 0.83114890046443,
      "learning_rate": 8.787326822389836e-06,
      "loss": 1.4239,
      "step": 571
    },
    {
      "epoch": 0.9152,
      "grad_norm": 0.8778762968636663,
      "learning_rate": 8.781241217921e-06,
      "loss": 1.3387,
      "step": 572
    },
    {
      "epoch": 0.9168,
      "grad_norm": 0.8126551106814256,
      "learning_rate": 8.775142500398513e-06,
      "loss": 1.3812,
      "step": 573
    },
    {
      "epoch": 0.9184,
      "grad_norm": 0.8700249543807123,
      "learning_rate": 8.769030690972262e-06,
      "loss": 1.3791,
      "step": 574
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6980772166243155,
      "learning_rate": 8.76290581083753e-06,
      "loss": 1.3671,
      "step": 575
    },
    {
      "epoch": 0.9216,
      "grad_norm": 0.8696530822789681,
      "learning_rate": 8.756767881234928e-06,
      "loss": 1.3771,
      "step": 576
    },
    {
      "epoch": 0.9232,
      "grad_norm": 0.7935768138474165,
      "learning_rate": 8.750616923450328e-06,
      "loss": 1.4007,
      "step": 577
    },
    {
      "epoch": 0.9248,
      "grad_norm": 0.8036805929225691,
      "learning_rate": 8.744452958814775e-06,
      "loss": 1.4196,
      "step": 578
    },
    {
      "epoch": 0.9264,
      "grad_norm": 0.7847864627511938,
      "learning_rate": 8.738276008704426e-06,
      "loss": 1.3935,
      "step": 579
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.7529665463414675,
      "learning_rate": 8.732086094540467e-06,
      "loss": 1.3355,
      "step": 580
    },
    {
      "epoch": 0.9296,
      "grad_norm": 0.7800856818901231,
      "learning_rate": 8.725883237789046e-06,
      "loss": 1.3463,
      "step": 581
    },
    {
      "epoch": 0.9312,
      "grad_norm": 0.7559193146779338,
      "learning_rate": 8.719667459961191e-06,
      "loss": 1.3471,
      "step": 582
    },
    {
      "epoch": 0.9328,
      "grad_norm": 0.7572296492124867,
      "learning_rate": 8.713438782612743e-06,
      "loss": 1.3964,
      "step": 583
    },
    {
      "epoch": 0.9344,
      "grad_norm": 0.7202177486392087,
      "learning_rate": 8.707197227344275e-06,
      "loss": 1.4221,
      "step": 584
    },
    {
      "epoch": 0.936,
      "grad_norm": 0.7494911244095489,
      "learning_rate": 8.700942815801023e-06,
      "loss": 1.3754,
      "step": 585
    },
    {
      "epoch": 0.9376,
      "grad_norm": 0.7616160295121883,
      "learning_rate": 8.6946755696728e-06,
      "loss": 1.3981,
      "step": 586
    },
    {
      "epoch": 0.9392,
      "grad_norm": 0.7368048157620367,
      "learning_rate": 8.688395510693939e-06,
      "loss": 1.3686,
      "step": 587
    },
    {
      "epoch": 0.9408,
      "grad_norm": 0.7265544031751773,
      "learning_rate": 8.682102660643196e-06,
      "loss": 1.3748,
      "step": 588
    },
    {
      "epoch": 0.9424,
      "grad_norm": 0.8485239766623587,
      "learning_rate": 8.675797041343696e-06,
      "loss": 1.3429,
      "step": 589
    },
    {
      "epoch": 0.944,
      "grad_norm": 0.7489094211323779,
      "learning_rate": 8.669478674662839e-06,
      "loss": 1.3931,
      "step": 590
    },
    {
      "epoch": 0.9456,
      "grad_norm": 0.7706545912133751,
      "learning_rate": 8.663147582512232e-06,
      "loss": 1.4284,
      "step": 591
    },
    {
      "epoch": 0.9472,
      "grad_norm": 0.7958261955655674,
      "learning_rate": 8.65680378684762e-06,
      "loss": 1.4734,
      "step": 592
    },
    {
      "epoch": 0.9488,
      "grad_norm": 0.7627018477381275,
      "learning_rate": 8.6504473096688e-06,
      "loss": 1.3959,
      "step": 593
    },
    {
      "epoch": 0.9504,
      "grad_norm": 0.8816765102136392,
      "learning_rate": 8.64407817301954e-06,
      "loss": 1.388,
      "step": 594
    },
    {
      "epoch": 0.952,
      "grad_norm": 0.8445238715646556,
      "learning_rate": 8.637696398987517e-06,
      "loss": 1.3261,
      "step": 595
    },
    {
      "epoch": 0.9536,
      "grad_norm": 0.8342961726857758,
      "learning_rate": 8.631302009704235e-06,
      "loss": 1.3252,
      "step": 596
    },
    {
      "epoch": 0.9552,
      "grad_norm": 0.8290124408833847,
      "learning_rate": 8.624895027344943e-06,
      "loss": 1.4278,
      "step": 597
    },
    {
      "epoch": 0.9568,
      "grad_norm": 0.7166594583514148,
      "learning_rate": 8.618475474128563e-06,
      "loss": 1.3993,
      "step": 598
    },
    {
      "epoch": 0.9584,
      "grad_norm": 0.8299142268041004,
      "learning_rate": 8.61204337231761e-06,
      "loss": 1.4521,
      "step": 599
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7634025359016703,
      "learning_rate": 8.605598744218122e-06,
      "loss": 1.35,
      "step": 600
    },
    {
      "epoch": 0.9616,
      "grad_norm": 0.7215796515817717,
      "learning_rate": 8.599141612179572e-06,
      "loss": 1.3913,
      "step": 601
    },
    {
      "epoch": 0.9632,
      "grad_norm": 0.691954346866529,
      "learning_rate": 8.592671998594794e-06,
      "loss": 1.374,
      "step": 602
    },
    {
      "epoch": 0.9648,
      "grad_norm": 0.7900813256144187,
      "learning_rate": 8.586189925899913e-06,
      "loss": 1.423,
      "step": 603
    },
    {
      "epoch": 0.9664,
      "grad_norm": 0.7466107243097425,
      "learning_rate": 8.57969541657426e-06,
      "loss": 1.3446,
      "step": 604
    },
    {
      "epoch": 0.968,
      "grad_norm": 0.8939956868844258,
      "learning_rate": 8.57318849314029e-06,
      "loss": 1.3242,
      "step": 605
    },
    {
      "epoch": 0.9696,
      "grad_norm": 0.7762098608849748,
      "learning_rate": 8.566669178163513e-06,
      "loss": 1.3713,
      "step": 606
    },
    {
      "epoch": 0.9712,
      "grad_norm": 0.7691026808384659,
      "learning_rate": 8.560137494252416e-06,
      "loss": 1.3931,
      "step": 607
    },
    {
      "epoch": 0.9728,
      "grad_norm": 0.7565517807290206,
      "learning_rate": 8.553593464058374e-06,
      "loss": 1.4672,
      "step": 608
    },
    {
      "epoch": 0.9744,
      "grad_norm": 0.7203792671056047,
      "learning_rate": 8.54703711027558e-06,
      "loss": 1.3735,
      "step": 609
    },
    {
      "epoch": 0.976,
      "grad_norm": 0.7986275181484728,
      "learning_rate": 8.540468455640964e-06,
      "loss": 1.3709,
      "step": 610
    },
    {
      "epoch": 0.9776,
      "grad_norm": 0.7887506735246259,
      "learning_rate": 8.533887522934114e-06,
      "loss": 1.4527,
      "step": 611
    },
    {
      "epoch": 0.9792,
      "grad_norm": 0.7277691765152821,
      "learning_rate": 8.527294334977201e-06,
      "loss": 1.4231,
      "step": 612
    },
    {
      "epoch": 0.9808,
      "grad_norm": 0.8847832730570527,
      "learning_rate": 8.520688914634894e-06,
      "loss": 1.5213,
      "step": 613
    },
    {
      "epoch": 0.9824,
      "grad_norm": 0.7512671531931401,
      "learning_rate": 8.51407128481428e-06,
      "loss": 1.4605,
      "step": 614
    },
    {
      "epoch": 0.984,
      "grad_norm": 0.9680047146118907,
      "learning_rate": 8.507441468464792e-06,
      "loss": 1.4016,
      "step": 615
    },
    {
      "epoch": 0.9856,
      "grad_norm": 0.7789829883004501,
      "learning_rate": 8.50079948857812e-06,
      "loss": 1.4172,
      "step": 616
    },
    {
      "epoch": 0.9872,
      "grad_norm": 0.9390164482594086,
      "learning_rate": 8.494145368188143e-06,
      "loss": 1.3901,
      "step": 617
    },
    {
      "epoch": 0.9888,
      "grad_norm": 0.8883511828034028,
      "learning_rate": 8.487479130370838e-06,
      "loss": 1.4116,
      "step": 618
    },
    {
      "epoch": 0.9904,
      "grad_norm": 0.8155403221649302,
      "learning_rate": 8.480800798244202e-06,
      "loss": 1.3582,
      "step": 619
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.8463552527671379,
      "learning_rate": 8.47411039496818e-06,
      "loss": 1.4555,
      "step": 620
    },
    {
      "epoch": 0.9936,
      "grad_norm": 0.8688354761175793,
      "learning_rate": 8.467407943744574e-06,
      "loss": 1.3741,
      "step": 621
    },
    {
      "epoch": 0.9952,
      "grad_norm": 0.849744871655598,
      "learning_rate": 8.460693467816972e-06,
      "loss": 1.3217,
      "step": 622
    },
    {
      "epoch": 0.9968,
      "grad_norm": 0.885520757521042,
      "learning_rate": 8.453966990470656e-06,
      "loss": 1.4777,
      "step": 623
    },
    {
      "epoch": 0.9984,
      "grad_norm": 0.7556120406017404,
      "learning_rate": 8.447228535032536e-06,
      "loss": 1.3359,
      "step": 624
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8823313839651297,
      "learning_rate": 8.440478124871054e-06,
      "loss": 1.4011,
      "step": 625
    },
    {
      "epoch": 1.0016,
      "grad_norm": 0.8593184628138082,
      "learning_rate": 8.433715783396115e-06,
      "loss": 1.325,
      "step": 626
    },
    {
      "epoch": 1.0032,
      "grad_norm": 0.7995063736540354,
      "learning_rate": 8.426941534058999e-06,
      "loss": 1.3224,
      "step": 627
    },
    {
      "epoch": 1.0048,
      "grad_norm": 0.8100847411960652,
      "learning_rate": 8.420155400352279e-06,
      "loss": 1.3007,
      "step": 628
    },
    {
      "epoch": 1.0064,
      "grad_norm": 0.8320497728892343,
      "learning_rate": 8.413357405809748e-06,
      "loss": 1.2983,
      "step": 629
    },
    {
      "epoch": 1.008,
      "grad_norm": 0.8178706569653971,
      "learning_rate": 8.406547574006326e-06,
      "loss": 1.3622,
      "step": 630
    },
    {
      "epoch": 1.0096,
      "grad_norm": 0.9369627320325458,
      "learning_rate": 8.399725928557985e-06,
      "loss": 1.3704,
      "step": 631
    },
    {
      "epoch": 1.0112,
      "grad_norm": 0.8681410877155201,
      "learning_rate": 8.39289249312167e-06,
      "loss": 1.3916,
      "step": 632
    },
    {
      "epoch": 1.0128,
      "grad_norm": 0.7860394519331632,
      "learning_rate": 8.386047291395208e-06,
      "loss": 1.2824,
      "step": 633
    },
    {
      "epoch": 1.0144,
      "grad_norm": 0.8614831886427855,
      "learning_rate": 8.37919034711723e-06,
      "loss": 1.3242,
      "step": 634
    },
    {
      "epoch": 1.016,
      "grad_norm": 0.7990188820206938,
      "learning_rate": 8.372321684067092e-06,
      "loss": 1.4018,
      "step": 635
    },
    {
      "epoch": 1.0176,
      "grad_norm": 1.0019118761746064,
      "learning_rate": 8.36544132606479e-06,
      "loss": 1.3451,
      "step": 636
    },
    {
      "epoch": 1.0192,
      "grad_norm": 0.7785104454956524,
      "learning_rate": 8.358549296970877e-06,
      "loss": 1.3651,
      "step": 637
    },
    {
      "epoch": 1.0208,
      "grad_norm": 0.7935226397363923,
      "learning_rate": 8.351645620686377e-06,
      "loss": 1.294,
      "step": 638
    },
    {
      "epoch": 1.0224,
      "grad_norm": 0.8070719801145307,
      "learning_rate": 8.34473032115271e-06,
      "loss": 1.3427,
      "step": 639
    },
    {
      "epoch": 1.024,
      "grad_norm": 0.8029922752674268,
      "learning_rate": 8.337803422351602e-06,
      "loss": 1.3168,
      "step": 640
    },
    {
      "epoch": 1.0256,
      "grad_norm": 0.9182084281277861,
      "learning_rate": 8.33086494830501e-06,
      "loss": 1.3276,
      "step": 641
    },
    {
      "epoch": 1.0272,
      "grad_norm": 0.8137851234743319,
      "learning_rate": 8.323914923075018e-06,
      "loss": 1.3724,
      "step": 642
    },
    {
      "epoch": 1.0288,
      "grad_norm": 0.904033398928434,
      "learning_rate": 8.316953370763788e-06,
      "loss": 1.3297,
      "step": 643
    },
    {
      "epoch": 1.0304,
      "grad_norm": 0.8276440733385142,
      "learning_rate": 8.309980315513444e-06,
      "loss": 1.3477,
      "step": 644
    },
    {
      "epoch": 1.032,
      "grad_norm": 0.9807982189151975,
      "learning_rate": 8.302995781506007e-06,
      "loss": 1.3382,
      "step": 645
    },
    {
      "epoch": 1.0336,
      "grad_norm": 0.9911951660875237,
      "learning_rate": 8.295999792963301e-06,
      "loss": 1.3367,
      "step": 646
    },
    {
      "epoch": 1.0352,
      "grad_norm": 0.8357509357497808,
      "learning_rate": 8.288992374146878e-06,
      "loss": 1.3837,
      "step": 647
    },
    {
      "epoch": 1.0368,
      "grad_norm": 0.9283501861320693,
      "learning_rate": 8.281973549357927e-06,
      "loss": 1.347,
      "step": 648
    },
    {
      "epoch": 1.0384,
      "grad_norm": 0.783705369268492,
      "learning_rate": 8.274943342937191e-06,
      "loss": 1.3972,
      "step": 649
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.80673226736088,
      "learning_rate": 8.267901779264889e-06,
      "loss": 1.3041,
      "step": 650
    },
    {
      "epoch": 1.0416,
      "grad_norm": 0.8057694519070149,
      "learning_rate": 8.260848882760616e-06,
      "loss": 1.3191,
      "step": 651
    },
    {
      "epoch": 1.0432,
      "grad_norm": 0.8801106309309326,
      "learning_rate": 8.25378467788328e-06,
      "loss": 1.3361,
      "step": 652
    },
    {
      "epoch": 1.0448,
      "grad_norm": 0.7620200662133425,
      "learning_rate": 8.246709189130997e-06,
      "loss": 1.2894,
      "step": 653
    },
    {
      "epoch": 1.0464,
      "grad_norm": 0.8897767164796212,
      "learning_rate": 8.23962244104102e-06,
      "loss": 1.3911,
      "step": 654
    },
    {
      "epoch": 1.048,
      "grad_norm": 0.8210387493603643,
      "learning_rate": 8.232524458189644e-06,
      "loss": 1.3181,
      "step": 655
    },
    {
      "epoch": 1.0496,
      "grad_norm": 0.7940387792692596,
      "learning_rate": 8.225415265192126e-06,
      "loss": 1.3497,
      "step": 656
    },
    {
      "epoch": 1.0512,
      "grad_norm": 0.8895934178753787,
      "learning_rate": 8.218294886702606e-06,
      "loss": 1.3602,
      "step": 657
    },
    {
      "epoch": 1.0528,
      "grad_norm": 0.7783252128720803,
      "learning_rate": 8.211163347414005e-06,
      "loss": 1.3953,
      "step": 658
    },
    {
      "epoch": 1.0544,
      "grad_norm": 0.9524842166019186,
      "learning_rate": 8.20402067205795e-06,
      "loss": 1.3713,
      "step": 659
    },
    {
      "epoch": 1.056,
      "grad_norm": 0.7538242605536091,
      "learning_rate": 8.196866885404697e-06,
      "loss": 1.279,
      "step": 660
    },
    {
      "epoch": 1.0576,
      "grad_norm": 0.8668643073297734,
      "learning_rate": 8.18970201226302e-06,
      "loss": 1.3479,
      "step": 661
    },
    {
      "epoch": 1.0592,
      "grad_norm": 0.7264526449351055,
      "learning_rate": 8.182526077480153e-06,
      "loss": 1.3124,
      "step": 662
    },
    {
      "epoch": 1.0608,
      "grad_norm": 0.8359555645558399,
      "learning_rate": 8.175339105941685e-06,
      "loss": 1.3226,
      "step": 663
    },
    {
      "epoch": 1.0624,
      "grad_norm": 0.8020330214895224,
      "learning_rate": 8.168141122571478e-06,
      "loss": 1.353,
      "step": 664
    },
    {
      "epoch": 1.064,
      "grad_norm": 0.7045122468551239,
      "learning_rate": 8.160932152331587e-06,
      "loss": 1.3365,
      "step": 665
    },
    {
      "epoch": 1.0656,
      "grad_norm": 0.8660230190266233,
      "learning_rate": 8.153712220222163e-06,
      "loss": 1.3242,
      "step": 666
    },
    {
      "epoch": 1.0672,
      "grad_norm": 0.8148939917079442,
      "learning_rate": 8.14648135128138e-06,
      "loss": 1.2957,
      "step": 667
    },
    {
      "epoch": 1.0688,
      "grad_norm": 0.7663314791126777,
      "learning_rate": 8.139239570585334e-06,
      "loss": 1.352,
      "step": 668
    },
    {
      "epoch": 1.0704,
      "grad_norm": 0.7497779061617293,
      "learning_rate": 8.131986903247959e-06,
      "loss": 1.3195,
      "step": 669
    },
    {
      "epoch": 1.072,
      "grad_norm": 0.7780950004209306,
      "learning_rate": 8.124723374420951e-06,
      "loss": 1.2785,
      "step": 670
    },
    {
      "epoch": 1.0735999999999999,
      "grad_norm": 0.7264921207801373,
      "learning_rate": 8.117449009293668e-06,
      "loss": 1.3304,
      "step": 671
    },
    {
      "epoch": 1.0752,
      "grad_norm": 0.7330451796448315,
      "learning_rate": 8.11016383309305e-06,
      "loss": 1.2835,
      "step": 672
    },
    {
      "epoch": 1.0768,
      "grad_norm": 0.6711736657487075,
      "learning_rate": 8.102867871083528e-06,
      "loss": 1.3257,
      "step": 673
    },
    {
      "epoch": 1.0784,
      "grad_norm": 0.7190991852111623,
      "learning_rate": 8.095561148566932e-06,
      "loss": 1.3448,
      "step": 674
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7674523426239592,
      "learning_rate": 8.088243690882421e-06,
      "loss": 1.3239,
      "step": 675
    },
    {
      "epoch": 1.0816,
      "grad_norm": 0.6528064990227597,
      "learning_rate": 8.080915523406371e-06,
      "loss": 1.3061,
      "step": 676
    },
    {
      "epoch": 1.0832,
      "grad_norm": 0.7768004145671703,
      "learning_rate": 8.073576671552303e-06,
      "loss": 1.299,
      "step": 677
    },
    {
      "epoch": 1.0848,
      "grad_norm": 0.7791454889933257,
      "learning_rate": 8.06622716077079e-06,
      "loss": 1.3696,
      "step": 678
    },
    {
      "epoch": 1.0864,
      "grad_norm": 0.7063396805443227,
      "learning_rate": 8.058867016549372e-06,
      "loss": 1.3129,
      "step": 679
    },
    {
      "epoch": 1.088,
      "grad_norm": 0.741646832419459,
      "learning_rate": 8.051496264412464e-06,
      "loss": 1.3513,
      "step": 680
    },
    {
      "epoch": 1.0896,
      "grad_norm": 0.6961803268938446,
      "learning_rate": 8.044114929921264e-06,
      "loss": 1.3223,
      "step": 681
    },
    {
      "epoch": 1.0912,
      "grad_norm": 0.7033884293811211,
      "learning_rate": 8.036723038673675e-06,
      "loss": 1.3182,
      "step": 682
    },
    {
      "epoch": 1.0928,
      "grad_norm": 0.7668570614069247,
      "learning_rate": 8.029320616304204e-06,
      "loss": 1.2926,
      "step": 683
    },
    {
      "epoch": 1.0944,
      "grad_norm": 0.7405073948828712,
      "learning_rate": 8.021907688483885e-06,
      "loss": 1.3374,
      "step": 684
    },
    {
      "epoch": 1.096,
      "grad_norm": 0.7528912255480958,
      "learning_rate": 8.01448428092018e-06,
      "loss": 1.3249,
      "step": 685
    },
    {
      "epoch": 1.0976,
      "grad_norm": 0.7552947350479217,
      "learning_rate": 8.007050419356898e-06,
      "loss": 1.329,
      "step": 686
    },
    {
      "epoch": 1.0992,
      "grad_norm": 0.7309855476716728,
      "learning_rate": 7.999606129574096e-06,
      "loss": 1.3866,
      "step": 687
    },
    {
      "epoch": 1.1008,
      "grad_norm": 0.7511301230256162,
      "learning_rate": 7.992151437387999e-06,
      "loss": 1.4042,
      "step": 688
    },
    {
      "epoch": 1.1024,
      "grad_norm": 0.7205097880838082,
      "learning_rate": 7.984686368650907e-06,
      "loss": 1.3371,
      "step": 689
    },
    {
      "epoch": 1.104,
      "grad_norm": 0.7422237555455701,
      "learning_rate": 7.977210949251102e-06,
      "loss": 1.3164,
      "step": 690
    },
    {
      "epoch": 1.1056,
      "grad_norm": 0.6918557507440017,
      "learning_rate": 7.969725205112766e-06,
      "loss": 1.3822,
      "step": 691
    },
    {
      "epoch": 1.1072,
      "grad_norm": 0.7763515205094746,
      "learning_rate": 7.962229162195882e-06,
      "loss": 1.3626,
      "step": 692
    },
    {
      "epoch": 1.1088,
      "grad_norm": 0.6817544097591205,
      "learning_rate": 7.95472284649615e-06,
      "loss": 1.3439,
      "step": 693
    },
    {
      "epoch": 1.1104,
      "grad_norm": 0.8819257089302646,
      "learning_rate": 7.947206284044896e-06,
      "loss": 1.3329,
      "step": 694
    },
    {
      "epoch": 1.112,
      "grad_norm": 0.7544655481133858,
      "learning_rate": 7.939679500908982e-06,
      "loss": 1.3532,
      "step": 695
    },
    {
      "epoch": 1.1136,
      "grad_norm": 0.7617702571000862,
      "learning_rate": 7.932142523190711e-06,
      "loss": 1.326,
      "step": 696
    },
    {
      "epoch": 1.1152,
      "grad_norm": 0.878138811882252,
      "learning_rate": 7.924595377027741e-06,
      "loss": 1.3202,
      "step": 697
    },
    {
      "epoch": 1.1168,
      "grad_norm": 0.7326150374148215,
      "learning_rate": 7.917038088592997e-06,
      "loss": 1.3665,
      "step": 698
    },
    {
      "epoch": 1.1184,
      "grad_norm": 1.152827817761174,
      "learning_rate": 7.90947068409457e-06,
      "loss": 1.418,
      "step": 699
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8250797253590365,
      "learning_rate": 7.90189318977564e-06,
      "loss": 1.3635,
      "step": 700
    },
    {
      "epoch": 1.1216,
      "grad_norm": 0.8609903255105108,
      "learning_rate": 7.894305631914373e-06,
      "loss": 1.3313,
      "step": 701
    },
    {
      "epoch": 1.1232,
      "grad_norm": 0.9108024633911359,
      "learning_rate": 7.886708036823838e-06,
      "loss": 1.3732,
      "step": 702
    },
    {
      "epoch": 1.1248,
      "grad_norm": 0.733469022746333,
      "learning_rate": 7.879100430851907e-06,
      "loss": 1.3209,
      "step": 703
    },
    {
      "epoch": 1.1264,
      "grad_norm": 0.9742998714755888,
      "learning_rate": 7.871482840381174e-06,
      "loss": 1.3218,
      "step": 704
    },
    {
      "epoch": 1.1280000000000001,
      "grad_norm": 0.7000034414874344,
      "learning_rate": 7.863855291828857e-06,
      "loss": 1.3501,
      "step": 705
    },
    {
      "epoch": 1.1296,
      "grad_norm": 0.8850055770561902,
      "learning_rate": 7.856217811646707e-06,
      "loss": 1.3957,
      "step": 706
    },
    {
      "epoch": 1.1312,
      "grad_norm": 0.8173667965041101,
      "learning_rate": 7.848570426320918e-06,
      "loss": 1.3175,
      "step": 707
    },
    {
      "epoch": 1.1328,
      "grad_norm": 0.8254473941166518,
      "learning_rate": 7.840913162372032e-06,
      "loss": 1.3515,
      "step": 708
    },
    {
      "epoch": 1.1344,
      "grad_norm": 3.5483967466429664,
      "learning_rate": 7.833246046354856e-06,
      "loss": 1.3649,
      "step": 709
    },
    {
      "epoch": 1.1360000000000001,
      "grad_norm": 1.1996505634541448,
      "learning_rate": 7.825569104858353e-06,
      "loss": 1.3272,
      "step": 710
    },
    {
      "epoch": 1.1376,
      "grad_norm": 0.7246481492273005,
      "learning_rate": 7.81788236450557e-06,
      "loss": 1.3876,
      "step": 711
    },
    {
      "epoch": 1.1392,
      "grad_norm": 0.9499456457409373,
      "learning_rate": 7.810185851953529e-06,
      "loss": 1.395,
      "step": 712
    },
    {
      "epoch": 1.1408,
      "grad_norm": 0.8948510614113747,
      "learning_rate": 7.802479593893142e-06,
      "loss": 1.4363,
      "step": 713
    },
    {
      "epoch": 1.1424,
      "grad_norm": 0.7635446069599047,
      "learning_rate": 7.794763617049124e-06,
      "loss": 1.3285,
      "step": 714
    },
    {
      "epoch": 1.144,
      "grad_norm": 0.8235295492991268,
      "learning_rate": 7.787037948179884e-06,
      "loss": 1.3321,
      "step": 715
    },
    {
      "epoch": 1.1456,
      "grad_norm": 0.8945545359391827,
      "learning_rate": 7.779302614077449e-06,
      "loss": 1.377,
      "step": 716
    },
    {
      "epoch": 1.1472,
      "grad_norm": 0.7540489596347976,
      "learning_rate": 7.771557641567363e-06,
      "loss": 1.3548,
      "step": 717
    },
    {
      "epoch": 1.1488,
      "grad_norm": 0.9098380367197593,
      "learning_rate": 7.763803057508594e-06,
      "loss": 1.2944,
      "step": 718
    },
    {
      "epoch": 1.1504,
      "grad_norm": 0.79541561392294,
      "learning_rate": 7.756038888793446e-06,
      "loss": 1.4317,
      "step": 719
    },
    {
      "epoch": 1.152,
      "grad_norm": 0.8104554005166604,
      "learning_rate": 7.748265162347455e-06,
      "loss": 1.4002,
      "step": 720
    },
    {
      "epoch": 1.1536,
      "grad_norm": 0.8488875440170629,
      "learning_rate": 7.740481905129307e-06,
      "loss": 1.3602,
      "step": 721
    },
    {
      "epoch": 1.1552,
      "grad_norm": 0.764137882301824,
      "learning_rate": 7.732689144130741e-06,
      "loss": 1.3616,
      "step": 722
    },
    {
      "epoch": 1.1568,
      "grad_norm": 0.8354888048072454,
      "learning_rate": 7.724886906376451e-06,
      "loss": 1.3358,
      "step": 723
    },
    {
      "epoch": 1.1584,
      "grad_norm": 3.216606401502655,
      "learning_rate": 7.717075218923998e-06,
      "loss": 1.3578,
      "step": 724
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0606244842240788,
      "learning_rate": 7.709254108863714e-06,
      "loss": 1.4128,
      "step": 725
    },
    {
      "epoch": 1.1616,
      "grad_norm": 0.7860075283395405,
      "learning_rate": 7.701423603318605e-06,
      "loss": 1.3572,
      "step": 726
    },
    {
      "epoch": 1.1632,
      "grad_norm": 0.8534694242473752,
      "learning_rate": 7.693583729444263e-06,
      "loss": 1.3155,
      "step": 727
    },
    {
      "epoch": 1.1648,
      "grad_norm": 0.8339474552839605,
      "learning_rate": 7.685734514428767e-06,
      "loss": 1.357,
      "step": 728
    },
    {
      "epoch": 1.1663999999999999,
      "grad_norm": 0.8277696927349593,
      "learning_rate": 7.677875985492591e-06,
      "loss": 1.2708,
      "step": 729
    },
    {
      "epoch": 1.168,
      "grad_norm": 0.9821065290614573,
      "learning_rate": 7.67000816988851e-06,
      "loss": 1.3377,
      "step": 730
    },
    {
      "epoch": 1.1696,
      "grad_norm": 0.8567141199261764,
      "learning_rate": 7.662131094901499e-06,
      "loss": 1.3337,
      "step": 731
    },
    {
      "epoch": 1.1712,
      "grad_norm": 1.1912757335269626,
      "learning_rate": 7.654244787848655e-06,
      "loss": 1.3152,
      "step": 732
    },
    {
      "epoch": 1.1728,
      "grad_norm": 0.8216721037396176,
      "learning_rate": 7.646349276079079e-06,
      "loss": 1.2996,
      "step": 733
    },
    {
      "epoch": 1.1743999999999999,
      "grad_norm": 0.8058971982498022,
      "learning_rate": 7.6384445869738e-06,
      "loss": 1.2942,
      "step": 734
    },
    {
      "epoch": 1.176,
      "grad_norm": 0.7689278212873505,
      "learning_rate": 7.630530747945672e-06,
      "loss": 1.3713,
      "step": 735
    },
    {
      "epoch": 1.1776,
      "grad_norm": 0.803837118735867,
      "learning_rate": 7.622607786439279e-06,
      "loss": 1.3637,
      "step": 736
    },
    {
      "epoch": 1.1792,
      "grad_norm": 0.6979574858792883,
      "learning_rate": 7.6146757299308406e-06,
      "loss": 1.3565,
      "step": 737
    },
    {
      "epoch": 1.1808,
      "grad_norm": 0.7487900494713055,
      "learning_rate": 7.606734605928123e-06,
      "loss": 1.3633,
      "step": 738
    },
    {
      "epoch": 1.1824,
      "grad_norm": 0.7276215392487727,
      "learning_rate": 7.598784441970329e-06,
      "loss": 1.3578,
      "step": 739
    },
    {
      "epoch": 1.184,
      "grad_norm": 0.7552024880159864,
      "learning_rate": 7.590825265628019e-06,
      "loss": 1.3511,
      "step": 740
    },
    {
      "epoch": 1.1856,
      "grad_norm": 0.7829632904295207,
      "learning_rate": 7.5828571045030005e-06,
      "loss": 1.3021,
      "step": 741
    },
    {
      "epoch": 1.1872,
      "grad_norm": 0.8070146762721719,
      "learning_rate": 7.574879986228245e-06,
      "loss": 1.257,
      "step": 742
    },
    {
      "epoch": 1.1888,
      "grad_norm": 0.7393011442886225,
      "learning_rate": 7.566893938467788e-06,
      "loss": 1.3584,
      "step": 743
    },
    {
      "epoch": 1.1904,
      "grad_norm": 0.874024971274496,
      "learning_rate": 7.558898988916624e-06,
      "loss": 1.2933,
      "step": 744
    },
    {
      "epoch": 1.192,
      "grad_norm": 0.6934970797266018,
      "learning_rate": 7.550895165300626e-06,
      "loss": 1.3716,
      "step": 745
    },
    {
      "epoch": 1.1936,
      "grad_norm": 0.8485448162115721,
      "learning_rate": 7.542882495376437e-06,
      "loss": 1.342,
      "step": 746
    },
    {
      "epoch": 1.1952,
      "grad_norm": 0.738522345931944,
      "learning_rate": 7.5348610069313795e-06,
      "loss": 1.3025,
      "step": 747
    },
    {
      "epoch": 1.1968,
      "grad_norm": 0.8165797986878177,
      "learning_rate": 7.5268307277833605e-06,
      "loss": 1.3701,
      "step": 748
    },
    {
      "epoch": 1.1984,
      "grad_norm": 0.6967206932572357,
      "learning_rate": 7.518791685780769e-06,
      "loss": 1.3221,
      "step": 749
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7472702983279,
      "learning_rate": 7.5107439088023845e-06,
      "loss": 1.367,
      "step": 750
    },
    {
      "epoch": 1.2016,
      "grad_norm": 0.7097469816112091,
      "learning_rate": 7.502687424757278e-06,
      "loss": 1.3171,
      "step": 751
    },
    {
      "epoch": 1.2032,
      "grad_norm": 0.7220994218378944,
      "learning_rate": 7.4946222615847165e-06,
      "loss": 1.3385,
      "step": 752
    },
    {
      "epoch": 1.2048,
      "grad_norm": 0.7056190186573942,
      "learning_rate": 7.486548447254065e-06,
      "loss": 1.2981,
      "step": 753
    },
    {
      "epoch": 1.2064,
      "grad_norm": 0.72009050279339,
      "learning_rate": 7.478466009764692e-06,
      "loss": 1.3325,
      "step": 754
    },
    {
      "epoch": 1.208,
      "grad_norm": 0.7144898475656866,
      "learning_rate": 7.470374977145867e-06,
      "loss": 1.3066,
      "step": 755
    },
    {
      "epoch": 1.2096,
      "grad_norm": 0.6810991161407781,
      "learning_rate": 7.462275377456671e-06,
      "loss": 1.371,
      "step": 756
    },
    {
      "epoch": 1.2112,
      "grad_norm": 0.7192152825297243,
      "learning_rate": 7.4541672387858895e-06,
      "loss": 1.3598,
      "step": 757
    },
    {
      "epoch": 1.2128,
      "grad_norm": 0.7732567133810291,
      "learning_rate": 7.446050589251928e-06,
      "loss": 1.262,
      "step": 758
    },
    {
      "epoch": 1.2144,
      "grad_norm": 0.751374089999309,
      "learning_rate": 7.437925457002697e-06,
      "loss": 1.4227,
      "step": 759
    },
    {
      "epoch": 1.216,
      "grad_norm": 0.7672410842295768,
      "learning_rate": 7.429791870215535e-06,
      "loss": 1.3197,
      "step": 760
    },
    {
      "epoch": 1.2176,
      "grad_norm": 0.7748581537831023,
      "learning_rate": 7.421649857097092e-06,
      "loss": 1.4,
      "step": 761
    },
    {
      "epoch": 1.2192,
      "grad_norm": 0.8555340394464912,
      "learning_rate": 7.413499445883245e-06,
      "loss": 1.3374,
      "step": 762
    },
    {
      "epoch": 1.2208,
      "grad_norm": 0.7699773335587452,
      "learning_rate": 7.405340664838994e-06,
      "loss": 1.3988,
      "step": 763
    },
    {
      "epoch": 1.2224,
      "grad_norm": 0.8051563488460314,
      "learning_rate": 7.39717354225836e-06,
      "loss": 1.3346,
      "step": 764
    },
    {
      "epoch": 1.224,
      "grad_norm": 0.7730802911468607,
      "learning_rate": 7.3889981064643e-06,
      "loss": 1.3647,
      "step": 765
    },
    {
      "epoch": 1.2256,
      "grad_norm": 0.7690403274999603,
      "learning_rate": 7.380814385808594e-06,
      "loss": 1.3276,
      "step": 766
    },
    {
      "epoch": 1.2272,
      "grad_norm": 0.7530707671324821,
      "learning_rate": 7.372622408671757e-06,
      "loss": 1.3716,
      "step": 767
    },
    {
      "epoch": 1.2288000000000001,
      "grad_norm": 0.8091232204220123,
      "learning_rate": 7.364422203462935e-06,
      "loss": 1.3597,
      "step": 768
    },
    {
      "epoch": 1.2304,
      "grad_norm": 0.6781773631462054,
      "learning_rate": 7.3562137986198065e-06,
      "loss": 1.3404,
      "step": 769
    },
    {
      "epoch": 1.232,
      "grad_norm": 0.8727105894319599,
      "learning_rate": 7.3479972226084925e-06,
      "loss": 1.346,
      "step": 770
    },
    {
      "epoch": 1.2336,
      "grad_norm": 0.6671477443702643,
      "learning_rate": 7.339772503923445e-06,
      "loss": 1.3452,
      "step": 771
    },
    {
      "epoch": 1.2352,
      "grad_norm": 0.7671989205024583,
      "learning_rate": 7.331539671087353e-06,
      "loss": 1.3275,
      "step": 772
    },
    {
      "epoch": 1.2368000000000001,
      "grad_norm": 0.7552877753245147,
      "learning_rate": 7.32329875265105e-06,
      "loss": 1.3158,
      "step": 773
    },
    {
      "epoch": 1.2384,
      "grad_norm": 0.7702745385121309,
      "learning_rate": 7.315049777193407e-06,
      "loss": 1.3291,
      "step": 774
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.6874768873823203,
      "learning_rate": 7.306792773321234e-06,
      "loss": 1.3576,
      "step": 775
    },
    {
      "epoch": 1.2416,
      "grad_norm": 0.7576348524021277,
      "learning_rate": 7.298527769669188e-06,
      "loss": 1.3794,
      "step": 776
    },
    {
      "epoch": 1.2432,
      "grad_norm": 0.7627002521011207,
      "learning_rate": 7.290254794899665e-06,
      "loss": 1.3223,
      "step": 777
    },
    {
      "epoch": 1.2448,
      "grad_norm": 0.7387657363346757,
      "learning_rate": 7.281973877702705e-06,
      "loss": 1.339,
      "step": 778
    },
    {
      "epoch": 1.2464,
      "grad_norm": 0.7787163012756043,
      "learning_rate": 7.2736850467958905e-06,
      "loss": 1.3565,
      "step": 779
    },
    {
      "epoch": 1.248,
      "grad_norm": 0.7340649415775544,
      "learning_rate": 7.26538833092425e-06,
      "loss": 1.333,
      "step": 780
    },
    {
      "epoch": 1.2496,
      "grad_norm": 0.7402311587953757,
      "learning_rate": 7.257083758860159e-06,
      "loss": 1.3204,
      "step": 781
    },
    {
      "epoch": 1.2511999999999999,
      "grad_norm": 0.8117946707112972,
      "learning_rate": 7.248771359403231e-06,
      "loss": 1.411,
      "step": 782
    },
    {
      "epoch": 1.2528000000000001,
      "grad_norm": 0.7858315958813772,
      "learning_rate": 7.240451161380226e-06,
      "loss": 1.2539,
      "step": 783
    },
    {
      "epoch": 1.2544,
      "grad_norm": 0.7574360086917487,
      "learning_rate": 7.232123193644957e-06,
      "loss": 1.3367,
      "step": 784
    },
    {
      "epoch": 1.256,
      "grad_norm": 0.7136657287077437,
      "learning_rate": 7.22378748507817e-06,
      "loss": 1.3785,
      "step": 785
    },
    {
      "epoch": 1.2576,
      "grad_norm": 0.7856241918754678,
      "learning_rate": 7.215444064587462e-06,
      "loss": 1.3822,
      "step": 786
    },
    {
      "epoch": 1.2591999999999999,
      "grad_norm": 0.8316523379088419,
      "learning_rate": 7.207092961107176e-06,
      "loss": 1.2994,
      "step": 787
    },
    {
      "epoch": 1.2608,
      "grad_norm": 0.7924063341560745,
      "learning_rate": 7.198734203598294e-06,
      "loss": 1.3808,
      "step": 788
    },
    {
      "epoch": 1.2624,
      "grad_norm": 0.807057424978597,
      "learning_rate": 7.190367821048346e-06,
      "loss": 1.384,
      "step": 789
    },
    {
      "epoch": 1.264,
      "grad_norm": 0.7119669904203823,
      "learning_rate": 7.181993842471301e-06,
      "loss": 1.3225,
      "step": 790
    },
    {
      "epoch": 1.2656,
      "grad_norm": 0.7787975618320615,
      "learning_rate": 7.173612296907473e-06,
      "loss": 1.3389,
      "step": 791
    },
    {
      "epoch": 1.2671999999999999,
      "grad_norm": 0.6779919997616739,
      "learning_rate": 7.165223213423416e-06,
      "loss": 1.3617,
      "step": 792
    },
    {
      "epoch": 1.2688,
      "grad_norm": 0.8090717943678012,
      "learning_rate": 7.15682662111183e-06,
      "loss": 1.3778,
      "step": 793
    },
    {
      "epoch": 1.2704,
      "grad_norm": 0.6824691001847115,
      "learning_rate": 7.148422549091447e-06,
      "loss": 1.431,
      "step": 794
    },
    {
      "epoch": 1.272,
      "grad_norm": 0.7708248822577349,
      "learning_rate": 7.140011026506945e-06,
      "loss": 1.2823,
      "step": 795
    },
    {
      "epoch": 1.2736,
      "grad_norm": 0.7299588124350784,
      "learning_rate": 7.131592082528837e-06,
      "loss": 1.3168,
      "step": 796
    },
    {
      "epoch": 1.2752,
      "grad_norm": 0.7343913192829972,
      "learning_rate": 7.12316574635337e-06,
      "loss": 1.3721,
      "step": 797
    },
    {
      "epoch": 1.2768,
      "grad_norm": 0.6977050630708325,
      "learning_rate": 7.114732047202433e-06,
      "loss": 1.2679,
      "step": 798
    },
    {
      "epoch": 1.2784,
      "grad_norm": 0.7004417266793006,
      "learning_rate": 7.106291014323445e-06,
      "loss": 1.2683,
      "step": 799
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7123324369251904,
      "learning_rate": 7.0978426769892585e-06,
      "loss": 1.2923,
      "step": 800
    },
    {
      "epoch": 1.2816,
      "grad_norm": 0.7050744002415377,
      "learning_rate": 7.089387064498057e-06,
      "loss": 1.3251,
      "step": 801
    },
    {
      "epoch": 1.2832,
      "grad_norm": 0.6906875743759789,
      "learning_rate": 7.080924206173253e-06,
      "loss": 1.32,
      "step": 802
    },
    {
      "epoch": 1.2848,
      "grad_norm": 0.6817390826299573,
      "learning_rate": 7.072454131363391e-06,
      "loss": 1.3507,
      "step": 803
    },
    {
      "epoch": 1.2864,
      "grad_norm": 0.676081292197881,
      "learning_rate": 7.063976869442037e-06,
      "loss": 1.3422,
      "step": 804
    },
    {
      "epoch": 1.288,
      "grad_norm": 0.6854331934904734,
      "learning_rate": 7.055492449807684e-06,
      "loss": 1.27,
      "step": 805
    },
    {
      "epoch": 1.2896,
      "grad_norm": 0.6731203425584832,
      "learning_rate": 7.047000901883646e-06,
      "loss": 1.4114,
      "step": 806
    },
    {
      "epoch": 1.2912,
      "grad_norm": 0.683201486472504,
      "learning_rate": 7.038502255117957e-06,
      "loss": 1.4081,
      "step": 807
    },
    {
      "epoch": 1.2928,
      "grad_norm": 0.6989211967990084,
      "learning_rate": 7.029996538983273e-06,
      "loss": 1.2512,
      "step": 808
    },
    {
      "epoch": 1.2944,
      "grad_norm": 0.7078667836380207,
      "learning_rate": 7.021483782976759e-06,
      "loss": 1.4018,
      "step": 809
    },
    {
      "epoch": 1.296,
      "grad_norm": 0.7326582679131063,
      "learning_rate": 7.012964016620002e-06,
      "loss": 1.3422,
      "step": 810
    },
    {
      "epoch": 1.2976,
      "grad_norm": 0.7196567986805879,
      "learning_rate": 7.004437269458894e-06,
      "loss": 1.4022,
      "step": 811
    },
    {
      "epoch": 1.2992,
      "grad_norm": 0.720777667030854,
      "learning_rate": 6.995903571063541e-06,
      "loss": 1.3492,
      "step": 812
    },
    {
      "epoch": 1.3008,
      "grad_norm": 0.7229245741216352,
      "learning_rate": 6.987362951028147e-06,
      "loss": 1.3312,
      "step": 813
    },
    {
      "epoch": 1.3024,
      "grad_norm": 0.6596887490824602,
      "learning_rate": 6.97881543897093e-06,
      "loss": 1.3342,
      "step": 814
    },
    {
      "epoch": 1.304,
      "grad_norm": 0.6749167303012193,
      "learning_rate": 6.970261064534003e-06,
      "loss": 1.3065,
      "step": 815
    },
    {
      "epoch": 1.3056,
      "grad_norm": 0.701773755028264,
      "learning_rate": 6.961699857383279e-06,
      "loss": 1.3177,
      "step": 816
    },
    {
      "epoch": 1.3072,
      "grad_norm": 0.7136601604172251,
      "learning_rate": 6.953131847208365e-06,
      "loss": 1.3575,
      "step": 817
    },
    {
      "epoch": 1.3088,
      "grad_norm": 0.7407754442336435,
      "learning_rate": 6.944557063722459e-06,
      "loss": 1.3687,
      "step": 818
    },
    {
      "epoch": 1.3104,
      "grad_norm": 0.6368885409177069,
      "learning_rate": 6.935975536662254e-06,
      "loss": 1.2823,
      "step": 819
    },
    {
      "epoch": 1.312,
      "grad_norm": 0.7085743548260826,
      "learning_rate": 6.9273872957878255e-06,
      "loss": 1.2999,
      "step": 820
    },
    {
      "epoch": 1.3136,
      "grad_norm": 0.6429668567611934,
      "learning_rate": 6.91879237088253e-06,
      "loss": 1.3024,
      "step": 821
    },
    {
      "epoch": 1.3152,
      "grad_norm": 0.685158912155651,
      "learning_rate": 6.910190791752907e-06,
      "loss": 1.2347,
      "step": 822
    },
    {
      "epoch": 1.3168,
      "grad_norm": 0.714254031593221,
      "learning_rate": 6.90158258822857e-06,
      "loss": 1.3204,
      "step": 823
    },
    {
      "epoch": 1.3184,
      "grad_norm": 0.6905774362476343,
      "learning_rate": 6.892967790162109e-06,
      "loss": 1.3478,
      "step": 824
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.6536606136952269,
      "learning_rate": 6.884346427428978e-06,
      "loss": 1.3383,
      "step": 825
    },
    {
      "epoch": 1.3216,
      "grad_norm": 0.7178340971662062,
      "learning_rate": 6.875718529927404e-06,
      "loss": 1.358,
      "step": 826
    },
    {
      "epoch": 1.3232,
      "grad_norm": 0.6924609850095303,
      "learning_rate": 6.867084127578267e-06,
      "loss": 1.3187,
      "step": 827
    },
    {
      "epoch": 1.3248,
      "grad_norm": 0.7318363903392718,
      "learning_rate": 6.858443250325013e-06,
      "loss": 1.3343,
      "step": 828
    },
    {
      "epoch": 1.3264,
      "grad_norm": 0.6925932554955396,
      "learning_rate": 6.849795928133538e-06,
      "loss": 1.3795,
      "step": 829
    },
    {
      "epoch": 1.328,
      "grad_norm": 0.719621592096467,
      "learning_rate": 6.841142190992092e-06,
      "loss": 1.3503,
      "step": 830
    },
    {
      "epoch": 1.3296000000000001,
      "grad_norm": 0.7222159923066307,
      "learning_rate": 6.832482068911167e-06,
      "loss": 1.2579,
      "step": 831
    },
    {
      "epoch": 1.3312,
      "grad_norm": 0.7264940386558071,
      "learning_rate": 6.823815591923402e-06,
      "loss": 1.318,
      "step": 832
    },
    {
      "epoch": 1.3328,
      "grad_norm": 0.7357912324303968,
      "learning_rate": 6.815142790083473e-06,
      "loss": 1.312,
      "step": 833
    },
    {
      "epoch": 1.3344,
      "grad_norm": 0.7553664760625635,
      "learning_rate": 6.8064636934679885e-06,
      "loss": 1.3204,
      "step": 834
    },
    {
      "epoch": 1.336,
      "grad_norm": 0.6479506044459742,
      "learning_rate": 6.797778332175387e-06,
      "loss": 1.2814,
      "step": 835
    },
    {
      "epoch": 1.3376000000000001,
      "grad_norm": 0.7352690128653675,
      "learning_rate": 6.789086736325834e-06,
      "loss": 1.2725,
      "step": 836
    },
    {
      "epoch": 1.3392,
      "grad_norm": 0.6934240285493395,
      "learning_rate": 6.780388936061118e-06,
      "loss": 1.2827,
      "step": 837
    },
    {
      "epoch": 1.3408,
      "grad_norm": 0.7027283866875061,
      "learning_rate": 6.771684961544537e-06,
      "loss": 1.2939,
      "step": 838
    },
    {
      "epoch": 1.3424,
      "grad_norm": 0.806462834740327,
      "learning_rate": 6.7629748429608076e-06,
      "loss": 1.3249,
      "step": 839
    },
    {
      "epoch": 1.3439999999999999,
      "grad_norm": 13.636912704940148,
      "learning_rate": 6.754258610515949e-06,
      "loss": 1.356,
      "step": 840
    },
    {
      "epoch": 1.3456000000000001,
      "grad_norm": 0.8552736154252054,
      "learning_rate": 6.745536294437187e-06,
      "loss": 1.3879,
      "step": 841
    },
    {
      "epoch": 1.3472,
      "grad_norm": 0.6935294578684352,
      "learning_rate": 6.736807924972841e-06,
      "loss": 1.2573,
      "step": 842
    },
    {
      "epoch": 1.3488,
      "grad_norm": 0.784550856392547,
      "learning_rate": 6.728073532392226e-06,
      "loss": 1.3571,
      "step": 843
    },
    {
      "epoch": 1.3504,
      "grad_norm": 0.8098221295339241,
      "learning_rate": 6.719333146985544e-06,
      "loss": 1.3279,
      "step": 844
    },
    {
      "epoch": 1.3519999999999999,
      "grad_norm": 0.7763389262028246,
      "learning_rate": 6.710586799063777e-06,
      "loss": 1.3173,
      "step": 845
    },
    {
      "epoch": 1.3536000000000001,
      "grad_norm": 0.8074885642289481,
      "learning_rate": 6.701834518958587e-06,
      "loss": 1.3384,
      "step": 846
    },
    {
      "epoch": 1.3552,
      "grad_norm": 0.7744601733105952,
      "learning_rate": 6.6930763370222104e-06,
      "loss": 1.3031,
      "step": 847
    },
    {
      "epoch": 1.3568,
      "grad_norm": 0.8826581252779889,
      "learning_rate": 6.684312283627348e-06,
      "loss": 1.3939,
      "step": 848
    },
    {
      "epoch": 1.3584,
      "grad_norm": 0.7055894443534096,
      "learning_rate": 6.6755423891670605e-06,
      "loss": 1.2879,
      "step": 849
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 0.7059854957329889,
      "learning_rate": 6.6667666840546685e-06,
      "loss": 1.3613,
      "step": 850
    },
    {
      "epoch": 1.3616,
      "grad_norm": 0.7190809331186211,
      "learning_rate": 6.6579851987236435e-06,
      "loss": 1.3389,
      "step": 851
    },
    {
      "epoch": 1.3632,
      "grad_norm": 0.7910716301337495,
      "learning_rate": 6.649197963627497e-06,
      "loss": 1.3119,
      "step": 852
    },
    {
      "epoch": 1.3648,
      "grad_norm": 0.6712234878792241,
      "learning_rate": 6.640405009239689e-06,
      "loss": 1.3958,
      "step": 853
    },
    {
      "epoch": 1.3664,
      "grad_norm": 0.7162298339524514,
      "learning_rate": 6.631606366053507e-06,
      "loss": 1.3095,
      "step": 854
    },
    {
      "epoch": 1.3679999999999999,
      "grad_norm": 0.6684418732587324,
      "learning_rate": 6.622802064581968e-06,
      "loss": 1.3787,
      "step": 855
    },
    {
      "epoch": 1.3696,
      "grad_norm": 0.7734627859475829,
      "learning_rate": 6.613992135357713e-06,
      "loss": 1.3553,
      "step": 856
    },
    {
      "epoch": 1.3712,
      "grad_norm": 0.6975427463184489,
      "learning_rate": 6.605176608932897e-06,
      "loss": 1.2859,
      "step": 857
    },
    {
      "epoch": 1.3728,
      "grad_norm": 0.7636418376843181,
      "learning_rate": 6.596355515879091e-06,
      "loss": 1.3609,
      "step": 858
    },
    {
      "epoch": 1.3744,
      "grad_norm": 0.7081692903418492,
      "learning_rate": 6.587528886787165e-06,
      "loss": 1.32,
      "step": 859
    },
    {
      "epoch": 1.376,
      "grad_norm": 0.7114175414075518,
      "learning_rate": 6.578696752267189e-06,
      "loss": 1.3286,
      "step": 860
    },
    {
      "epoch": 1.3776,
      "grad_norm": 0.6759631865878251,
      "learning_rate": 6.5698591429483286e-06,
      "loss": 1.3232,
      "step": 861
    },
    {
      "epoch": 1.3792,
      "grad_norm": 0.7506720385807686,
      "learning_rate": 6.5610160894787275e-06,
      "loss": 1.2874,
      "step": 862
    },
    {
      "epoch": 1.3808,
      "grad_norm": 0.6516799654448546,
      "learning_rate": 6.552167622525421e-06,
      "loss": 1.3405,
      "step": 863
    },
    {
      "epoch": 1.3824,
      "grad_norm": 0.7857368993778627,
      "learning_rate": 6.543313772774209e-06,
      "loss": 1.3136,
      "step": 864
    },
    {
      "epoch": 1.384,
      "grad_norm": 0.6575670464397381,
      "learning_rate": 6.534454570929563e-06,
      "loss": 1.3445,
      "step": 865
    },
    {
      "epoch": 1.3856,
      "grad_norm": 0.7278378822815618,
      "learning_rate": 6.52559004771451e-06,
      "loss": 1.3463,
      "step": 866
    },
    {
      "epoch": 1.3872,
      "grad_norm": 0.6799136019300637,
      "learning_rate": 6.516720233870538e-06,
      "loss": 1.3414,
      "step": 867
    },
    {
      "epoch": 1.3888,
      "grad_norm": 0.6890577149546877,
      "learning_rate": 6.507845160157476e-06,
      "loss": 1.391,
      "step": 868
    },
    {
      "epoch": 1.3904,
      "grad_norm": 0.6656809331851256,
      "learning_rate": 6.498964857353401e-06,
      "loss": 1.3522,
      "step": 869
    },
    {
      "epoch": 1.392,
      "grad_norm": 0.7284615708828512,
      "learning_rate": 6.4900793562545165e-06,
      "loss": 1.3611,
      "step": 870
    },
    {
      "epoch": 1.3936,
      "grad_norm": 0.6312689874917573,
      "learning_rate": 6.481188687675057e-06,
      "loss": 1.2954,
      "step": 871
    },
    {
      "epoch": 1.3952,
      "grad_norm": 0.7348936589657888,
      "learning_rate": 6.47229288244718e-06,
      "loss": 1.3216,
      "step": 872
    },
    {
      "epoch": 1.3968,
      "grad_norm": 0.6951222813786351,
      "learning_rate": 6.46339197142085e-06,
      "loss": 1.3642,
      "step": 873
    },
    {
      "epoch": 1.3984,
      "grad_norm": 0.6941393810338808,
      "learning_rate": 6.454485985463742e-06,
      "loss": 1.3647,
      "step": 874
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.6847824607374816,
      "learning_rate": 6.445574955461134e-06,
      "loss": 1.3446,
      "step": 875
    },
    {
      "epoch": 1.4016,
      "grad_norm": 0.7076695542477842,
      "learning_rate": 6.436658912315789e-06,
      "loss": 1.3764,
      "step": 876
    },
    {
      "epoch": 1.4032,
      "grad_norm": 0.6721453621572587,
      "learning_rate": 6.427737886947859e-06,
      "loss": 1.3502,
      "step": 877
    },
    {
      "epoch": 1.4048,
      "grad_norm": 0.642149531762323,
      "learning_rate": 6.418811910294776e-06,
      "loss": 1.3514,
      "step": 878
    },
    {
      "epoch": 1.4064,
      "grad_norm": 0.696793725370131,
      "learning_rate": 6.409881013311136e-06,
      "loss": 1.3108,
      "step": 879
    },
    {
      "epoch": 1.408,
      "grad_norm": 0.6901302554126103,
      "learning_rate": 6.400945226968607e-06,
      "loss": 1.2959,
      "step": 880
    },
    {
      "epoch": 1.4096,
      "grad_norm": 0.7585811302698048,
      "learning_rate": 6.392004582255807e-06,
      "loss": 1.3586,
      "step": 881
    },
    {
      "epoch": 1.4112,
      "grad_norm": 0.7078910758662755,
      "learning_rate": 6.383059110178205e-06,
      "loss": 1.3267,
      "step": 882
    },
    {
      "epoch": 1.4128,
      "grad_norm": 0.6946358163699178,
      "learning_rate": 6.374108841758006e-06,
      "loss": 1.3039,
      "step": 883
    },
    {
      "epoch": 1.4144,
      "grad_norm": 0.7714889342703183,
      "learning_rate": 6.365153808034057e-06,
      "loss": 1.3391,
      "step": 884
    },
    {
      "epoch": 1.416,
      "grad_norm": 0.7408737883919189,
      "learning_rate": 6.356194040061725e-06,
      "loss": 1.337,
      "step": 885
    },
    {
      "epoch": 1.4176,
      "grad_norm": 0.6904742085843393,
      "learning_rate": 6.3472295689127946e-06,
      "loss": 1.2925,
      "step": 886
    },
    {
      "epoch": 1.4192,
      "grad_norm": 0.6920467541517021,
      "learning_rate": 6.338260425675365e-06,
      "loss": 1.2445,
      "step": 887
    },
    {
      "epoch": 1.4208,
      "grad_norm": 0.749825929546616,
      "learning_rate": 6.329286641453729e-06,
      "loss": 1.3523,
      "step": 888
    },
    {
      "epoch": 1.4224,
      "grad_norm": 0.6651275260312463,
      "learning_rate": 6.320308247368285e-06,
      "loss": 1.3326,
      "step": 889
    },
    {
      "epoch": 1.424,
      "grad_norm": 0.8375367542697988,
      "learning_rate": 6.311325274555413e-06,
      "loss": 1.3147,
      "step": 890
    },
    {
      "epoch": 1.4256,
      "grad_norm": 0.6678257181187496,
      "learning_rate": 6.302337754167369e-06,
      "loss": 1.3672,
      "step": 891
    },
    {
      "epoch": 1.4272,
      "grad_norm": 0.8476325476659016,
      "learning_rate": 6.2933457173721855e-06,
      "loss": 1.3733,
      "step": 892
    },
    {
      "epoch": 1.4288,
      "grad_norm": 0.7113441201102606,
      "learning_rate": 6.2843491953535515e-06,
      "loss": 1.3459,
      "step": 893
    },
    {
      "epoch": 1.4304000000000001,
      "grad_norm": 0.7111552791101371,
      "learning_rate": 6.275348219310715e-06,
      "loss": 1.3044,
      "step": 894
    },
    {
      "epoch": 1.432,
      "grad_norm": 0.6927697599108797,
      "learning_rate": 6.266342820458366e-06,
      "loss": 1.3594,
      "step": 895
    },
    {
      "epoch": 1.4336,
      "grad_norm": 0.8341027195436078,
      "learning_rate": 6.2573330300265375e-06,
      "loss": 1.3492,
      "step": 896
    },
    {
      "epoch": 1.4352,
      "grad_norm": 0.7751199663517662,
      "learning_rate": 6.248318879260488e-06,
      "loss": 1.3903,
      "step": 897
    },
    {
      "epoch": 1.4368,
      "grad_norm": 0.6770318389478522,
      "learning_rate": 6.239300399420601e-06,
      "loss": 1.3093,
      "step": 898
    },
    {
      "epoch": 1.4384000000000001,
      "grad_norm": 0.7758456722865342,
      "learning_rate": 6.230277621782269e-06,
      "loss": 1.3376,
      "step": 899
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7295095680816364,
      "learning_rate": 6.221250577635791e-06,
      "loss": 1.3196,
      "step": 900
    },
    {
      "epoch": 1.4416,
      "grad_norm": 0.7185292330612116,
      "learning_rate": 6.2122192982862615e-06,
      "loss": 1.4081,
      "step": 901
    },
    {
      "epoch": 1.4432,
      "grad_norm": 0.7134905935849686,
      "learning_rate": 6.203183815053463e-06,
      "loss": 1.3077,
      "step": 902
    },
    {
      "epoch": 1.4447999999999999,
      "grad_norm": 0.7176324396977447,
      "learning_rate": 6.1941441592717564e-06,
      "loss": 1.3296,
      "step": 903
    },
    {
      "epoch": 1.4464000000000001,
      "grad_norm": 0.7233624953206502,
      "learning_rate": 6.185100362289972e-06,
      "loss": 1.3981,
      "step": 904
    },
    {
      "epoch": 1.448,
      "grad_norm": 0.7274610273392916,
      "learning_rate": 6.176052455471302e-06,
      "loss": 1.3049,
      "step": 905
    },
    {
      "epoch": 1.4496,
      "grad_norm": 0.7148130189061632,
      "learning_rate": 6.167000470193189e-06,
      "loss": 1.3054,
      "step": 906
    },
    {
      "epoch": 1.4512,
      "grad_norm": 0.7901173661631029,
      "learning_rate": 6.157944437847226e-06,
      "loss": 1.285,
      "step": 907
    },
    {
      "epoch": 1.4527999999999999,
      "grad_norm": 0.7387489114871357,
      "learning_rate": 6.148884389839035e-06,
      "loss": 1.3408,
      "step": 908
    },
    {
      "epoch": 1.4544000000000001,
      "grad_norm": 0.7558057761440851,
      "learning_rate": 6.1398203575881645e-06,
      "loss": 1.2942,
      "step": 909
    },
    {
      "epoch": 1.456,
      "grad_norm": 0.7588792178682182,
      "learning_rate": 6.130752372527981e-06,
      "loss": 1.3358,
      "step": 910
    },
    {
      "epoch": 1.4576,
      "grad_norm": 0.7204111081985519,
      "learning_rate": 6.121680466105559e-06,
      "loss": 1.3185,
      "step": 911
    },
    {
      "epoch": 1.4592,
      "grad_norm": 0.7778871153537111,
      "learning_rate": 6.112604669781572e-06,
      "loss": 1.3282,
      "step": 912
    },
    {
      "epoch": 1.4607999999999999,
      "grad_norm": 0.7681143653229145,
      "learning_rate": 6.1035250150301864e-06,
      "loss": 1.3062,
      "step": 913
    },
    {
      "epoch": 1.4624,
      "grad_norm": 0.7713744306707856,
      "learning_rate": 6.0944415333389405e-06,
      "loss": 1.4001,
      "step": 914
    },
    {
      "epoch": 1.464,
      "grad_norm": 0.7282823810716962,
      "learning_rate": 6.085354256208655e-06,
      "loss": 1.3046,
      "step": 915
    },
    {
      "epoch": 1.4656,
      "grad_norm": 0.771345539583643,
      "learning_rate": 6.076263215153308e-06,
      "loss": 1.3453,
      "step": 916
    },
    {
      "epoch": 1.4672,
      "grad_norm": 0.7022662776606637,
      "learning_rate": 6.067168441699927e-06,
      "loss": 1.2911,
      "step": 917
    },
    {
      "epoch": 1.4687999999999999,
      "grad_norm": 0.7448858648637333,
      "learning_rate": 6.058069967388489e-06,
      "loss": 1.2603,
      "step": 918
    },
    {
      "epoch": 1.4704,
      "grad_norm": 0.7129212853031672,
      "learning_rate": 6.048967823771802e-06,
      "loss": 1.2828,
      "step": 919
    },
    {
      "epoch": 1.472,
      "grad_norm": 0.674879963197364,
      "learning_rate": 6.039862042415401e-06,
      "loss": 1.3592,
      "step": 920
    },
    {
      "epoch": 1.4736,
      "grad_norm": 0.8453315739114547,
      "learning_rate": 6.030752654897435e-06,
      "loss": 1.3915,
      "step": 921
    },
    {
      "epoch": 1.4752,
      "grad_norm": 0.7244279115073962,
      "learning_rate": 6.021639692808558e-06,
      "loss": 1.2473,
      "step": 922
    },
    {
      "epoch": 1.4768,
      "grad_norm": 0.6710246828792319,
      "learning_rate": 6.0125231877518205e-06,
      "loss": 1.3001,
      "step": 923
    },
    {
      "epoch": 1.4784,
      "grad_norm": 0.7764481964436588,
      "learning_rate": 6.0034031713425636e-06,
      "loss": 1.3442,
      "step": 924
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.6668709078441614,
      "learning_rate": 5.994279675208302e-06,
      "loss": 1.3241,
      "step": 925
    },
    {
      "epoch": 1.4816,
      "grad_norm": 0.7245179292213216,
      "learning_rate": 5.985152730988617e-06,
      "loss": 1.3442,
      "step": 926
    },
    {
      "epoch": 1.4832,
      "grad_norm": 0.7848097983481358,
      "learning_rate": 5.9760223703350495e-06,
      "loss": 1.3369,
      "step": 927
    },
    {
      "epoch": 1.4848,
      "grad_norm": 0.7263010108224064,
      "learning_rate": 5.966888624910989e-06,
      "loss": 1.2867,
      "step": 928
    },
    {
      "epoch": 1.4864,
      "grad_norm": 0.6542346789991754,
      "learning_rate": 5.957751526391558e-06,
      "loss": 1.2437,
      "step": 929
    },
    {
      "epoch": 1.488,
      "grad_norm": 0.7467339881715739,
      "learning_rate": 5.948611106463518e-06,
      "loss": 1.2559,
      "step": 930
    },
    {
      "epoch": 1.4896,
      "grad_norm": 0.680127050499239,
      "learning_rate": 5.939467396825137e-06,
      "loss": 1.332,
      "step": 931
    },
    {
      "epoch": 1.4912,
      "grad_norm": 0.7018793487304368,
      "learning_rate": 5.9303204291860975e-06,
      "loss": 1.3051,
      "step": 932
    },
    {
      "epoch": 1.4928,
      "grad_norm": 0.7743606652563313,
      "learning_rate": 5.92117023526738e-06,
      "loss": 1.3249,
      "step": 933
    },
    {
      "epoch": 1.4944,
      "grad_norm": 0.6681133803508406,
      "learning_rate": 5.912016846801153e-06,
      "loss": 1.3692,
      "step": 934
    },
    {
      "epoch": 1.496,
      "grad_norm": 0.7509320805789341,
      "learning_rate": 5.902860295530665e-06,
      "loss": 1.2882,
      "step": 935
    },
    {
      "epoch": 1.4976,
      "grad_norm": 0.6887965636964528,
      "learning_rate": 5.893700613210128e-06,
      "loss": 1.281,
      "step": 936
    },
    {
      "epoch": 1.4992,
      "grad_norm": 0.6503656544447964,
      "learning_rate": 5.88453783160462e-06,
      "loss": 1.325,
      "step": 937
    },
    {
      "epoch": 1.5008,
      "grad_norm": 0.7348585895825063,
      "learning_rate": 5.875371982489959e-06,
      "loss": 1.335,
      "step": 938
    },
    {
      "epoch": 1.5024,
      "grad_norm": 0.7172621876246941,
      "learning_rate": 5.866203097652605e-06,
      "loss": 1.3153,
      "step": 939
    },
    {
      "epoch": 1.504,
      "grad_norm": 0.6769678850593607,
      "learning_rate": 5.857031208889548e-06,
      "loss": 1.4055,
      "step": 940
    },
    {
      "epoch": 1.5056,
      "grad_norm": 0.8018999028376177,
      "learning_rate": 5.847856348008188e-06,
      "loss": 1.3679,
      "step": 941
    },
    {
      "epoch": 1.5072,
      "grad_norm": 0.6948629121134091,
      "learning_rate": 5.838678546826242e-06,
      "loss": 1.3281,
      "step": 942
    },
    {
      "epoch": 1.5088,
      "grad_norm": 0.740780057042295,
      "learning_rate": 5.829497837171616e-06,
      "loss": 1.3662,
      "step": 943
    },
    {
      "epoch": 1.5104,
      "grad_norm": 0.6935742896884737,
      "learning_rate": 5.820314250882304e-06,
      "loss": 1.3416,
      "step": 944
    },
    {
      "epoch": 1.512,
      "grad_norm": 0.6841784826525678,
      "learning_rate": 5.811127819806277e-06,
      "loss": 1.252,
      "step": 945
    },
    {
      "epoch": 1.5135999999999998,
      "grad_norm": 0.7895724506829804,
      "learning_rate": 5.801938575801372e-06,
      "loss": 1.3954,
      "step": 946
    },
    {
      "epoch": 1.5152,
      "grad_norm": 0.6701910942440477,
      "learning_rate": 5.792746550735182e-06,
      "loss": 1.3852,
      "step": 947
    },
    {
      "epoch": 1.5168,
      "grad_norm": 0.7376868989764289,
      "learning_rate": 5.7835517764849395e-06,
      "loss": 1.4055,
      "step": 948
    },
    {
      "epoch": 1.5184,
      "grad_norm": 0.6487968997874293,
      "learning_rate": 5.7743542849374155e-06,
      "loss": 1.3798,
      "step": 949
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7016398676794361,
      "learning_rate": 5.765154107988803e-06,
      "loss": 1.3898,
      "step": 950
    },
    {
      "epoch": 1.5215999999999998,
      "grad_norm": 0.6558587270930889,
      "learning_rate": 5.755951277544607e-06,
      "loss": 1.3519,
      "step": 951
    },
    {
      "epoch": 1.5232,
      "grad_norm": 0.7395057561112114,
      "learning_rate": 5.746745825519539e-06,
      "loss": 1.4152,
      "step": 952
    },
    {
      "epoch": 1.5248,
      "grad_norm": 0.6797600382229331,
      "learning_rate": 5.737537783837395e-06,
      "loss": 1.3862,
      "step": 953
    },
    {
      "epoch": 1.5264,
      "grad_norm": 0.7368460404121541,
      "learning_rate": 5.728327184430955e-06,
      "loss": 1.3696,
      "step": 954
    },
    {
      "epoch": 1.528,
      "grad_norm": 0.6967971755755309,
      "learning_rate": 5.719114059241871e-06,
      "loss": 1.3499,
      "step": 955
    },
    {
      "epoch": 1.5295999999999998,
      "grad_norm": 0.6787468307903893,
      "learning_rate": 5.709898440220552e-06,
      "loss": 1.285,
      "step": 956
    },
    {
      "epoch": 1.5312000000000001,
      "grad_norm": 0.7755051190324879,
      "learning_rate": 5.700680359326055e-06,
      "loss": 1.3684,
      "step": 957
    },
    {
      "epoch": 1.5328,
      "grad_norm": 0.7477081412773476,
      "learning_rate": 5.691459848525977e-06,
      "loss": 1.3203,
      "step": 958
    },
    {
      "epoch": 1.5344,
      "grad_norm": 0.7835462590812435,
      "learning_rate": 5.682236939796337e-06,
      "loss": 1.3394,
      "step": 959
    },
    {
      "epoch": 1.536,
      "grad_norm": 0.7290448051158535,
      "learning_rate": 5.673011665121477e-06,
      "loss": 1.3161,
      "step": 960
    },
    {
      "epoch": 1.5375999999999999,
      "grad_norm": 0.7006741114158486,
      "learning_rate": 5.663784056493936e-06,
      "loss": 1.35,
      "step": 961
    },
    {
      "epoch": 1.5392000000000001,
      "grad_norm": 0.7143865943325339,
      "learning_rate": 5.6545541459143535e-06,
      "loss": 1.335,
      "step": 962
    },
    {
      "epoch": 1.5408,
      "grad_norm": 0.7269911951745034,
      "learning_rate": 5.6453219653913495e-06,
      "loss": 1.3528,
      "step": 963
    },
    {
      "epoch": 1.5424,
      "grad_norm": 0.729909423022284,
      "learning_rate": 5.636087546941413e-06,
      "loss": 1.374,
      "step": 964
    },
    {
      "epoch": 1.544,
      "grad_norm": 0.737137677001826,
      "learning_rate": 5.6268509225888005e-06,
      "loss": 1.2972,
      "step": 965
    },
    {
      "epoch": 1.5455999999999999,
      "grad_norm": 0.6716770689251331,
      "learning_rate": 5.617612124365411e-06,
      "loss": 1.3214,
      "step": 966
    },
    {
      "epoch": 1.5472000000000001,
      "grad_norm": 0.7295792686490288,
      "learning_rate": 5.608371184310688e-06,
      "loss": 1.2918,
      "step": 967
    },
    {
      "epoch": 1.5488,
      "grad_norm": 0.6434759061172859,
      "learning_rate": 5.5991281344714984e-06,
      "loss": 1.3388,
      "step": 968
    },
    {
      "epoch": 1.5504,
      "grad_norm": 0.629371764305478,
      "learning_rate": 5.5898830069020325e-06,
      "loss": 1.3359,
      "step": 969
    },
    {
      "epoch": 1.552,
      "grad_norm": 0.6962053894397091,
      "learning_rate": 5.580635833663679e-06,
      "loss": 1.3108,
      "step": 970
    },
    {
      "epoch": 1.5535999999999999,
      "grad_norm": 0.670789525237759,
      "learning_rate": 5.5713866468249235e-06,
      "loss": 1.3127,
      "step": 971
    },
    {
      "epoch": 1.5552000000000001,
      "grad_norm": 0.6645144038994726,
      "learning_rate": 5.562135478461234e-06,
      "loss": 1.3306,
      "step": 972
    },
    {
      "epoch": 1.5568,
      "grad_norm": 1.5391234435643324,
      "learning_rate": 5.55288236065495e-06,
      "loss": 1.3079,
      "step": 973
    },
    {
      "epoch": 1.5584,
      "grad_norm": 0.6900638055678654,
      "learning_rate": 5.5436273254951734e-06,
      "loss": 1.2784,
      "step": 974
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.6449562414353665,
      "learning_rate": 5.5343704050776535e-06,
      "loss": 1.3077,
      "step": 975
    },
    {
      "epoch": 1.5615999999999999,
      "grad_norm": 0.6476185898072332,
      "learning_rate": 5.5251116315046785e-06,
      "loss": 1.3839,
      "step": 976
    },
    {
      "epoch": 1.5632000000000001,
      "grad_norm": 0.7076137896596729,
      "learning_rate": 5.515851036884964e-06,
      "loss": 1.3421,
      "step": 977
    },
    {
      "epoch": 1.5648,
      "grad_norm": 0.6841634852821264,
      "learning_rate": 5.5065886533335355e-06,
      "loss": 1.3328,
      "step": 978
    },
    {
      "epoch": 1.5664,
      "grad_norm": 0.6554668853677373,
      "learning_rate": 5.497324512971632e-06,
      "loss": 1.309,
      "step": 979
    },
    {
      "epoch": 1.568,
      "grad_norm": 0.6471339962911499,
      "learning_rate": 5.4880586479265774e-06,
      "loss": 1.25,
      "step": 980
    },
    {
      "epoch": 1.5695999999999999,
      "grad_norm": 0.6492692412580385,
      "learning_rate": 5.478791090331677e-06,
      "loss": 1.3326,
      "step": 981
    },
    {
      "epoch": 1.5712000000000002,
      "grad_norm": 0.6844015364517265,
      "learning_rate": 5.4695218723261115e-06,
      "loss": 1.3636,
      "step": 982
    },
    {
      "epoch": 1.5728,
      "grad_norm": 0.6656709476344331,
      "learning_rate": 5.46025102605481e-06,
      "loss": 1.3809,
      "step": 983
    },
    {
      "epoch": 1.5744,
      "grad_norm": 0.6610761086537226,
      "learning_rate": 5.4509785836683606e-06,
      "loss": 1.3135,
      "step": 984
    },
    {
      "epoch": 1.576,
      "grad_norm": 0.6435616878562068,
      "learning_rate": 5.441704577322877e-06,
      "loss": 1.3204,
      "step": 985
    },
    {
      "epoch": 1.5776,
      "grad_norm": 0.6837641107280251,
      "learning_rate": 5.4324290391798995e-06,
      "loss": 1.28,
      "step": 986
    },
    {
      "epoch": 1.5792000000000002,
      "grad_norm": 0.6726183092470666,
      "learning_rate": 5.423152001406282e-06,
      "loss": 1.3363,
      "step": 987
    },
    {
      "epoch": 1.5808,
      "grad_norm": 0.6462398654211154,
      "learning_rate": 5.413873496174077e-06,
      "loss": 1.2621,
      "step": 988
    },
    {
      "epoch": 1.5824,
      "grad_norm": 0.6836270321301225,
      "learning_rate": 5.404593555660424e-06,
      "loss": 1.3049,
      "step": 989
    },
    {
      "epoch": 1.584,
      "grad_norm": 0.6596917501552024,
      "learning_rate": 5.39531221204745e-06,
      "loss": 1.327,
      "step": 990
    },
    {
      "epoch": 1.5856,
      "grad_norm": 0.6851466736853775,
      "learning_rate": 5.3860294975221335e-06,
      "loss": 1.3016,
      "step": 991
    },
    {
      "epoch": 1.5872000000000002,
      "grad_norm": 0.7060571846121407,
      "learning_rate": 5.376745444276219e-06,
      "loss": 1.3935,
      "step": 992
    },
    {
      "epoch": 1.5888,
      "grad_norm": 0.7078995136976644,
      "learning_rate": 5.3674600845060856e-06,
      "loss": 1.3087,
      "step": 993
    },
    {
      "epoch": 1.5904,
      "grad_norm": 0.6490403868644476,
      "learning_rate": 5.358173450412649e-06,
      "loss": 1.3372,
      "step": 994
    },
    {
      "epoch": 1.592,
      "grad_norm": 0.7398129647070868,
      "learning_rate": 5.34888557420124e-06,
      "loss": 1.3776,
      "step": 995
    },
    {
      "epoch": 1.5936,
      "grad_norm": 0.6454021048955672,
      "learning_rate": 5.339596488081501e-06,
      "loss": 1.355,
      "step": 996
    },
    {
      "epoch": 1.5952,
      "grad_norm": 0.8950095391106841,
      "learning_rate": 5.330306224267268e-06,
      "loss": 1.3811,
      "step": 997
    },
    {
      "epoch": 1.5968,
      "grad_norm": 0.7003017668880928,
      "learning_rate": 5.321014814976459e-06,
      "loss": 1.377,
      "step": 998
    },
    {
      "epoch": 1.5984,
      "grad_norm": 0.6416789919464462,
      "learning_rate": 5.311722292430966e-06,
      "loss": 1.2461,
      "step": 999
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.6780035805445138,
      "learning_rate": 5.302428688856544e-06,
      "loss": 1.2604,
      "step": 1000
    },
    {
      "epoch": 1.6016,
      "grad_norm": 0.6738471936101198,
      "learning_rate": 5.293134036482697e-06,
      "loss": 1.3057,
      "step": 1001
    },
    {
      "epoch": 1.6032,
      "grad_norm": 0.7182871763205134,
      "learning_rate": 5.283838367542562e-06,
      "loss": 1.3087,
      "step": 1002
    },
    {
      "epoch": 1.6048,
      "grad_norm": 0.6600177230510398,
      "learning_rate": 5.274541714272805e-06,
      "loss": 1.3308,
      "step": 1003
    },
    {
      "epoch": 1.6064,
      "grad_norm": 0.7137026902177019,
      "learning_rate": 5.265244108913503e-06,
      "loss": 1.355,
      "step": 1004
    },
    {
      "epoch": 1.608,
      "grad_norm": 0.6746349471069771,
      "learning_rate": 5.255945583708037e-06,
      "loss": 1.3881,
      "step": 1005
    },
    {
      "epoch": 1.6096,
      "grad_norm": 0.7235567545790804,
      "learning_rate": 5.2466461709029755e-06,
      "loss": 1.3531,
      "step": 1006
    },
    {
      "epoch": 1.6112,
      "grad_norm": 0.6466103253901367,
      "learning_rate": 5.237345902747969e-06,
      "loss": 1.2727,
      "step": 1007
    },
    {
      "epoch": 1.6128,
      "grad_norm": 0.7083184523817438,
      "learning_rate": 5.228044811495632e-06,
      "loss": 1.3941,
      "step": 1008
    },
    {
      "epoch": 1.6143999999999998,
      "grad_norm": 0.6883205679827926,
      "learning_rate": 5.218742929401432e-06,
      "loss": 1.3024,
      "step": 1009
    },
    {
      "epoch": 1.616,
      "grad_norm": 0.682904825753002,
      "learning_rate": 5.2094402887235805e-06,
      "loss": 1.3212,
      "step": 1010
    },
    {
      "epoch": 1.6176,
      "grad_norm": 0.6907862195922169,
      "learning_rate": 5.200136921722919e-06,
      "loss": 1.2902,
      "step": 1011
    },
    {
      "epoch": 1.6192,
      "grad_norm": 0.6996489738717504,
      "learning_rate": 5.1908328606628114e-06,
      "loss": 1.3387,
      "step": 1012
    },
    {
      "epoch": 1.6208,
      "grad_norm": 0.6372917921910929,
      "learning_rate": 5.181528137809023e-06,
      "loss": 1.3372,
      "step": 1013
    },
    {
      "epoch": 1.6223999999999998,
      "grad_norm": 0.6433137191104882,
      "learning_rate": 5.1722227854296195e-06,
      "loss": 1.3437,
      "step": 1014
    },
    {
      "epoch": 1.624,
      "grad_norm": 0.6484516687091758,
      "learning_rate": 5.162916835794843e-06,
      "loss": 1.3576,
      "step": 1015
    },
    {
      "epoch": 1.6256,
      "grad_norm": 0.6169243047358949,
      "learning_rate": 5.1536103211770135e-06,
      "loss": 1.2985,
      "step": 1016
    },
    {
      "epoch": 1.6272,
      "grad_norm": 0.6563304183182529,
      "learning_rate": 5.14430327385041e-06,
      "loss": 1.3343,
      "step": 1017
    },
    {
      "epoch": 1.6288,
      "grad_norm": 0.6764479078017436,
      "learning_rate": 5.134995726091152e-06,
      "loss": 1.3639,
      "step": 1018
    },
    {
      "epoch": 1.6303999999999998,
      "grad_norm": 0.6539592759248707,
      "learning_rate": 5.1256877101771015e-06,
      "loss": 1.2945,
      "step": 1019
    },
    {
      "epoch": 1.6320000000000001,
      "grad_norm": 0.6197864129351625,
      "learning_rate": 5.116379258387742e-06,
      "loss": 1.3042,
      "step": 1020
    },
    {
      "epoch": 1.6336,
      "grad_norm": 0.6479291908676431,
      "learning_rate": 5.1070704030040675e-06,
      "loss": 1.3077,
      "step": 1021
    },
    {
      "epoch": 1.6352,
      "grad_norm": 0.6385618899045838,
      "learning_rate": 5.097761176308471e-06,
      "loss": 1.3057,
      "step": 1022
    },
    {
      "epoch": 1.6368,
      "grad_norm": 0.6546390312104393,
      "learning_rate": 5.088451610584638e-06,
      "loss": 1.3114,
      "step": 1023
    },
    {
      "epoch": 1.6383999999999999,
      "grad_norm": 0.6097482218376914,
      "learning_rate": 5.079141738117423e-06,
      "loss": 1.2617,
      "step": 1024
    },
    {
      "epoch": 1.6400000000000001,
      "grad_norm": 0.6734636828572604,
      "learning_rate": 5.06983159119275e-06,
      "loss": 1.3641,
      "step": 1025
    },
    {
      "epoch": 1.6416,
      "grad_norm": 0.6279900180348446,
      "learning_rate": 5.060521202097491e-06,
      "loss": 1.2851,
      "step": 1026
    },
    {
      "epoch": 1.6432,
      "grad_norm": 0.649206340359319,
      "learning_rate": 5.051210603119358e-06,
      "loss": 1.3035,
      "step": 1027
    },
    {
      "epoch": 1.6448,
      "grad_norm": 0.6490333452076913,
      "learning_rate": 5.041899826546791e-06,
      "loss": 1.3726,
      "step": 1028
    },
    {
      "epoch": 1.6463999999999999,
      "grad_norm": 0.6416496158226107,
      "learning_rate": 5.032588904668851e-06,
      "loss": 1.3225,
      "step": 1029
    },
    {
      "epoch": 1.6480000000000001,
      "grad_norm": 0.7167274611777741,
      "learning_rate": 5.023277869775097e-06,
      "loss": 1.3639,
      "step": 1030
    },
    {
      "epoch": 1.6496,
      "grad_norm": 0.618727684796022,
      "learning_rate": 5.013966754155482e-06,
      "loss": 1.3185,
      "step": 1031
    },
    {
      "epoch": 1.6512,
      "grad_norm": 0.6841829839769985,
      "learning_rate": 5.004655590100238e-06,
      "loss": 1.3808,
      "step": 1032
    },
    {
      "epoch": 1.6528,
      "grad_norm": 0.680680733970545,
      "learning_rate": 4.995344409899764e-06,
      "loss": 1.3052,
      "step": 1033
    },
    {
      "epoch": 1.6543999999999999,
      "grad_norm": 0.6063536194516561,
      "learning_rate": 4.986033245844519e-06,
      "loss": 1.2818,
      "step": 1034
    },
    {
      "epoch": 1.6560000000000001,
      "grad_norm": 0.6591831585669239,
      "learning_rate": 4.976722130224904e-06,
      "loss": 1.3504,
      "step": 1035
    },
    {
      "epoch": 1.6576,
      "grad_norm": 0.6362563402013064,
      "learning_rate": 4.967411095331149e-06,
      "loss": 1.3578,
      "step": 1036
    },
    {
      "epoch": 1.6592,
      "grad_norm": 0.7424648454002032,
      "learning_rate": 4.95810017345321e-06,
      "loss": 1.3412,
      "step": 1037
    },
    {
      "epoch": 1.6608,
      "grad_norm": 0.6580462730189933,
      "learning_rate": 4.948789396880644e-06,
      "loss": 1.3048,
      "step": 1038
    },
    {
      "epoch": 1.6623999999999999,
      "grad_norm": 0.7118539525030807,
      "learning_rate": 4.939478797902512e-06,
      "loss": 1.2995,
      "step": 1039
    },
    {
      "epoch": 1.6640000000000001,
      "grad_norm": 0.7040740474755023,
      "learning_rate": 4.930168408807252e-06,
      "loss": 1.4057,
      "step": 1040
    },
    {
      "epoch": 1.6656,
      "grad_norm": 0.657348326911951,
      "learning_rate": 4.920858261882578e-06,
      "loss": 1.2696,
      "step": 1041
    },
    {
      "epoch": 1.6672,
      "grad_norm": 0.6532360509520155,
      "learning_rate": 4.911548389415363e-06,
      "loss": 1.3146,
      "step": 1042
    },
    {
      "epoch": 1.6688,
      "grad_norm": 0.6746679805669085,
      "learning_rate": 4.9022388236915306e-06,
      "loss": 1.2571,
      "step": 1043
    },
    {
      "epoch": 1.6703999999999999,
      "grad_norm": 0.6547899683393446,
      "learning_rate": 4.892929596995934e-06,
      "loss": 1.3941,
      "step": 1044
    },
    {
      "epoch": 1.6720000000000002,
      "grad_norm": 0.6958074387530667,
      "learning_rate": 4.883620741612259e-06,
      "loss": 1.2915,
      "step": 1045
    },
    {
      "epoch": 1.6736,
      "grad_norm": 0.6297878286910985,
      "learning_rate": 4.8743122898229e-06,
      "loss": 1.2969,
      "step": 1046
    },
    {
      "epoch": 1.6752,
      "grad_norm": 0.6970411721958757,
      "learning_rate": 4.865004273908851e-06,
      "loss": 1.2784,
      "step": 1047
    },
    {
      "epoch": 1.6768,
      "grad_norm": 0.6846313436171685,
      "learning_rate": 4.855696726149593e-06,
      "loss": 1.3419,
      "step": 1048
    },
    {
      "epoch": 1.6784,
      "grad_norm": 0.6337024230878959,
      "learning_rate": 4.846389678822987e-06,
      "loss": 1.3726,
      "step": 1049
    },
    {
      "epoch": 1.6800000000000002,
      "grad_norm": 0.6138030818374821,
      "learning_rate": 4.837083164205159e-06,
      "loss": 1.3235,
      "step": 1050
    },
    {
      "epoch": 1.6816,
      "grad_norm": 13.383234772931994,
      "learning_rate": 4.827777214570384e-06,
      "loss": 1.3931,
      "step": 1051
    },
    {
      "epoch": 1.6832,
      "grad_norm": 0.7581099495703584,
      "learning_rate": 4.818471862190979e-06,
      "loss": 1.2815,
      "step": 1052
    },
    {
      "epoch": 1.6848,
      "grad_norm": 0.6627743352368185,
      "learning_rate": 4.809167139337191e-06,
      "loss": 1.2875,
      "step": 1053
    },
    {
      "epoch": 1.6864,
      "grad_norm": 0.664673826918059,
      "learning_rate": 4.799863078277082e-06,
      "loss": 1.3269,
      "step": 1054
    },
    {
      "epoch": 1.688,
      "grad_norm": 0.6940645845271229,
      "learning_rate": 4.790559711276422e-06,
      "loss": 1.2764,
      "step": 1055
    },
    {
      "epoch": 1.6896,
      "grad_norm": 0.6426114617461507,
      "learning_rate": 4.781257070598571e-06,
      "loss": 1.3574,
      "step": 1056
    },
    {
      "epoch": 1.6912,
      "grad_norm": 0.6772670624042596,
      "learning_rate": 4.771955188504371e-06,
      "loss": 1.3012,
      "step": 1057
    },
    {
      "epoch": 1.6928,
      "grad_norm": 0.697794550659188,
      "learning_rate": 4.762654097252033e-06,
      "loss": 1.3503,
      "step": 1058
    },
    {
      "epoch": 1.6944,
      "grad_norm": 0.6308245140461917,
      "learning_rate": 4.753353829097025e-06,
      "loss": 1.3623,
      "step": 1059
    },
    {
      "epoch": 1.696,
      "grad_norm": 0.6487411726608782,
      "learning_rate": 4.7440544162919645e-06,
      "loss": 1.3595,
      "step": 1060
    },
    {
      "epoch": 1.6976,
      "grad_norm": 0.6753969105862622,
      "learning_rate": 4.734755891086498e-06,
      "loss": 1.3376,
      "step": 1061
    },
    {
      "epoch": 1.6992,
      "grad_norm": 0.6391856092920328,
      "learning_rate": 4.725458285727195e-06,
      "loss": 1.2875,
      "step": 1062
    },
    {
      "epoch": 1.7008,
      "grad_norm": 0.6826561375127412,
      "learning_rate": 4.716161632457438e-06,
      "loss": 1.3395,
      "step": 1063
    },
    {
      "epoch": 1.7024,
      "grad_norm": 0.6175998602194827,
      "learning_rate": 4.7068659635173034e-06,
      "loss": 1.3174,
      "step": 1064
    },
    {
      "epoch": 1.704,
      "grad_norm": 0.6630018103624823,
      "learning_rate": 4.6975713111434556e-06,
      "loss": 1.2974,
      "step": 1065
    },
    {
      "epoch": 1.7056,
      "grad_norm": 0.681483048644118,
      "learning_rate": 4.688277707569035e-06,
      "loss": 1.3933,
      "step": 1066
    },
    {
      "epoch": 1.7072,
      "grad_norm": 0.6426792068595732,
      "learning_rate": 4.678985185023542e-06,
      "loss": 1.2916,
      "step": 1067
    },
    {
      "epoch": 1.7088,
      "grad_norm": 0.6549016353970986,
      "learning_rate": 4.669693775732733e-06,
      "loss": 1.2989,
      "step": 1068
    },
    {
      "epoch": 1.7104,
      "grad_norm": 0.6754679545612037,
      "learning_rate": 4.660403511918499e-06,
      "loss": 1.357,
      "step": 1069
    },
    {
      "epoch": 1.712,
      "grad_norm": 0.682906262227491,
      "learning_rate": 4.65111442579876e-06,
      "loss": 1.2332,
      "step": 1070
    },
    {
      "epoch": 1.7136,
      "grad_norm": 0.6638825784009132,
      "learning_rate": 4.641826549587352e-06,
      "loss": 1.3595,
      "step": 1071
    },
    {
      "epoch": 1.7151999999999998,
      "grad_norm": 0.670087633729932,
      "learning_rate": 4.632539915493915e-06,
      "loss": 1.3208,
      "step": 1072
    },
    {
      "epoch": 1.7168,
      "grad_norm": 0.6437902163419233,
      "learning_rate": 4.623254555723783e-06,
      "loss": 1.3118,
      "step": 1073
    },
    {
      "epoch": 1.7184,
      "grad_norm": 0.603949538028129,
      "learning_rate": 4.613970502477867e-06,
      "loss": 1.2712,
      "step": 1074
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.7196483748548921,
      "learning_rate": 4.604687787952552e-06,
      "loss": 1.3075,
      "step": 1075
    },
    {
      "epoch": 1.7216,
      "grad_norm": 0.6735740517878172,
      "learning_rate": 4.5954064443395765e-06,
      "loss": 1.3386,
      "step": 1076
    },
    {
      "epoch": 1.7231999999999998,
      "grad_norm": 0.6294632128863833,
      "learning_rate": 4.586126503825925e-06,
      "loss": 1.2701,
      "step": 1077
    },
    {
      "epoch": 1.7248,
      "grad_norm": 0.6468436835545782,
      "learning_rate": 4.57684799859372e-06,
      "loss": 1.2517,
      "step": 1078
    },
    {
      "epoch": 1.7264,
      "grad_norm": 0.6403918790106707,
      "learning_rate": 4.567570960820101e-06,
      "loss": 1.3009,
      "step": 1079
    },
    {
      "epoch": 1.728,
      "grad_norm": 0.6923324461185102,
      "learning_rate": 4.558295422677124e-06,
      "loss": 1.3841,
      "step": 1080
    },
    {
      "epoch": 1.7296,
      "grad_norm": 0.6816413154009893,
      "learning_rate": 4.54902141633164e-06,
      "loss": 1.2968,
      "step": 1081
    },
    {
      "epoch": 1.7311999999999999,
      "grad_norm": 0.6487567015576441,
      "learning_rate": 4.539748973945191e-06,
      "loss": 1.3045,
      "step": 1082
    },
    {
      "epoch": 1.7328000000000001,
      "grad_norm": 0.7612026975087938,
      "learning_rate": 4.53047812767389e-06,
      "loss": 1.3,
      "step": 1083
    },
    {
      "epoch": 1.7344,
      "grad_norm": 0.6383887703195028,
      "learning_rate": 4.5212089096683234e-06,
      "loss": 1.3207,
      "step": 1084
    },
    {
      "epoch": 1.736,
      "grad_norm": 0.6598727939221881,
      "learning_rate": 4.511941352073424e-06,
      "loss": 1.3504,
      "step": 1085
    },
    {
      "epoch": 1.7376,
      "grad_norm": 0.6816383705251093,
      "learning_rate": 4.5026754870283695e-06,
      "loss": 1.3517,
      "step": 1086
    },
    {
      "epoch": 1.7391999999999999,
      "grad_norm": 0.6760278574772198,
      "learning_rate": 4.493411346666465e-06,
      "loss": 1.379,
      "step": 1087
    },
    {
      "epoch": 1.7408000000000001,
      "grad_norm": 0.6367359057645615,
      "learning_rate": 4.484148963115038e-06,
      "loss": 1.2858,
      "step": 1088
    },
    {
      "epoch": 1.7424,
      "grad_norm": 0.6550873408237964,
      "learning_rate": 4.474888368495322e-06,
      "loss": 1.2866,
      "step": 1089
    },
    {
      "epoch": 1.744,
      "grad_norm": 0.6262905091671155,
      "learning_rate": 4.465629594922348e-06,
      "loss": 1.3767,
      "step": 1090
    },
    {
      "epoch": 1.7456,
      "grad_norm": 0.6631050182829783,
      "learning_rate": 4.456372674504828e-06,
      "loss": 1.3238,
      "step": 1091
    },
    {
      "epoch": 1.7471999999999999,
      "grad_norm": 0.6527922100513226,
      "learning_rate": 4.447117639345052e-06,
      "loss": 1.3489,
      "step": 1092
    },
    {
      "epoch": 1.7488000000000001,
      "grad_norm": 0.615892368972,
      "learning_rate": 4.437864521538768e-06,
      "loss": 1.3204,
      "step": 1093
    },
    {
      "epoch": 1.7504,
      "grad_norm": 0.6853309442088157,
      "learning_rate": 4.428613353175078e-06,
      "loss": 1.3938,
      "step": 1094
    },
    {
      "epoch": 1.752,
      "grad_norm": 0.6181524470478899,
      "learning_rate": 4.4193641663363214e-06,
      "loss": 1.3885,
      "step": 1095
    },
    {
      "epoch": 1.7536,
      "grad_norm": 0.9197127002134523,
      "learning_rate": 4.410116993097968e-06,
      "loss": 1.3455,
      "step": 1096
    },
    {
      "epoch": 1.7551999999999999,
      "grad_norm": 0.6381521136093241,
      "learning_rate": 4.400871865528502e-06,
      "loss": 1.3615,
      "step": 1097
    },
    {
      "epoch": 1.7568000000000001,
      "grad_norm": 0.6767894630170835,
      "learning_rate": 4.391628815689314e-06,
      "loss": 1.33,
      "step": 1098
    },
    {
      "epoch": 1.7584,
      "grad_norm": 0.6376221036519258,
      "learning_rate": 4.382387875634592e-06,
      "loss": 1.3575,
      "step": 1099
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.6298256529817223,
      "learning_rate": 4.373149077411203e-06,
      "loss": 1.2983,
      "step": 1100
    },
    {
      "epoch": 1.7616,
      "grad_norm": 0.6241264398230109,
      "learning_rate": 4.363912453058589e-06,
      "loss": 1.3043,
      "step": 1101
    },
    {
      "epoch": 1.7631999999999999,
      "grad_norm": 0.6212968839603045,
      "learning_rate": 4.354678034608654e-06,
      "loss": 1.3198,
      "step": 1102
    },
    {
      "epoch": 1.7648000000000001,
      "grad_norm": 0.6127025633089962,
      "learning_rate": 4.345445854085649e-06,
      "loss": 1.3119,
      "step": 1103
    },
    {
      "epoch": 1.7664,
      "grad_norm": 0.6315114202548601,
      "learning_rate": 4.336215943506066e-06,
      "loss": 1.3424,
      "step": 1104
    },
    {
      "epoch": 1.768,
      "grad_norm": 0.6655339984125158,
      "learning_rate": 4.326988334878526e-06,
      "loss": 1.3158,
      "step": 1105
    },
    {
      "epoch": 1.7696,
      "grad_norm": 0.6776872335794549,
      "learning_rate": 4.317763060203665e-06,
      "loss": 1.3002,
      "step": 1106
    },
    {
      "epoch": 1.7711999999999999,
      "grad_norm": 0.6093917972043034,
      "learning_rate": 4.308540151474027e-06,
      "loss": 1.2887,
      "step": 1107
    },
    {
      "epoch": 1.7728000000000002,
      "grad_norm": 0.6454554950028193,
      "learning_rate": 4.299319640673948e-06,
      "loss": 1.4322,
      "step": 1108
    },
    {
      "epoch": 1.7744,
      "grad_norm": 0.6606218023590151,
      "learning_rate": 4.290101559779451e-06,
      "loss": 1.3423,
      "step": 1109
    },
    {
      "epoch": 1.776,
      "grad_norm": 0.6087462256102949,
      "learning_rate": 4.280885940758131e-06,
      "loss": 1.3566,
      "step": 1110
    },
    {
      "epoch": 1.7776,
      "grad_norm": 0.8448228132893343,
      "learning_rate": 4.271672815569047e-06,
      "loss": 1.3516,
      "step": 1111
    },
    {
      "epoch": 1.7792,
      "grad_norm": 0.6879383571051513,
      "learning_rate": 4.262462216162606e-06,
      "loss": 1.3157,
      "step": 1112
    },
    {
      "epoch": 1.7808000000000002,
      "grad_norm": 0.5968134558735962,
      "learning_rate": 4.253254174480462e-06,
      "loss": 1.3469,
      "step": 1113
    },
    {
      "epoch": 1.7824,
      "grad_norm": 0.6516062714006348,
      "learning_rate": 4.244048722455393e-06,
      "loss": 1.3237,
      "step": 1114
    },
    {
      "epoch": 1.784,
      "grad_norm": 0.6674639563307888,
      "learning_rate": 4.234845892011198e-06,
      "loss": 1.3305,
      "step": 1115
    },
    {
      "epoch": 1.7856,
      "grad_norm": 0.6915784844671417,
      "learning_rate": 4.225645715062585e-06,
      "loss": 1.3284,
      "step": 1116
    },
    {
      "epoch": 1.7872,
      "grad_norm": 0.6039239394294024,
      "learning_rate": 4.216448223515061e-06,
      "loss": 1.3448,
      "step": 1117
    },
    {
      "epoch": 1.7888,
      "grad_norm": 0.6406950264634086,
      "learning_rate": 4.2072534492648184e-06,
      "loss": 1.2967,
      "step": 1118
    },
    {
      "epoch": 1.7904,
      "grad_norm": 0.6684919339133126,
      "learning_rate": 4.198061424198627e-06,
      "loss": 1.3131,
      "step": 1119
    },
    {
      "epoch": 1.792,
      "grad_norm": 0.6250856170858864,
      "learning_rate": 4.188872180193723e-06,
      "loss": 1.3566,
      "step": 1120
    },
    {
      "epoch": 1.7936,
      "grad_norm": 0.6759371586355923,
      "learning_rate": 4.179685749117698e-06,
      "loss": 1.3617,
      "step": 1121
    },
    {
      "epoch": 1.7952,
      "grad_norm": 0.6399048309532533,
      "learning_rate": 4.170502162828385e-06,
      "loss": 1.2499,
      "step": 1122
    },
    {
      "epoch": 1.7968,
      "grad_norm": 0.6520800340158223,
      "learning_rate": 4.161321453173759e-06,
      "loss": 1.3582,
      "step": 1123
    },
    {
      "epoch": 1.7984,
      "grad_norm": 0.7093764595110739,
      "learning_rate": 4.152143651991812e-06,
      "loss": 1.2527,
      "step": 1124
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.6171482189985221,
      "learning_rate": 4.142968791110455e-06,
      "loss": 1.3207,
      "step": 1125
    },
    {
      "epoch": 1.8016,
      "grad_norm": 0.7173909454333728,
      "learning_rate": 4.133796902347397e-06,
      "loss": 1.349,
      "step": 1126
    },
    {
      "epoch": 1.8032,
      "grad_norm": 0.6762788065478339,
      "learning_rate": 4.124628017510043e-06,
      "loss": 1.3012,
      "step": 1127
    },
    {
      "epoch": 1.8048,
      "grad_norm": 0.668915635970412,
      "learning_rate": 4.115462168395382e-06,
      "loss": 1.3074,
      "step": 1128
    },
    {
      "epoch": 1.8064,
      "grad_norm": 0.7401349106755826,
      "learning_rate": 4.106299386789873e-06,
      "loss": 1.3394,
      "step": 1129
    },
    {
      "epoch": 1.808,
      "grad_norm": 0.6112719183338718,
      "learning_rate": 4.097139704469337e-06,
      "loss": 1.3122,
      "step": 1130
    },
    {
      "epoch": 1.8096,
      "grad_norm": 0.6614627155288405,
      "learning_rate": 4.0879831531988485e-06,
      "loss": 1.3119,
      "step": 1131
    },
    {
      "epoch": 1.8112,
      "grad_norm": 0.7096752084192162,
      "learning_rate": 4.078829764732621e-06,
      "loss": 1.4435,
      "step": 1132
    },
    {
      "epoch": 1.8128,
      "grad_norm": 0.6165225147863865,
      "learning_rate": 4.069679570813903e-06,
      "loss": 1.384,
      "step": 1133
    },
    {
      "epoch": 1.8144,
      "grad_norm": 0.708517656410055,
      "learning_rate": 4.060532603174865e-06,
      "loss": 1.2838,
      "step": 1134
    },
    {
      "epoch": 1.8159999999999998,
      "grad_norm": 0.6704052809695206,
      "learning_rate": 4.051388893536484e-06,
      "loss": 1.3169,
      "step": 1135
    },
    {
      "epoch": 1.8176,
      "grad_norm": 0.6477888537959446,
      "learning_rate": 4.042248473608442e-06,
      "loss": 1.3312,
      "step": 1136
    },
    {
      "epoch": 1.8192,
      "grad_norm": 0.7014558775149993,
      "learning_rate": 4.033111375089013e-06,
      "loss": 1.2295,
      "step": 1137
    },
    {
      "epoch": 1.8208,
      "grad_norm": 0.6218200809907051,
      "learning_rate": 4.023977629664951e-06,
      "loss": 1.3191,
      "step": 1138
    },
    {
      "epoch": 1.8224,
      "grad_norm": 0.6918694841221974,
      "learning_rate": 4.0148472690113845e-06,
      "loss": 1.3479,
      "step": 1139
    },
    {
      "epoch": 1.8239999999999998,
      "grad_norm": 4.048825436397009,
      "learning_rate": 4.0057203247917e-06,
      "loss": 1.3179,
      "step": 1140
    },
    {
      "epoch": 1.8256000000000001,
      "grad_norm": 0.7133239602284827,
      "learning_rate": 3.996596828657437e-06,
      "loss": 1.3499,
      "step": 1141
    },
    {
      "epoch": 1.8272,
      "grad_norm": 0.6994569468431421,
      "learning_rate": 3.987476812248181e-06,
      "loss": 1.3234,
      "step": 1142
    },
    {
      "epoch": 1.8288,
      "grad_norm": 0.6299715612034581,
      "learning_rate": 3.978360307191444e-06,
      "loss": 1.2891,
      "step": 1143
    },
    {
      "epoch": 1.8304,
      "grad_norm": 0.6680297319987071,
      "learning_rate": 3.969247345102567e-06,
      "loss": 1.4024,
      "step": 1144
    },
    {
      "epoch": 1.8319999999999999,
      "grad_norm": 0.634670449185284,
      "learning_rate": 3.960137957584601e-06,
      "loss": 1.3377,
      "step": 1145
    },
    {
      "epoch": 1.8336000000000001,
      "grad_norm": 0.6549563403405082,
      "learning_rate": 3.9510321762282e-06,
      "loss": 1.3455,
      "step": 1146
    },
    {
      "epoch": 1.8352,
      "grad_norm": 0.6910731470202456,
      "learning_rate": 3.941930032611513e-06,
      "loss": 1.2729,
      "step": 1147
    },
    {
      "epoch": 1.8368,
      "grad_norm": 0.6566036466559674,
      "learning_rate": 3.932831558300074e-06,
      "loss": 1.3317,
      "step": 1148
    },
    {
      "epoch": 1.8384,
      "grad_norm": 0.6080916343163177,
      "learning_rate": 3.923736784846693e-06,
      "loss": 1.3366,
      "step": 1149
    },
    {
      "epoch": 1.8399999999999999,
      "grad_norm": 0.709880643883824,
      "learning_rate": 3.914645743791346e-06,
      "loss": 1.3688,
      "step": 1150
    },
    {
      "epoch": 1.8416000000000001,
      "grad_norm": 0.6149835653990718,
      "learning_rate": 3.90555846666106e-06,
      "loss": 1.3192,
      "step": 1151
    },
    {
      "epoch": 1.8432,
      "grad_norm": 0.6970986322373535,
      "learning_rate": 3.896474984969817e-06,
      "loss": 1.3431,
      "step": 1152
    },
    {
      "epoch": 1.8448,
      "grad_norm": 0.6613420980668205,
      "learning_rate": 3.887395330218429e-06,
      "loss": 1.2925,
      "step": 1153
    },
    {
      "epoch": 1.8464,
      "grad_norm": 0.6403259324526558,
      "learning_rate": 3.878319533894443e-06,
      "loss": 1.3132,
      "step": 1154
    },
    {
      "epoch": 1.8479999999999999,
      "grad_norm": 0.7149049694598982,
      "learning_rate": 3.869247627472021e-06,
      "loss": 1.3116,
      "step": 1155
    },
    {
      "epoch": 1.8496000000000001,
      "grad_norm": 0.6549062773283906,
      "learning_rate": 3.860179642411838e-06,
      "loss": 1.3058,
      "step": 1156
    },
    {
      "epoch": 1.8512,
      "grad_norm": 0.6540876727645114,
      "learning_rate": 3.851115610160967e-06,
      "loss": 1.268,
      "step": 1157
    },
    {
      "epoch": 1.8528,
      "grad_norm": 0.6382583917002316,
      "learning_rate": 3.842055562152775e-06,
      "loss": 1.2859,
      "step": 1158
    },
    {
      "epoch": 1.8544,
      "grad_norm": 0.6951618039629249,
      "learning_rate": 3.8329995298068114e-06,
      "loss": 1.3281,
      "step": 1159
    },
    {
      "epoch": 1.8559999999999999,
      "grad_norm": 0.6998750981667256,
      "learning_rate": 3.8239475445287015e-06,
      "loss": 1.4041,
      "step": 1160
    },
    {
      "epoch": 1.8576000000000001,
      "grad_norm": 0.9882816251274343,
      "learning_rate": 3.814899637710031e-06,
      "loss": 1.3311,
      "step": 1161
    },
    {
      "epoch": 1.8592,
      "grad_norm": 0.7455332134940613,
      "learning_rate": 3.8058558407282465e-06,
      "loss": 1.301,
      "step": 1162
    },
    {
      "epoch": 1.8608,
      "grad_norm": 0.6773453763048289,
      "learning_rate": 3.7968161849465395e-06,
      "loss": 1.311,
      "step": 1163
    },
    {
      "epoch": 1.8624,
      "grad_norm": 0.6654162074909247,
      "learning_rate": 3.78778070171374e-06,
      "loss": 1.3091,
      "step": 1164
    },
    {
      "epoch": 1.8639999999999999,
      "grad_norm": 0.720610745444871,
      "learning_rate": 3.7787494223642096e-06,
      "loss": 1.2851,
      "step": 1165
    },
    {
      "epoch": 1.8656000000000001,
      "grad_norm": 0.6892793130749405,
      "learning_rate": 3.7697223782177304e-06,
      "loss": 1.3817,
      "step": 1166
    },
    {
      "epoch": 1.8672,
      "grad_norm": 0.6579321974308626,
      "learning_rate": 3.760699600579399e-06,
      "loss": 1.3276,
      "step": 1167
    },
    {
      "epoch": 1.8688,
      "grad_norm": 0.6912314742633009,
      "learning_rate": 3.7516811207395116e-06,
      "loss": 1.274,
      "step": 1168
    },
    {
      "epoch": 1.8704,
      "grad_norm": 0.6751693770763482,
      "learning_rate": 3.742666969973463e-06,
      "loss": 1.3431,
      "step": 1169
    },
    {
      "epoch": 1.8719999999999999,
      "grad_norm": 0.6390354082506068,
      "learning_rate": 3.733657179541635e-06,
      "loss": 1.316,
      "step": 1170
    },
    {
      "epoch": 1.8736000000000002,
      "grad_norm": 0.6313333713215293,
      "learning_rate": 3.724651780689286e-06,
      "loss": 1.3025,
      "step": 1171
    },
    {
      "epoch": 1.8752,
      "grad_norm": 0.6729499395314861,
      "learning_rate": 3.715650804646449e-06,
      "loss": 1.2677,
      "step": 1172
    },
    {
      "epoch": 1.8768,
      "grad_norm": 0.6377141149958804,
      "learning_rate": 3.7066542826278153e-06,
      "loss": 1.2818,
      "step": 1173
    },
    {
      "epoch": 1.8784,
      "grad_norm": 0.6302211735985253,
      "learning_rate": 3.6976622458326308e-06,
      "loss": 1.3353,
      "step": 1174
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.7003876117517727,
      "learning_rate": 3.6886747254445877e-06,
      "loss": 1.3056,
      "step": 1175
    },
    {
      "epoch": 1.8816000000000002,
      "grad_norm": 0.6106346888651357,
      "learning_rate": 3.6796917526317153e-06,
      "loss": 1.2915,
      "step": 1176
    },
    {
      "epoch": 1.8832,
      "grad_norm": 0.6311188977625891,
      "learning_rate": 3.6707133585462713e-06,
      "loss": 1.3516,
      "step": 1177
    },
    {
      "epoch": 1.8848,
      "grad_norm": 0.6884908761249647,
      "learning_rate": 3.6617395743246375e-06,
      "loss": 1.3355,
      "step": 1178
    },
    {
      "epoch": 1.8864,
      "grad_norm": 0.6124845850106975,
      "learning_rate": 3.652770431087206e-06,
      "loss": 1.308,
      "step": 1179
    },
    {
      "epoch": 1.888,
      "grad_norm": 0.6226475641545722,
      "learning_rate": 3.6438059599382765e-06,
      "loss": 1.3139,
      "step": 1180
    },
    {
      "epoch": 1.8896,
      "grad_norm": 0.6632115292848585,
      "learning_rate": 3.634846191965944e-06,
      "loss": 1.2614,
      "step": 1181
    },
    {
      "epoch": 1.8912,
      "grad_norm": 0.6325949822936453,
      "learning_rate": 3.625891158241994e-06,
      "loss": 1.2995,
      "step": 1182
    },
    {
      "epoch": 1.8928,
      "grad_norm": 0.6131510768212218,
      "learning_rate": 3.6169408898217973e-06,
      "loss": 1.3006,
      "step": 1183
    },
    {
      "epoch": 1.8944,
      "grad_norm": 0.6122334854926342,
      "learning_rate": 3.6079954177441945e-06,
      "loss": 1.3295,
      "step": 1184
    },
    {
      "epoch": 1.896,
      "grad_norm": 0.6363679764888363,
      "learning_rate": 3.599054773031394e-06,
      "loss": 1.3598,
      "step": 1185
    },
    {
      "epoch": 1.8976,
      "grad_norm": 0.6444447675647113,
      "learning_rate": 3.5901189866888654e-06,
      "loss": 1.253,
      "step": 1186
    },
    {
      "epoch": 1.8992,
      "grad_norm": 0.6190530802224519,
      "learning_rate": 3.581188089705226e-06,
      "loss": 1.2665,
      "step": 1187
    },
    {
      "epoch": 1.9008,
      "grad_norm": 0.5798608566019152,
      "learning_rate": 3.572262113052142e-06,
      "loss": 1.2758,
      "step": 1188
    },
    {
      "epoch": 1.9024,
      "grad_norm": 0.6194788568150374,
      "learning_rate": 3.563341087684213e-06,
      "loss": 1.3196,
      "step": 1189
    },
    {
      "epoch": 1.904,
      "grad_norm": 0.6199567750775735,
      "learning_rate": 3.554425044538868e-06,
      "loss": 1.2855,
      "step": 1190
    },
    {
      "epoch": 1.9056,
      "grad_norm": 0.5998461900007648,
      "learning_rate": 3.5455140145362587e-06,
      "loss": 1.3322,
      "step": 1191
    },
    {
      "epoch": 1.9072,
      "grad_norm": 0.612106693393711,
      "learning_rate": 3.5366080285791516e-06,
      "loss": 1.3078,
      "step": 1192
    },
    {
      "epoch": 1.9088,
      "grad_norm": 0.6659761967537392,
      "learning_rate": 3.527707117552822e-06,
      "loss": 1.309,
      "step": 1193
    },
    {
      "epoch": 1.9104,
      "grad_norm": 0.5666878093877382,
      "learning_rate": 3.5188113123249435e-06,
      "loss": 1.3594,
      "step": 1194
    },
    {
      "epoch": 1.912,
      "grad_norm": 0.6249739984905636,
      "learning_rate": 3.5099206437454852e-06,
      "loss": 1.331,
      "step": 1195
    },
    {
      "epoch": 1.9136,
      "grad_norm": 0.6166416945937175,
      "learning_rate": 3.5010351426466006e-06,
      "loss": 1.3333,
      "step": 1196
    },
    {
      "epoch": 1.9152,
      "grad_norm": 0.5986213643145921,
      "learning_rate": 3.4921548398425246e-06,
      "loss": 1.2876,
      "step": 1197
    },
    {
      "epoch": 1.9167999999999998,
      "grad_norm": 0.6413042573654271,
      "learning_rate": 3.4832797661294633e-06,
      "loss": 1.3103,
      "step": 1198
    },
    {
      "epoch": 1.9184,
      "grad_norm": 0.590133376410799,
      "learning_rate": 3.4744099522854914e-06,
      "loss": 1.3193,
      "step": 1199
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7573245540756742,
      "learning_rate": 3.4655454290704393e-06,
      "loss": 1.343,
      "step": 1200
    },
    {
      "epoch": 1.9216,
      "grad_norm": 0.6584893116425911,
      "learning_rate": 3.4566862272257923e-06,
      "loss": 1.3439,
      "step": 1201
    },
    {
      "epoch": 1.9232,
      "grad_norm": 0.5885388426411419,
      "learning_rate": 3.44783237747458e-06,
      "loss": 1.2976,
      "step": 1202
    },
    {
      "epoch": 1.9247999999999998,
      "grad_norm": 0.6991640985075909,
      "learning_rate": 3.438983910521273e-06,
      "loss": 1.3672,
      "step": 1203
    },
    {
      "epoch": 1.9264000000000001,
      "grad_norm": 0.6339619261396773,
      "learning_rate": 3.430140857051675e-06,
      "loss": 1.3157,
      "step": 1204
    },
    {
      "epoch": 1.928,
      "grad_norm": 0.6548451484816092,
      "learning_rate": 3.421303247732813e-06,
      "loss": 1.3489,
      "step": 1205
    },
    {
      "epoch": 1.9296,
      "grad_norm": 0.599780709326664,
      "learning_rate": 3.4124711132128374e-06,
      "loss": 1.3049,
      "step": 1206
    },
    {
      "epoch": 1.9312,
      "grad_norm": 0.6136370385499912,
      "learning_rate": 3.4036444841209113e-06,
      "loss": 1.2331,
      "step": 1207
    },
    {
      "epoch": 1.9327999999999999,
      "grad_norm": 0.6280928565524572,
      "learning_rate": 3.3948233910671036e-06,
      "loss": 1.329,
      "step": 1208
    },
    {
      "epoch": 1.9344000000000001,
      "grad_norm": 0.5910824617286615,
      "learning_rate": 3.3860078646422894e-06,
      "loss": 1.3937,
      "step": 1209
    },
    {
      "epoch": 1.936,
      "grad_norm": 0.6494772725990872,
      "learning_rate": 3.3771979354180343e-06,
      "loss": 1.3459,
      "step": 1210
    },
    {
      "epoch": 1.9376,
      "grad_norm": 0.6080845520070994,
      "learning_rate": 3.3683936339464957e-06,
      "loss": 1.2751,
      "step": 1211
    },
    {
      "epoch": 1.9392,
      "grad_norm": 0.6043129308720684,
      "learning_rate": 3.359594990760313e-06,
      "loss": 1.2744,
      "step": 1212
    },
    {
      "epoch": 1.9407999999999999,
      "grad_norm": 0.6360940377587104,
      "learning_rate": 3.3508020363725043e-06,
      "loss": 1.3352,
      "step": 1213
    },
    {
      "epoch": 1.9424000000000001,
      "grad_norm": 0.6450328713238515,
      "learning_rate": 3.34201480127636e-06,
      "loss": 1.2939,
      "step": 1214
    },
    {
      "epoch": 1.944,
      "grad_norm": 0.6318388560106656,
      "learning_rate": 3.333233315945333e-06,
      "loss": 1.3401,
      "step": 1215
    },
    {
      "epoch": 1.9456,
      "grad_norm": 0.5977154514541712,
      "learning_rate": 3.324457610832942e-06,
      "loss": 1.2625,
      "step": 1216
    },
    {
      "epoch": 1.9472,
      "grad_norm": 0.6267366863803461,
      "learning_rate": 3.315687716372655e-06,
      "loss": 1.3429,
      "step": 1217
    },
    {
      "epoch": 1.9487999999999999,
      "grad_norm": 0.6417295851062828,
      "learning_rate": 3.306923662977789e-06,
      "loss": 1.3376,
      "step": 1218
    },
    {
      "epoch": 1.9504000000000001,
      "grad_norm": 0.6123826071436064,
      "learning_rate": 3.2981654810414128e-06,
      "loss": 1.3482,
      "step": 1219
    },
    {
      "epoch": 1.952,
      "grad_norm": 0.5872913748622864,
      "learning_rate": 3.2894132009362245e-06,
      "loss": 1.2884,
      "step": 1220
    },
    {
      "epoch": 1.9536,
      "grad_norm": 0.6270368392858394,
      "learning_rate": 3.280666853014457e-06,
      "loss": 1.3509,
      "step": 1221
    },
    {
      "epoch": 1.9552,
      "grad_norm": 0.6876731067848509,
      "learning_rate": 3.271926467607774e-06,
      "loss": 1.3602,
      "step": 1222
    },
    {
      "epoch": 1.9567999999999999,
      "grad_norm": 0.5989859400903079,
      "learning_rate": 3.2631920750271594e-06,
      "loss": 1.3135,
      "step": 1223
    },
    {
      "epoch": 1.9584000000000001,
      "grad_norm": 0.6809150085301863,
      "learning_rate": 3.2544637055628135e-06,
      "loss": 1.3493,
      "step": 1224
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.6402833822548974,
      "learning_rate": 3.2457413894840516e-06,
      "loss": 1.2576,
      "step": 1225
    },
    {
      "epoch": 1.9616,
      "grad_norm": 0.6072730659797738,
      "learning_rate": 3.2370251570391933e-06,
      "loss": 1.3299,
      "step": 1226
    },
    {
      "epoch": 1.9632,
      "grad_norm": 14.468809712375457,
      "learning_rate": 3.2283150384554642e-06,
      "loss": 1.6002,
      "step": 1227
    },
    {
      "epoch": 1.9647999999999999,
      "grad_norm": 0.7700356274758257,
      "learning_rate": 3.219611063938883e-06,
      "loss": 1.3305,
      "step": 1228
    },
    {
      "epoch": 1.9664000000000001,
      "grad_norm": 0.647255832117017,
      "learning_rate": 3.210913263674166e-06,
      "loss": 1.2945,
      "step": 1229
    },
    {
      "epoch": 1.968,
      "grad_norm": 0.5908351536353945,
      "learning_rate": 3.2022216678246145e-06,
      "loss": 1.311,
      "step": 1230
    },
    {
      "epoch": 1.9696,
      "grad_norm": 0.7236263094620875,
      "learning_rate": 3.193536306532013e-06,
      "loss": 1.3085,
      "step": 1231
    },
    {
      "epoch": 1.9712,
      "grad_norm": 0.6573904719839839,
      "learning_rate": 3.184857209916528e-06,
      "loss": 1.3243,
      "step": 1232
    },
    {
      "epoch": 1.9727999999999999,
      "grad_norm": 0.6625321012113893,
      "learning_rate": 3.1761844080765993e-06,
      "loss": 1.2656,
      "step": 1233
    },
    {
      "epoch": 1.9744000000000002,
      "grad_norm": 0.6553575840177628,
      "learning_rate": 3.1675179310888344e-06,
      "loss": 1.32,
      "step": 1234
    },
    {
      "epoch": 1.976,
      "grad_norm": 0.6368289506322888,
      "learning_rate": 3.15885780900791e-06,
      "loss": 1.3062,
      "step": 1235
    },
    {
      "epoch": 1.9776,
      "grad_norm": 0.7069922894835612,
      "learning_rate": 3.150204071866464e-06,
      "loss": 1.301,
      "step": 1236
    },
    {
      "epoch": 1.9792,
      "grad_norm": 0.6654820104409741,
      "learning_rate": 3.141556749674988e-06,
      "loss": 1.2604,
      "step": 1237
    },
    {
      "epoch": 1.9808,
      "grad_norm": 0.6089240990561239,
      "learning_rate": 3.132915872421734e-06,
      "loss": 1.2659,
      "step": 1238
    },
    {
      "epoch": 1.9824000000000002,
      "grad_norm": 0.6870573908969106,
      "learning_rate": 3.1242814700725977e-06,
      "loss": 1.2747,
      "step": 1239
    },
    {
      "epoch": 1.984,
      "grad_norm": 0.6522515840787557,
      "learning_rate": 3.1156535725710224e-06,
      "loss": 1.2735,
      "step": 1240
    },
    {
      "epoch": 1.9856,
      "grad_norm": 0.5975499492188486,
      "learning_rate": 3.1070322098378925e-06,
      "loss": 1.3058,
      "step": 1241
    },
    {
      "epoch": 1.9872,
      "grad_norm": 0.6812613426222909,
      "learning_rate": 3.0984174117714306e-06,
      "loss": 1.2752,
      "step": 1242
    },
    {
      "epoch": 1.9888,
      "grad_norm": 0.5965624756760142,
      "learning_rate": 3.0898092082470943e-06,
      "loss": 1.3192,
      "step": 1243
    },
    {
      "epoch": 1.9904,
      "grad_norm": 0.6082612784496522,
      "learning_rate": 3.081207629117472e-06,
      "loss": 1.2993,
      "step": 1244
    },
    {
      "epoch": 1.992,
      "grad_norm": 0.6087021007198163,
      "learning_rate": 3.0726127042121766e-06,
      "loss": 1.4174,
      "step": 1245
    },
    {
      "epoch": 1.9936,
      "grad_norm": 0.6238406654539388,
      "learning_rate": 3.064024463337747e-06,
      "loss": 1.3493,
      "step": 1246
    },
    {
      "epoch": 1.9952,
      "grad_norm": 0.61369377892187,
      "learning_rate": 3.0554429362775417e-06,
      "loss": 1.3153,
      "step": 1247
    },
    {
      "epoch": 1.9968,
      "grad_norm": 0.5798667652334559,
      "learning_rate": 3.046868152791638e-06,
      "loss": 1.2629,
      "step": 1248
    },
    {
      "epoch": 1.9984,
      "grad_norm": 0.6412835652968717,
      "learning_rate": 3.038300142616723e-06,
      "loss": 1.3351,
      "step": 1249
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.6246035637368326,
      "learning_rate": 3.0297389354659984e-06,
      "loss": 1.2868,
      "step": 1250
    },
    {
      "epoch": 2.0016,
      "grad_norm": 0.8467921912059534,
      "learning_rate": 3.021184561029071e-06,
      "loss": 1.3158,
      "step": 1251
    },
    {
      "epoch": 2.0032,
      "grad_norm": 0.7026382946914086,
      "learning_rate": 3.0126370489718537e-06,
      "loss": 1.3137,
      "step": 1252
    },
    {
      "epoch": 2.0048,
      "grad_norm": 0.7263063641735807,
      "learning_rate": 3.0040964289364618e-06,
      "loss": 1.2614,
      "step": 1253
    },
    {
      "epoch": 2.0064,
      "grad_norm": 0.7027840695636279,
      "learning_rate": 2.9955627305411074e-06,
      "loss": 1.218,
      "step": 1254
    },
    {
      "epoch": 2.008,
      "grad_norm": 0.699895540245262,
      "learning_rate": 2.9870359833799994e-06,
      "loss": 1.2001,
      "step": 1255
    },
    {
      "epoch": 2.0096,
      "grad_norm": 0.7038698256144013,
      "learning_rate": 2.978516217023243e-06,
      "loss": 1.2316,
      "step": 1256
    },
    {
      "epoch": 2.0112,
      "grad_norm": 0.6676232969039125,
      "learning_rate": 2.97000346101673e-06,
      "loss": 1.2872,
      "step": 1257
    },
    {
      "epoch": 2.0128,
      "grad_norm": 0.6950691122015595,
      "learning_rate": 2.9614977448820444e-06,
      "loss": 1.3088,
      "step": 1258
    },
    {
      "epoch": 2.0144,
      "grad_norm": 0.6773357407276979,
      "learning_rate": 2.952999098116356e-06,
      "loss": 1.2872,
      "step": 1259
    },
    {
      "epoch": 2.016,
      "grad_norm": 0.6945931585633626,
      "learning_rate": 2.944507550192318e-06,
      "loss": 1.2941,
      "step": 1260
    },
    {
      "epoch": 2.0176,
      "grad_norm": 0.6781016935364389,
      "learning_rate": 2.9360231305579645e-06,
      "loss": 1.2231,
      "step": 1261
    },
    {
      "epoch": 2.0192,
      "grad_norm": 0.6702239548677627,
      "learning_rate": 2.9275458686366108e-06,
      "loss": 1.3095,
      "step": 1262
    },
    {
      "epoch": 2.0208,
      "grad_norm": 0.6698215842323128,
      "learning_rate": 2.9190757938267477e-06,
      "loss": 1.276,
      "step": 1263
    },
    {
      "epoch": 2.0224,
      "grad_norm": 0.6671591410003507,
      "learning_rate": 2.9106129355019464e-06,
      "loss": 1.2999,
      "step": 1264
    },
    {
      "epoch": 2.024,
      "grad_norm": 0.6252431119065879,
      "learning_rate": 2.9021573230107436e-06,
      "loss": 1.2807,
      "step": 1265
    },
    {
      "epoch": 2.0256,
      "grad_norm": 0.6649236379593464,
      "learning_rate": 2.8937089856765564e-06,
      "loss": 1.2071,
      "step": 1266
    },
    {
      "epoch": 2.0272,
      "grad_norm": 0.6541283112079199,
      "learning_rate": 2.885267952797569e-06,
      "loss": 1.3383,
      "step": 1267
    },
    {
      "epoch": 2.0288,
      "grad_norm": 0.5991725134400661,
      "learning_rate": 2.876834253646631e-06,
      "loss": 1.2791,
      "step": 1268
    },
    {
      "epoch": 2.0304,
      "grad_norm": 0.6281708245177252,
      "learning_rate": 2.8684079174711665e-06,
      "loss": 1.2221,
      "step": 1269
    },
    {
      "epoch": 2.032,
      "grad_norm": 0.6448187829769144,
      "learning_rate": 2.8599889734930548e-06,
      "loss": 1.2545,
      "step": 1270
    },
    {
      "epoch": 2.0336,
      "grad_norm": 0.596735844346817,
      "learning_rate": 2.8515774509085535e-06,
      "loss": 1.302,
      "step": 1271
    },
    {
      "epoch": 2.0352,
      "grad_norm": 0.6049269976199287,
      "learning_rate": 2.8431733788881703e-06,
      "loss": 1.2712,
      "step": 1272
    },
    {
      "epoch": 2.0368,
      "grad_norm": 0.6626117944540131,
      "learning_rate": 2.8347767865765828e-06,
      "loss": 1.2916,
      "step": 1273
    },
    {
      "epoch": 2.0384,
      "grad_norm": 0.6054861127171095,
      "learning_rate": 2.826387703092528e-06,
      "loss": 1.2563,
      "step": 1274
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.5978986848219142,
      "learning_rate": 2.8180061575286995e-06,
      "loss": 1.2254,
      "step": 1275
    },
    {
      "epoch": 2.0416,
      "grad_norm": 0.6307146576412761,
      "learning_rate": 2.8096321789516557e-06,
      "loss": 1.2544,
      "step": 1276
    },
    {
      "epoch": 2.0432,
      "grad_norm": 0.686306419024282,
      "learning_rate": 2.8012657964017056e-06,
      "loss": 1.2391,
      "step": 1277
    },
    {
      "epoch": 2.0448,
      "grad_norm": 0.6637599571414298,
      "learning_rate": 2.792907038892823e-06,
      "loss": 1.3435,
      "step": 1278
    },
    {
      "epoch": 2.0464,
      "grad_norm": 0.6261326730504239,
      "learning_rate": 2.784555935412538e-06,
      "loss": 1.2968,
      "step": 1279
    },
    {
      "epoch": 2.048,
      "grad_norm": 0.6172898312670141,
      "learning_rate": 2.77621251492183e-06,
      "loss": 1.2737,
      "step": 1280
    },
    {
      "epoch": 2.0496,
      "grad_norm": 0.6785677975592969,
      "learning_rate": 2.7678768063550454e-06,
      "loss": 1.1795,
      "step": 1281
    },
    {
      "epoch": 2.0512,
      "grad_norm": 0.6301863795258116,
      "learning_rate": 2.759548838619774e-06,
      "loss": 1.2544,
      "step": 1282
    },
    {
      "epoch": 2.0528,
      "grad_norm": 0.6259883467804318,
      "learning_rate": 2.7512286405967726e-06,
      "loss": 1.2468,
      "step": 1283
    },
    {
      "epoch": 2.0544,
      "grad_norm": 0.635411677515335,
      "learning_rate": 2.742916241139843e-06,
      "loss": 1.2807,
      "step": 1284
    },
    {
      "epoch": 2.056,
      "grad_norm": 0.5864147617008509,
      "learning_rate": 2.7346116690757496e-06,
      "loss": 1.3197,
      "step": 1285
    },
    {
      "epoch": 2.0576,
      "grad_norm": 0.6164926047571148,
      "learning_rate": 2.726314953204111e-06,
      "loss": 1.2836,
      "step": 1286
    },
    {
      "epoch": 2.0592,
      "grad_norm": 0.6618379096496074,
      "learning_rate": 2.718026122297297e-06,
      "loss": 1.2933,
      "step": 1287
    },
    {
      "epoch": 2.0608,
      "grad_norm": 0.6389135995035234,
      "learning_rate": 2.7097452051003375e-06,
      "loss": 1.2865,
      "step": 1288
    },
    {
      "epoch": 2.0624,
      "grad_norm": 0.6231229063191868,
      "learning_rate": 2.701472230330813e-06,
      "loss": 1.1952,
      "step": 1289
    },
    {
      "epoch": 2.064,
      "grad_norm": 0.6079207952712706,
      "learning_rate": 2.693207226678767e-06,
      "loss": 1.2541,
      "step": 1290
    },
    {
      "epoch": 2.0656,
      "grad_norm": 0.6605384316816444,
      "learning_rate": 2.684950222806596e-06,
      "loss": 1.2606,
      "step": 1291
    },
    {
      "epoch": 2.0672,
      "grad_norm": 0.6376121517946606,
      "learning_rate": 2.676701247348951e-06,
      "loss": 1.2589,
      "step": 1292
    },
    {
      "epoch": 2.0688,
      "grad_norm": 0.6164619333904917,
      "learning_rate": 2.6684603289126492e-06,
      "loss": 1.277,
      "step": 1293
    },
    {
      "epoch": 2.0704,
      "grad_norm": 0.6303927420865051,
      "learning_rate": 2.660227496076557e-06,
      "loss": 1.3001,
      "step": 1294
    },
    {
      "epoch": 2.072,
      "grad_norm": 0.6195516490507735,
      "learning_rate": 2.6520027773915075e-06,
      "loss": 1.2442,
      "step": 1295
    },
    {
      "epoch": 2.0736,
      "grad_norm": 0.6330567754034269,
      "learning_rate": 2.643786201380194e-06,
      "loss": 1.2979,
      "step": 1296
    },
    {
      "epoch": 2.0752,
      "grad_norm": 0.5937374023549747,
      "learning_rate": 2.6355777965370665e-06,
      "loss": 1.2592,
      "step": 1297
    },
    {
      "epoch": 2.0768,
      "grad_norm": 0.6132277119820595,
      "learning_rate": 2.627377591328245e-06,
      "loss": 1.2675,
      "step": 1298
    },
    {
      "epoch": 2.0784,
      "grad_norm": 0.6286136854433573,
      "learning_rate": 2.6191856141914074e-06,
      "loss": 1.3343,
      "step": 1299
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.6061867722517124,
      "learning_rate": 2.6110018935357005e-06,
      "loss": 1.259,
      "step": 1300
    },
    {
      "epoch": 2.0816,
      "grad_norm": 0.6180346556750618,
      "learning_rate": 2.6028264577416418e-06,
      "loss": 1.2345,
      "step": 1301
    },
    {
      "epoch": 2.0832,
      "grad_norm": 0.6476183336329412,
      "learning_rate": 2.594659335161008e-06,
      "loss": 1.3113,
      "step": 1302
    },
    {
      "epoch": 2.0848,
      "grad_norm": 0.6277139975959528,
      "learning_rate": 2.586500554116757e-06,
      "loss": 1.2383,
      "step": 1303
    },
    {
      "epoch": 2.0864,
      "grad_norm": 0.6294314086297267,
      "learning_rate": 2.578350142902909e-06,
      "loss": 1.2933,
      "step": 1304
    },
    {
      "epoch": 2.088,
      "grad_norm": 0.6194080260179882,
      "learning_rate": 2.570208129784466e-06,
      "loss": 1.2102,
      "step": 1305
    },
    {
      "epoch": 2.0896,
      "grad_norm": 0.6445750405269448,
      "learning_rate": 2.562074542997305e-06,
      "loss": 1.3045,
      "step": 1306
    },
    {
      "epoch": 2.0912,
      "grad_norm": 0.631376292546275,
      "learning_rate": 2.5539494107480746e-06,
      "loss": 1.3036,
      "step": 1307
    },
    {
      "epoch": 2.0928,
      "grad_norm": 0.7314997624641156,
      "learning_rate": 2.545832761214112e-06,
      "loss": 1.2794,
      "step": 1308
    },
    {
      "epoch": 2.0944,
      "grad_norm": 0.6986000186881405,
      "learning_rate": 2.5377246225433306e-06,
      "loss": 1.2994,
      "step": 1309
    },
    {
      "epoch": 2.096,
      "grad_norm": 0.6860933019298338,
      "learning_rate": 2.529625022854133e-06,
      "loss": 1.3158,
      "step": 1310
    },
    {
      "epoch": 2.0976,
      "grad_norm": 0.6248486540981119,
      "learning_rate": 2.5215339902353097e-06,
      "loss": 1.2855,
      "step": 1311
    },
    {
      "epoch": 2.0992,
      "grad_norm": 0.6033498043000807,
      "learning_rate": 2.513451552745936e-06,
      "loss": 1.3026,
      "step": 1312
    },
    {
      "epoch": 2.1008,
      "grad_norm": 0.6316978898876466,
      "learning_rate": 2.505377738415286e-06,
      "loss": 1.3356,
      "step": 1313
    },
    {
      "epoch": 2.1024,
      "grad_norm": 0.6670772293647546,
      "learning_rate": 2.4973125752427243e-06,
      "loss": 1.2709,
      "step": 1314
    },
    {
      "epoch": 2.104,
      "grad_norm": 0.6209905085553976,
      "learning_rate": 2.4892560911976167e-06,
      "loss": 1.3127,
      "step": 1315
    },
    {
      "epoch": 2.1056,
      "grad_norm": 0.6519265994705039,
      "learning_rate": 2.481208314219233e-06,
      "loss": 1.257,
      "step": 1316
    },
    {
      "epoch": 2.1072,
      "grad_norm": 0.6402494067527705,
      "learning_rate": 2.4731692722166408e-06,
      "loss": 1.2693,
      "step": 1317
    },
    {
      "epoch": 2.1088,
      "grad_norm": 0.5991729786131702,
      "learning_rate": 2.4651389930686226e-06,
      "loss": 1.2141,
      "step": 1318
    },
    {
      "epoch": 2.1104,
      "grad_norm": 0.6334786479858866,
      "learning_rate": 2.457117504623565e-06,
      "loss": 1.2199,
      "step": 1319
    },
    {
      "epoch": 2.112,
      "grad_norm": 0.6840373801856968,
      "learning_rate": 2.4491048346993756e-06,
      "loss": 1.2789,
      "step": 1320
    },
    {
      "epoch": 2.1136,
      "grad_norm": 0.6551061635576735,
      "learning_rate": 2.4411010110833783e-06,
      "loss": 1.2299,
      "step": 1321
    },
    {
      "epoch": 2.1152,
      "grad_norm": 0.6207153246227417,
      "learning_rate": 2.433106061532214e-06,
      "loss": 1.2182,
      "step": 1322
    },
    {
      "epoch": 2.1168,
      "grad_norm": 0.6179551330293862,
      "learning_rate": 2.4251200137717545e-06,
      "loss": 1.2299,
      "step": 1323
    },
    {
      "epoch": 2.1184,
      "grad_norm": 0.6473218730321081,
      "learning_rate": 2.4171428954969995e-06,
      "loss": 1.2474,
      "step": 1324
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.6204347158900584,
      "learning_rate": 2.4091747343719828e-06,
      "loss": 1.2513,
      "step": 1325
    },
    {
      "epoch": 2.1216,
      "grad_norm": 0.6967678297045189,
      "learning_rate": 2.401215558029671e-06,
      "loss": 1.2474,
      "step": 1326
    },
    {
      "epoch": 2.1232,
      "grad_norm": 0.6543990979898268,
      "learning_rate": 2.3932653940718784e-06,
      "loss": 1.3256,
      "step": 1327
    },
    {
      "epoch": 2.1248,
      "grad_norm": 0.6071638588975897,
      "learning_rate": 2.3853242700691594e-06,
      "loss": 1.2837,
      "step": 1328
    },
    {
      "epoch": 2.1264,
      "grad_norm": 0.6597937251657913,
      "learning_rate": 2.3773922135607217e-06,
      "loss": 1.2531,
      "step": 1329
    },
    {
      "epoch": 2.128,
      "grad_norm": 0.6723284078718527,
      "learning_rate": 2.3694692520543293e-06,
      "loss": 1.2744,
      "step": 1330
    },
    {
      "epoch": 2.1296,
      "grad_norm": 0.6402094615960296,
      "learning_rate": 2.3615554130262003e-06,
      "loss": 1.3014,
      "step": 1331
    },
    {
      "epoch": 2.1312,
      "grad_norm": 0.6113375434921613,
      "learning_rate": 2.3536507239209223e-06,
      "loss": 1.1988,
      "step": 1332
    },
    {
      "epoch": 2.1328,
      "grad_norm": 0.6264078792172373,
      "learning_rate": 2.3457552121513455e-06,
      "loss": 1.3186,
      "step": 1333
    },
    {
      "epoch": 2.1344,
      "grad_norm": 0.643359381627878,
      "learning_rate": 2.337868905098499e-06,
      "loss": 1.2708,
      "step": 1334
    },
    {
      "epoch": 2.136,
      "grad_norm": 0.6196431058643349,
      "learning_rate": 2.329991830111492e-06,
      "loss": 1.2901,
      "step": 1335
    },
    {
      "epoch": 2.1376,
      "grad_norm": 0.6343146423229109,
      "learning_rate": 2.32212401450741e-06,
      "loss": 1.2168,
      "step": 1336
    },
    {
      "epoch": 2.1391999999999998,
      "grad_norm": 0.6067879792541507,
      "learning_rate": 2.3142654855712353e-06,
      "loss": 1.2338,
      "step": 1337
    },
    {
      "epoch": 2.1408,
      "grad_norm": 0.618830791561787,
      "learning_rate": 2.3064162705557387e-06,
      "loss": 1.2149,
      "step": 1338
    },
    {
      "epoch": 2.1424,
      "grad_norm": 0.6193758212854783,
      "learning_rate": 2.2985763966813963e-06,
      "loss": 1.31,
      "step": 1339
    },
    {
      "epoch": 2.144,
      "grad_norm": 0.6444034391351889,
      "learning_rate": 2.2907458911362885e-06,
      "loss": 1.2348,
      "step": 1340
    },
    {
      "epoch": 2.1456,
      "grad_norm": 0.6235749686145161,
      "learning_rate": 2.2829247810760023e-06,
      "loss": 1.2214,
      "step": 1341
    },
    {
      "epoch": 2.1471999999999998,
      "grad_norm": 0.6114023346020294,
      "learning_rate": 2.275113093623551e-06,
      "loss": 1.2686,
      "step": 1342
    },
    {
      "epoch": 2.1488,
      "grad_norm": 0.6655316689323277,
      "learning_rate": 2.2673108558692603e-06,
      "loss": 1.2442,
      "step": 1343
    },
    {
      "epoch": 2.1504,
      "grad_norm": 0.6268025742329077,
      "learning_rate": 2.259518094870693e-06,
      "loss": 1.2281,
      "step": 1344
    },
    {
      "epoch": 2.152,
      "grad_norm": 0.6371675972052467,
      "learning_rate": 2.251734837652547e-06,
      "loss": 1.2877,
      "step": 1345
    },
    {
      "epoch": 2.1536,
      "grad_norm": 0.6183826641780498,
      "learning_rate": 2.243961111206555e-06,
      "loss": 1.2507,
      "step": 1346
    },
    {
      "epoch": 2.1552,
      "grad_norm": 0.6267951892834761,
      "learning_rate": 2.236196942491407e-06,
      "loss": 1.265,
      "step": 1347
    },
    {
      "epoch": 2.1568,
      "grad_norm": 0.633052940600082,
      "learning_rate": 2.228442358432638e-06,
      "loss": 1.2504,
      "step": 1348
    },
    {
      "epoch": 2.1584,
      "grad_norm": 0.6164245575935143,
      "learning_rate": 2.2206973859225518e-06,
      "loss": 1.2584,
      "step": 1349
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.6494025786288816,
      "learning_rate": 2.2129620518201184e-06,
      "loss": 1.3038,
      "step": 1350
    },
    {
      "epoch": 2.1616,
      "grad_norm": 0.6661259056058949,
      "learning_rate": 2.2052363829508776e-06,
      "loss": 1.2244,
      "step": 1351
    },
    {
      "epoch": 2.1632,
      "grad_norm": 0.6478775796366764,
      "learning_rate": 2.1975204061068594e-06,
      "loss": 1.3614,
      "step": 1352
    },
    {
      "epoch": 2.1648,
      "grad_norm": 0.5888178336108175,
      "learning_rate": 2.189814148046473e-06,
      "loss": 1.275,
      "step": 1353
    },
    {
      "epoch": 2.1664,
      "grad_norm": 0.6144664327755119,
      "learning_rate": 2.182117635494431e-06,
      "loss": 1.2375,
      "step": 1354
    },
    {
      "epoch": 2.168,
      "grad_norm": 0.6496904569133986,
      "learning_rate": 2.1744308951416483e-06,
      "loss": 1.3446,
      "step": 1355
    },
    {
      "epoch": 2.1696,
      "grad_norm": 0.6286191687938929,
      "learning_rate": 2.1667539536451455e-06,
      "loss": 1.2295,
      "step": 1356
    },
    {
      "epoch": 2.1712,
      "grad_norm": 0.5971434890046149,
      "learning_rate": 2.1590868376279693e-06,
      "loss": 1.2395,
      "step": 1357
    },
    {
      "epoch": 2.1728,
      "grad_norm": 0.6344850196267976,
      "learning_rate": 2.151429573679084e-06,
      "loss": 1.2265,
      "step": 1358
    },
    {
      "epoch": 2.1744,
      "grad_norm": 0.6186019085298168,
      "learning_rate": 2.1437821883532956e-06,
      "loss": 1.2441,
      "step": 1359
    },
    {
      "epoch": 2.176,
      "grad_norm": 0.6164781134971133,
      "learning_rate": 2.136144708171145e-06,
      "loss": 1.3038,
      "step": 1360
    },
    {
      "epoch": 2.1776,
      "grad_norm": 0.5820867574983208,
      "learning_rate": 2.128517159618827e-06,
      "loss": 1.3014,
      "step": 1361
    },
    {
      "epoch": 2.1792,
      "grad_norm": 0.6276672700979027,
      "learning_rate": 2.1208995691480947e-06,
      "loss": 1.2924,
      "step": 1362
    },
    {
      "epoch": 2.1808,
      "grad_norm": 0.6011247208515993,
      "learning_rate": 2.1132919631761637e-06,
      "loss": 1.2643,
      "step": 1363
    },
    {
      "epoch": 2.1824,
      "grad_norm": 0.5813604064944615,
      "learning_rate": 2.1056943680856286e-06,
      "loss": 1.2453,
      "step": 1364
    },
    {
      "epoch": 2.184,
      "grad_norm": 0.6011975725495142,
      "learning_rate": 2.098106810224362e-06,
      "loss": 1.261,
      "step": 1365
    },
    {
      "epoch": 2.1856,
      "grad_norm": 0.621493294584029,
      "learning_rate": 2.0905293159054315e-06,
      "loss": 1.2174,
      "step": 1366
    },
    {
      "epoch": 2.1872,
      "grad_norm": 0.6271665292026299,
      "learning_rate": 2.0829619114070068e-06,
      "loss": 1.2781,
      "step": 1367
    },
    {
      "epoch": 2.1888,
      "grad_norm": 0.5940837305555879,
      "learning_rate": 2.075404622972261e-06,
      "loss": 1.3112,
      "step": 1368
    },
    {
      "epoch": 2.1904,
      "grad_norm": 0.6156030908361901,
      "learning_rate": 2.0678574768092926e-06,
      "loss": 1.3353,
      "step": 1369
    },
    {
      "epoch": 2.192,
      "grad_norm": 0.6277009236002056,
      "learning_rate": 2.0603204990910195e-06,
      "loss": 1.2952,
      "step": 1370
    },
    {
      "epoch": 2.1936,
      "grad_norm": 0.5939071610543329,
      "learning_rate": 2.0527937159551044e-06,
      "loss": 1.2404,
      "step": 1371
    },
    {
      "epoch": 2.1952,
      "grad_norm": 0.6145367350018467,
      "learning_rate": 2.0452771535038518e-06,
      "loss": 1.263,
      "step": 1372
    },
    {
      "epoch": 2.1968,
      "grad_norm": 0.5835076576640534,
      "learning_rate": 2.03777083780412e-06,
      "loss": 1.242,
      "step": 1373
    },
    {
      "epoch": 2.1984,
      "grad_norm": 0.6140788157506267,
      "learning_rate": 2.030274794887237e-06,
      "loss": 1.3365,
      "step": 1374
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.593625003297278,
      "learning_rate": 2.0227890507488993e-06,
      "loss": 1.2813,
      "step": 1375
    },
    {
      "epoch": 2.2016,
      "grad_norm": 0.6008159880359805,
      "learning_rate": 2.0153136313490945e-06,
      "loss": 1.3148,
      "step": 1376
    },
    {
      "epoch": 2.2032,
      "grad_norm": 0.6109629216594358,
      "learning_rate": 2.0078485626120015e-06,
      "loss": 1.2923,
      "step": 1377
    },
    {
      "epoch": 2.2048,
      "grad_norm": 0.6138043280129014,
      "learning_rate": 2.000393870425904e-06,
      "loss": 1.2865,
      "step": 1378
    },
    {
      "epoch": 2.2064,
      "grad_norm": 0.6055776192469358,
      "learning_rate": 1.9929495806431024e-06,
      "loss": 1.269,
      "step": 1379
    },
    {
      "epoch": 2.208,
      "grad_norm": 0.5939238621491928,
      "learning_rate": 1.985515719079819e-06,
      "loss": 1.2262,
      "step": 1380
    },
    {
      "epoch": 2.2096,
      "grad_norm": 0.5918131077392118,
      "learning_rate": 1.978092311516116e-06,
      "loss": 1.2823,
      "step": 1381
    },
    {
      "epoch": 2.2112,
      "grad_norm": 0.5957104305340708,
      "learning_rate": 1.9706793836957964e-06,
      "loss": 1.2355,
      "step": 1382
    },
    {
      "epoch": 2.2128,
      "grad_norm": 0.6022030880009704,
      "learning_rate": 1.963276961326326e-06,
      "loss": 1.237,
      "step": 1383
    },
    {
      "epoch": 2.2144,
      "grad_norm": 0.6177498074135761,
      "learning_rate": 1.955885070078737e-06,
      "loss": 1.3081,
      "step": 1384
    },
    {
      "epoch": 2.216,
      "grad_norm": 0.6011806600562563,
      "learning_rate": 1.948503735587537e-06,
      "loss": 1.3297,
      "step": 1385
    },
    {
      "epoch": 2.2176,
      "grad_norm": 0.5926795107794901,
      "learning_rate": 1.9411329834506286e-06,
      "loss": 1.3352,
      "step": 1386
    },
    {
      "epoch": 2.2192,
      "grad_norm": 0.5939791257296538,
      "learning_rate": 1.9337728392292104e-06,
      "loss": 1.2969,
      "step": 1387
    },
    {
      "epoch": 2.2208,
      "grad_norm": 0.6062433771420146,
      "learning_rate": 1.926423328447698e-06,
      "loss": 1.238,
      "step": 1388
    },
    {
      "epoch": 2.2224,
      "grad_norm": 0.595723467834509,
      "learning_rate": 1.919084476593631e-06,
      "loss": 1.3237,
      "step": 1389
    },
    {
      "epoch": 2.224,
      "grad_norm": 0.6055238583298779,
      "learning_rate": 1.9117563091175795e-06,
      "loss": 1.3335,
      "step": 1390
    },
    {
      "epoch": 2.2256,
      "grad_norm": 0.5904632093264371,
      "learning_rate": 1.904438851433068e-06,
      "loss": 1.2593,
      "step": 1391
    },
    {
      "epoch": 2.2272,
      "grad_norm": 0.6038115199502553,
      "learning_rate": 1.897132128916474e-06,
      "loss": 1.3183,
      "step": 1392
    },
    {
      "epoch": 2.2288,
      "grad_norm": 0.6063700545480967,
      "learning_rate": 1.8898361669069497e-06,
      "loss": 1.2982,
      "step": 1393
    },
    {
      "epoch": 2.2304,
      "grad_norm": 0.6303800250057439,
      "learning_rate": 1.8825509907063328e-06,
      "loss": 1.2495,
      "step": 1394
    },
    {
      "epoch": 2.232,
      "grad_norm": 0.5897741038730985,
      "learning_rate": 1.87527662557905e-06,
      "loss": 1.2872,
      "step": 1395
    },
    {
      "epoch": 2.2336,
      "grad_norm": 0.5998205740247918,
      "learning_rate": 1.8680130967520433e-06,
      "loss": 1.3015,
      "step": 1396
    },
    {
      "epoch": 2.2352,
      "grad_norm": 0.5937712414917767,
      "learning_rate": 1.8607604294146685e-06,
      "loss": 1.2533,
      "step": 1397
    },
    {
      "epoch": 2.2368,
      "grad_norm": 0.7523509256336717,
      "learning_rate": 1.8535186487186213e-06,
      "loss": 1.2803,
      "step": 1398
    },
    {
      "epoch": 2.2384,
      "grad_norm": 0.6106710322517287,
      "learning_rate": 1.8462877797778367e-06,
      "loss": 1.3116,
      "step": 1399
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.5872998993983781,
      "learning_rate": 1.8390678476684143e-06,
      "loss": 1.2816,
      "step": 1400
    },
    {
      "epoch": 2.2416,
      "grad_norm": 0.5829582905211955,
      "learning_rate": 1.831858877428524e-06,
      "loss": 1.2419,
      "step": 1401
    },
    {
      "epoch": 2.2432,
      "grad_norm": 0.6061899997853949,
      "learning_rate": 1.8246608940583166e-06,
      "loss": 1.2598,
      "step": 1402
    },
    {
      "epoch": 2.2448,
      "grad_norm": 0.6072613965967586,
      "learning_rate": 1.8174739225198485e-06,
      "loss": 1.2972,
      "step": 1403
    },
    {
      "epoch": 2.2464,
      "grad_norm": 0.5681076202414032,
      "learning_rate": 1.8102979877369808e-06,
      "loss": 1.2754,
      "step": 1404
    },
    {
      "epoch": 2.248,
      "grad_norm": 0.6046143952742274,
      "learning_rate": 1.8031331145953047e-06,
      "loss": 1.3302,
      "step": 1405
    },
    {
      "epoch": 2.2496,
      "grad_norm": 0.6184618408031163,
      "learning_rate": 1.7959793279420507e-06,
      "loss": 1.3061,
      "step": 1406
    },
    {
      "epoch": 2.2512,
      "grad_norm": 0.5888074918611663,
      "learning_rate": 1.7888366525859968e-06,
      "loss": 1.2496,
      "step": 1407
    },
    {
      "epoch": 2.2528,
      "grad_norm": 0.6092198138080548,
      "learning_rate": 1.781705113297396e-06,
      "loss": 1.2633,
      "step": 1408
    },
    {
      "epoch": 2.2544,
      "grad_norm": 0.6241041774018643,
      "learning_rate": 1.7745847348078742e-06,
      "loss": 1.2774,
      "step": 1409
    },
    {
      "epoch": 2.2560000000000002,
      "grad_norm": 0.6057515707830918,
      "learning_rate": 1.7674755418103578e-06,
      "loss": 1.2533,
      "step": 1410
    },
    {
      "epoch": 2.2576,
      "grad_norm": 0.5849943299440921,
      "learning_rate": 1.7603775589589821e-06,
      "loss": 1.24,
      "step": 1411
    },
    {
      "epoch": 2.2592,
      "grad_norm": 0.5817127406897575,
      "learning_rate": 1.7532908108690038e-06,
      "loss": 1.2728,
      "step": 1412
    },
    {
      "epoch": 2.2608,
      "grad_norm": 0.5842738267853037,
      "learning_rate": 1.7462153221167222e-06,
      "loss": 1.314,
      "step": 1413
    },
    {
      "epoch": 2.2624,
      "grad_norm": 0.6082984009290396,
      "learning_rate": 1.7391511172393849e-06,
      "loss": 1.2703,
      "step": 1414
    },
    {
      "epoch": 2.2640000000000002,
      "grad_norm": 0.5817703451373666,
      "learning_rate": 1.7320982207351128e-06,
      "loss": 1.256,
      "step": 1415
    },
    {
      "epoch": 2.2656,
      "grad_norm": 0.5869093020733668,
      "learning_rate": 1.7250566570628103e-06,
      "loss": 1.2739,
      "step": 1416
    },
    {
      "epoch": 2.2672,
      "grad_norm": 0.5844048909421451,
      "learning_rate": 1.7180264506420746e-06,
      "loss": 1.2706,
      "step": 1417
    },
    {
      "epoch": 2.2688,
      "grad_norm": 0.6028341752965313,
      "learning_rate": 1.7110076258531244e-06,
      "loss": 1.2763,
      "step": 1418
    },
    {
      "epoch": 2.2704,
      "grad_norm": 0.5850445638234866,
      "learning_rate": 1.7040002070367006e-06,
      "loss": 1.2349,
      "step": 1419
    },
    {
      "epoch": 2.2720000000000002,
      "grad_norm": 0.5989332470420787,
      "learning_rate": 1.6970042184939943e-06,
      "loss": 1.226,
      "step": 1420
    },
    {
      "epoch": 2.2736,
      "grad_norm": 0.6026195174922316,
      "learning_rate": 1.6900196844865575e-06,
      "loss": 1.3187,
      "step": 1421
    },
    {
      "epoch": 2.2752,
      "grad_norm": 0.6072628794059648,
      "learning_rate": 1.683046629236213e-06,
      "loss": 1.2076,
      "step": 1422
    },
    {
      "epoch": 2.2768,
      "grad_norm": 0.6274847185554071,
      "learning_rate": 1.6760850769249837e-06,
      "loss": 1.3304,
      "step": 1423
    },
    {
      "epoch": 2.2784,
      "grad_norm": 0.5985768183638025,
      "learning_rate": 1.669135051694994e-06,
      "loss": 1.2518,
      "step": 1424
    },
    {
      "epoch": 2.2800000000000002,
      "grad_norm": 0.5983431008971208,
      "learning_rate": 1.662196577648398e-06,
      "loss": 1.327,
      "step": 1425
    },
    {
      "epoch": 2.2816,
      "grad_norm": 0.5957440585165525,
      "learning_rate": 1.6552696788472921e-06,
      "loss": 1.2748,
      "step": 1426
    },
    {
      "epoch": 2.2832,
      "grad_norm": 0.5933032310476537,
      "learning_rate": 1.6483543793136247e-06,
      "loss": 1.2089,
      "step": 1427
    },
    {
      "epoch": 2.2848,
      "grad_norm": 0.5926285455021892,
      "learning_rate": 1.6414507030291249e-06,
      "loss": 1.3369,
      "step": 1428
    },
    {
      "epoch": 2.2864,
      "grad_norm": 0.5843230800247733,
      "learning_rate": 1.6345586739352105e-06,
      "loss": 1.2353,
      "step": 1429
    },
    {
      "epoch": 2.288,
      "grad_norm": 0.6375263939084986,
      "learning_rate": 1.6276783159329095e-06,
      "loss": 1.2768,
      "step": 1430
    },
    {
      "epoch": 2.2896,
      "grad_norm": 0.5912913810684892,
      "learning_rate": 1.6208096528827717e-06,
      "loss": 1.2974,
      "step": 1431
    },
    {
      "epoch": 2.2912,
      "grad_norm": 0.5669276257372398,
      "learning_rate": 1.6139527086047929e-06,
      "loss": 1.2104,
      "step": 1432
    },
    {
      "epoch": 2.2928,
      "grad_norm": 0.5907823742640246,
      "learning_rate": 1.6071075068783303e-06,
      "loss": 1.2569,
      "step": 1433
    },
    {
      "epoch": 2.2944,
      "grad_norm": 0.5805763206441698,
      "learning_rate": 1.600274071442014e-06,
      "loss": 1.2477,
      "step": 1434
    },
    {
      "epoch": 2.296,
      "grad_norm": 0.588115411496204,
      "learning_rate": 1.5934524259936757e-06,
      "loss": 1.1945,
      "step": 1435
    },
    {
      "epoch": 2.2976,
      "grad_norm": 0.5876466328123485,
      "learning_rate": 1.5866425941902524e-06,
      "loss": 1.2774,
      "step": 1436
    },
    {
      "epoch": 2.2992,
      "grad_norm": 0.5720067037253567,
      "learning_rate": 1.5798445996477219e-06,
      "loss": 1.2009,
      "step": 1437
    },
    {
      "epoch": 2.3008,
      "grad_norm": 0.5654308403790613,
      "learning_rate": 1.573058465941002e-06,
      "loss": 1.2303,
      "step": 1438
    },
    {
      "epoch": 2.3024,
      "grad_norm": 0.606110260648582,
      "learning_rate": 1.5662842166038844e-06,
      "loss": 1.2481,
      "step": 1439
    },
    {
      "epoch": 2.304,
      "grad_norm": 0.5741199273123376,
      "learning_rate": 1.5595218751289465e-06,
      "loss": 1.2516,
      "step": 1440
    },
    {
      "epoch": 2.3056,
      "grad_norm": 0.5817343252919832,
      "learning_rate": 1.5527714649674641e-06,
      "loss": 1.2805,
      "step": 1441
    },
    {
      "epoch": 2.3072,
      "grad_norm": 0.5876185739609885,
      "learning_rate": 1.5460330095293447e-06,
      "loss": 1.2488,
      "step": 1442
    },
    {
      "epoch": 2.3088,
      "grad_norm": 0.5930620401707185,
      "learning_rate": 1.5393065321830292e-06,
      "loss": 1.2435,
      "step": 1443
    },
    {
      "epoch": 2.3104,
      "grad_norm": 0.5852327850364768,
      "learning_rate": 1.5325920562554259e-06,
      "loss": 1.2295,
      "step": 1444
    },
    {
      "epoch": 2.312,
      "grad_norm": 0.5670052158995248,
      "learning_rate": 1.5258896050318217e-06,
      "loss": 1.2376,
      "step": 1445
    },
    {
      "epoch": 2.3136,
      "grad_norm": 0.5964852835748327,
      "learning_rate": 1.5191992017557994e-06,
      "loss": 1.2713,
      "step": 1446
    },
    {
      "epoch": 2.3152,
      "grad_norm": 0.6110865742600388,
      "learning_rate": 1.512520869629165e-06,
      "loss": 1.2398,
      "step": 1447
    },
    {
      "epoch": 2.3168,
      "grad_norm": 0.5830895136077373,
      "learning_rate": 1.5058546318118583e-06,
      "loss": 1.236,
      "step": 1448
    },
    {
      "epoch": 2.3184,
      "grad_norm": 0.5961703918430933,
      "learning_rate": 1.4992005114218805e-06,
      "loss": 1.2364,
      "step": 1449
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.5686096553614677,
      "learning_rate": 1.4925585315352108e-06,
      "loss": 1.2624,
      "step": 1450
    },
    {
      "epoch": 2.3216,
      "grad_norm": 0.5873353561341068,
      "learning_rate": 1.485928715185721e-06,
      "loss": 1.2385,
      "step": 1451
    },
    {
      "epoch": 2.3232,
      "grad_norm": 0.5854157452395402,
      "learning_rate": 1.4793110853651077e-06,
      "loss": 1.2128,
      "step": 1452
    },
    {
      "epoch": 2.3247999999999998,
      "grad_norm": 0.5850855772343744,
      "learning_rate": 1.472705665022799e-06,
      "loss": 1.273,
      "step": 1453
    },
    {
      "epoch": 2.3264,
      "grad_norm": 0.5927517418088193,
      "learning_rate": 1.4661124770658857e-06,
      "loss": 1.3105,
      "step": 1454
    },
    {
      "epoch": 2.328,
      "grad_norm": 0.5962187752085701,
      "learning_rate": 1.459531544359038e-06,
      "loss": 1.2862,
      "step": 1455
    },
    {
      "epoch": 2.3296,
      "grad_norm": 0.6467463307647424,
      "learning_rate": 1.4529628897244214e-06,
      "loss": 1.2253,
      "step": 1456
    },
    {
      "epoch": 2.3312,
      "grad_norm": 0.5850601922031406,
      "learning_rate": 1.4464065359416274e-06,
      "loss": 1.2251,
      "step": 1457
    },
    {
      "epoch": 2.3327999999999998,
      "grad_norm": 0.6264856673722831,
      "learning_rate": 1.4398625057475845e-06,
      "loss": 1.3281,
      "step": 1458
    },
    {
      "epoch": 2.3344,
      "grad_norm": 0.593124719059151,
      "learning_rate": 1.4333308218364861e-06,
      "loss": 1.2433,
      "step": 1459
    },
    {
      "epoch": 2.336,
      "grad_norm": 0.5857309531828387,
      "learning_rate": 1.4268115068597122e-06,
      "loss": 1.2722,
      "step": 1460
    },
    {
      "epoch": 2.3376,
      "grad_norm": 0.5863012253051065,
      "learning_rate": 1.4203045834257418e-06,
      "loss": 1.306,
      "step": 1461
    },
    {
      "epoch": 2.3392,
      "grad_norm": 0.6051708554791881,
      "learning_rate": 1.4138100741000888e-06,
      "loss": 1.2871,
      "step": 1462
    },
    {
      "epoch": 2.3407999999999998,
      "grad_norm": 0.6175046016766729,
      "learning_rate": 1.4073280014052077e-06,
      "loss": 1.3407,
      "step": 1463
    },
    {
      "epoch": 2.3424,
      "grad_norm": 0.5720692448496913,
      "learning_rate": 1.4008583878204297e-06,
      "loss": 1.2538,
      "step": 1464
    },
    {
      "epoch": 2.344,
      "grad_norm": 0.5864675401363624,
      "learning_rate": 1.3944012557818793e-06,
      "loss": 1.2641,
      "step": 1465
    },
    {
      "epoch": 2.3456,
      "grad_norm": 0.5809285761014837,
      "learning_rate": 1.3879566276823896e-06,
      "loss": 1.2929,
      "step": 1466
    },
    {
      "epoch": 2.3472,
      "grad_norm": 0.6076827487050531,
      "learning_rate": 1.3815245258714393e-06,
      "loss": 1.2158,
      "step": 1467
    },
    {
      "epoch": 2.3487999999999998,
      "grad_norm": 0.5926585469322596,
      "learning_rate": 1.3751049726550587e-06,
      "loss": 1.2576,
      "step": 1468
    },
    {
      "epoch": 2.3504,
      "grad_norm": 0.5974953173060833,
      "learning_rate": 1.368697990295766e-06,
      "loss": 1.2419,
      "step": 1469
    },
    {
      "epoch": 2.352,
      "grad_norm": 0.6187341516212043,
      "learning_rate": 1.3623036010124845e-06,
      "loss": 1.274,
      "step": 1470
    },
    {
      "epoch": 2.3536,
      "grad_norm": 0.6155393991249734,
      "learning_rate": 1.3559218269804624e-06,
      "loss": 1.2125,
      "step": 1471
    },
    {
      "epoch": 2.3552,
      "grad_norm": 0.6434502135560404,
      "learning_rate": 1.3495526903312029e-06,
      "loss": 1.3013,
      "step": 1472
    },
    {
      "epoch": 2.3568,
      "grad_norm": 0.607337027887684,
      "learning_rate": 1.3431962131523796e-06,
      "loss": 1.2972,
      "step": 1473
    },
    {
      "epoch": 2.3584,
      "grad_norm": 0.6105217510937985,
      "learning_rate": 1.3368524174877679e-06,
      "loss": 1.2562,
      "step": 1474
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.573848495251055,
      "learning_rate": 1.330521325337164e-06,
      "loss": 1.3097,
      "step": 1475
    },
    {
      "epoch": 2.3616,
      "grad_norm": 0.6063011004160049,
      "learning_rate": 1.3242029586563054e-06,
      "loss": 1.2942,
      "step": 1476
    },
    {
      "epoch": 2.3632,
      "grad_norm": 0.6016002633370061,
      "learning_rate": 1.3178973393568055e-06,
      "loss": 1.3231,
      "step": 1477
    },
    {
      "epoch": 2.3648,
      "grad_norm": 0.6168886461048435,
      "learning_rate": 1.3116044893060637e-06,
      "loss": 1.3093,
      "step": 1478
    },
    {
      "epoch": 2.3664,
      "grad_norm": 0.5942643720458348,
      "learning_rate": 1.3053244303272022e-06,
      "loss": 1.2567,
      "step": 1479
    },
    {
      "epoch": 2.368,
      "grad_norm": 0.5782965254522152,
      "learning_rate": 1.2990571841989796e-06,
      "loss": 1.2192,
      "step": 1480
    },
    {
      "epoch": 2.3696,
      "grad_norm": 0.60901594278907,
      "learning_rate": 1.2928027726557257e-06,
      "loss": 1.2912,
      "step": 1481
    },
    {
      "epoch": 2.3712,
      "grad_norm": 0.5904296273365327,
      "learning_rate": 1.2865612173872577e-06,
      "loss": 1.2748,
      "step": 1482
    },
    {
      "epoch": 2.3728,
      "grad_norm": 0.5805931137401499,
      "learning_rate": 1.2803325400388095e-06,
      "loss": 1.275,
      "step": 1483
    },
    {
      "epoch": 2.3744,
      "grad_norm": 0.5887141369641821,
      "learning_rate": 1.2741167622109557e-06,
      "loss": 1.2721,
      "step": 1484
    },
    {
      "epoch": 2.376,
      "grad_norm": 0.588923430399329,
      "learning_rate": 1.2679139054595335e-06,
      "loss": 1.2931,
      "step": 1485
    },
    {
      "epoch": 2.3776,
      "grad_norm": 0.5832367876964788,
      "learning_rate": 1.261723991295576e-06,
      "loss": 1.253,
      "step": 1486
    },
    {
      "epoch": 2.3792,
      "grad_norm": 0.6406613698714704,
      "learning_rate": 1.2555470411852262e-06,
      "loss": 1.2733,
      "step": 1487
    },
    {
      "epoch": 2.3808,
      "grad_norm": 0.6139802371712987,
      "learning_rate": 1.2493830765496724e-06,
      "loss": 1.2905,
      "step": 1488
    },
    {
      "epoch": 2.3824,
      "grad_norm": 0.6038424401217812,
      "learning_rate": 1.2432321187650726e-06,
      "loss": 1.2406,
      "step": 1489
    },
    {
      "epoch": 2.384,
      "grad_norm": 0.5751170335514236,
      "learning_rate": 1.237094189162471e-06,
      "loss": 1.2453,
      "step": 1490
    },
    {
      "epoch": 2.3856,
      "grad_norm": 1.1909120798834427,
      "learning_rate": 1.2309693090277392e-06,
      "loss": 1.2326,
      "step": 1491
    },
    {
      "epoch": 2.3872,
      "grad_norm": 0.5626360551012151,
      "learning_rate": 1.2248574996014872e-06,
      "loss": 1.3086,
      "step": 1492
    },
    {
      "epoch": 2.3888,
      "grad_norm": 0.5929660428338006,
      "learning_rate": 1.218758782079001e-06,
      "loss": 1.2887,
      "step": 1493
    },
    {
      "epoch": 2.3904,
      "grad_norm": 0.5668110186980315,
      "learning_rate": 1.2126731776101657e-06,
      "loss": 1.2523,
      "step": 1494
    },
    {
      "epoch": 2.392,
      "grad_norm": 0.6070538460929025,
      "learning_rate": 1.2066007072993856e-06,
      "loss": 1.2961,
      "step": 1495
    },
    {
      "epoch": 2.3936,
      "grad_norm": 0.5830786948175447,
      "learning_rate": 1.2005413922055247e-06,
      "loss": 1.2385,
      "step": 1496
    },
    {
      "epoch": 2.3952,
      "grad_norm": 0.5771514510429985,
      "learning_rate": 1.194495253341818e-06,
      "loss": 1.2481,
      "step": 1497
    },
    {
      "epoch": 2.3968,
      "grad_norm": 0.5694762578718512,
      "learning_rate": 1.1884623116758121e-06,
      "loss": 1.3176,
      "step": 1498
    },
    {
      "epoch": 2.3984,
      "grad_norm": 0.5994080359822693,
      "learning_rate": 1.1824425881292846e-06,
      "loss": 1.2718,
      "step": 1499
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.6021131606712326,
      "learning_rate": 1.1764361035781718e-06,
      "loss": 1.2441,
      "step": 1500
    },
    {
      "epoch": 2.4016,
      "grad_norm": 0.582423894157119,
      "learning_rate": 1.170442878852503e-06,
      "loss": 1.2639,
      "step": 1501
    },
    {
      "epoch": 2.4032,
      "grad_norm": 0.5870076597852442,
      "learning_rate": 1.1644629347363173e-06,
      "loss": 1.2333,
      "step": 1502
    },
    {
      "epoch": 2.4048,
      "grad_norm": 0.5858371121457101,
      "learning_rate": 1.1584962919676024e-06,
      "loss": 1.2515,
      "step": 1503
    },
    {
      "epoch": 2.4064,
      "grad_norm": 0.5645223436018748,
      "learning_rate": 1.1525429712382175e-06,
      "loss": 1.1771,
      "step": 1504
    },
    {
      "epoch": 2.408,
      "grad_norm": 0.5761829668375421,
      "learning_rate": 1.1466029931938182e-06,
      "loss": 1.3198,
      "step": 1505
    },
    {
      "epoch": 2.4096,
      "grad_norm": 0.5839160436028913,
      "learning_rate": 1.1406763784337948e-06,
      "loss": 1.2386,
      "step": 1506
    },
    {
      "epoch": 2.4112,
      "grad_norm": 0.5791472786627343,
      "learning_rate": 1.1347631475111882e-06,
      "loss": 1.2452,
      "step": 1507
    },
    {
      "epoch": 2.4128,
      "grad_norm": 0.5804192094822178,
      "learning_rate": 1.1288633209326288e-06,
      "loss": 1.2743,
      "step": 1508
    },
    {
      "epoch": 2.4144,
      "grad_norm": 0.6138200026204995,
      "learning_rate": 1.122976919158264e-06,
      "loss": 1.2575,
      "step": 1509
    },
    {
      "epoch": 2.416,
      "grad_norm": 0.5829956998350908,
      "learning_rate": 1.1171039626016789e-06,
      "loss": 1.2622,
      "step": 1510
    },
    {
      "epoch": 2.4176,
      "grad_norm": 0.6074407688025479,
      "learning_rate": 1.1112444716298381e-06,
      "loss": 1.3056,
      "step": 1511
    },
    {
      "epoch": 2.4192,
      "grad_norm": 0.6377262466713383,
      "learning_rate": 1.1053984665630025e-06,
      "loss": 1.2965,
      "step": 1512
    },
    {
      "epoch": 2.4208,
      "grad_norm": 0.5801538921917387,
      "learning_rate": 1.0995659676746706e-06,
      "loss": 1.2236,
      "step": 1513
    },
    {
      "epoch": 2.4224,
      "grad_norm": 0.5954247409420429,
      "learning_rate": 1.093746995191497e-06,
      "loss": 1.2429,
      "step": 1514
    },
    {
      "epoch": 2.424,
      "grad_norm": 0.5749588052202351,
      "learning_rate": 1.0879415692932328e-06,
      "loss": 1.2696,
      "step": 1515
    },
    {
      "epoch": 2.4256,
      "grad_norm": 0.5907680629611292,
      "learning_rate": 1.0821497101126487e-06,
      "loss": 1.3256,
      "step": 1516
    },
    {
      "epoch": 2.4272,
      "grad_norm": 0.5822877071437347,
      "learning_rate": 1.076371437735465e-06,
      "loss": 1.214,
      "step": 1517
    },
    {
      "epoch": 2.4288,
      "grad_norm": 0.609491268051046,
      "learning_rate": 1.0706067722002877e-06,
      "loss": 1.2244,
      "step": 1518
    },
    {
      "epoch": 2.4304,
      "grad_norm": 0.5917165709545086,
      "learning_rate": 1.064855733498531e-06,
      "loss": 1.294,
      "step": 1519
    },
    {
      "epoch": 2.432,
      "grad_norm": 0.5784117391285912,
      "learning_rate": 1.0591183415743562e-06,
      "loss": 1.2835,
      "step": 1520
    },
    {
      "epoch": 2.4336,
      "grad_norm": 0.5787106442487396,
      "learning_rate": 1.0533946163245984e-06,
      "loss": 1.282,
      "step": 1521
    },
    {
      "epoch": 2.4352,
      "grad_norm": 0.5770579689025486,
      "learning_rate": 1.047684577598694e-06,
      "loss": 1.2919,
      "step": 1522
    },
    {
      "epoch": 2.4368,
      "grad_norm": 0.5920100549990316,
      "learning_rate": 1.0419882451986197e-06,
      "loss": 1.2505,
      "step": 1523
    },
    {
      "epoch": 2.4384,
      "grad_norm": 0.5713563145659873,
      "learning_rate": 1.0363056388788162e-06,
      "loss": 1.2239,
      "step": 1524
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.5581966840953551,
      "learning_rate": 1.0306367783461258e-06,
      "loss": 1.1804,
      "step": 1525
    },
    {
      "epoch": 2.4416,
      "grad_norm": 0.5798395250538267,
      "learning_rate": 1.024981683259723e-06,
      "loss": 1.2324,
      "step": 1526
    },
    {
      "epoch": 2.4432,
      "grad_norm": 0.5774239931246444,
      "learning_rate": 1.0193403732310392e-06,
      "loss": 1.2176,
      "step": 1527
    },
    {
      "epoch": 2.4448,
      "grad_norm": 0.5687799078955014,
      "learning_rate": 1.0137128678237062e-06,
      "loss": 1.289,
      "step": 1528
    },
    {
      "epoch": 2.4464,
      "grad_norm": 0.5676797299012579,
      "learning_rate": 1.0080991865534773e-06,
      "loss": 1.2817,
      "step": 1529
    },
    {
      "epoch": 2.448,
      "grad_norm": 0.5622556466918845,
      "learning_rate": 1.002499348888169e-06,
      "loss": 1.2342,
      "step": 1530
    },
    {
      "epoch": 2.4496,
      "grad_norm": 0.5660176982848062,
      "learning_rate": 9.969133742475883e-07,
      "loss": 1.2961,
      "step": 1531
    },
    {
      "epoch": 2.4512,
      "grad_norm": 0.5727078034675379,
      "learning_rate": 9.913412820034629e-07,
      "loss": 1.2543,
      "step": 1532
    },
    {
      "epoch": 2.4528,
      "grad_norm": 0.594710916734317,
      "learning_rate": 9.857830914793827e-07,
      "loss": 1.1825,
      "step": 1533
    },
    {
      "epoch": 2.4544,
      "grad_norm": 0.598906817973598,
      "learning_rate": 9.802388219507215e-07,
      "loss": 1.2619,
      "step": 1534
    },
    {
      "epoch": 2.456,
      "grad_norm": 0.5679240925575452,
      "learning_rate": 9.747084926445839e-07,
      "loss": 1.2246,
      "step": 1535
    },
    {
      "epoch": 2.4576000000000002,
      "grad_norm": 0.5960787241293773,
      "learning_rate": 9.691921227397227e-07,
      "loss": 1.2466,
      "step": 1536
    },
    {
      "epoch": 2.4592,
      "grad_norm": 0.5929037319996285,
      "learning_rate": 9.63689731366486e-07,
      "loss": 1.246,
      "step": 1537
    },
    {
      "epoch": 2.4608,
      "grad_norm": 0.5667904177192885,
      "learning_rate": 9.58201337606745e-07,
      "loss": 1.3073,
      "step": 1538
    },
    {
      "epoch": 2.4624,
      "grad_norm": 0.5653620921283392,
      "learning_rate": 9.527269604938249e-07,
      "loss": 1.2518,
      "step": 1539
    },
    {
      "epoch": 2.464,
      "grad_norm": 0.6130592860947571,
      "learning_rate": 9.472666190124457e-07,
      "loss": 1.2935,
      "step": 1540
    },
    {
      "epoch": 2.4656000000000002,
      "grad_norm": 0.5693456703401221,
      "learning_rate": 9.418203320986502e-07,
      "loss": 1.2387,
      "step": 1541
    },
    {
      "epoch": 2.4672,
      "grad_norm": 0.6000173889942658,
      "learning_rate": 9.363881186397434e-07,
      "loss": 1.2372,
      "step": 1542
    },
    {
      "epoch": 2.4688,
      "grad_norm": 0.5855477992480824,
      "learning_rate": 9.309699974742243e-07,
      "loss": 1.2925,
      "step": 1543
    },
    {
      "epoch": 2.4704,
      "grad_norm": 0.5704859778042619,
      "learning_rate": 9.255659873917183e-07,
      "loss": 1.2067,
      "step": 1544
    },
    {
      "epoch": 2.472,
      "grad_norm": 0.5670682778563786,
      "learning_rate": 9.201761071329196e-07,
      "loss": 1.3266,
      "step": 1545
    },
    {
      "epoch": 2.4736000000000002,
      "grad_norm": 0.5821544292961685,
      "learning_rate": 9.148003753895146e-07,
      "loss": 1.2935,
      "step": 1546
    },
    {
      "epoch": 2.4752,
      "grad_norm": 0.5669999213442118,
      "learning_rate": 9.094388108041302e-07,
      "loss": 1.31,
      "step": 1547
    },
    {
      "epoch": 2.4768,
      "grad_norm": 0.5853563134238543,
      "learning_rate": 9.040914319702598e-07,
      "loss": 1.3308,
      "step": 1548
    },
    {
      "epoch": 2.4784,
      "grad_norm": 0.564214272697882,
      "learning_rate": 8.987582574321996e-07,
      "loss": 1.2669,
      "step": 1549
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.5837206458396095,
      "learning_rate": 8.934393056849921e-07,
      "loss": 1.3091,
      "step": 1550
    },
    {
      "epoch": 2.4816,
      "grad_norm": 0.5708241014677848,
      "learning_rate": 8.881345951743486e-07,
      "loss": 1.2965,
      "step": 1551
    },
    {
      "epoch": 2.4832,
      "grad_norm": 0.5753140958223908,
      "learning_rate": 8.828441442966013e-07,
      "loss": 1.2102,
      "step": 1552
    },
    {
      "epoch": 2.4848,
      "grad_norm": 0.588317987834062,
      "learning_rate": 8.775679713986235e-07,
      "loss": 1.2896,
      "step": 1553
    },
    {
      "epoch": 2.4864,
      "grad_norm": 0.5822708645030097,
      "learning_rate": 8.723060947777778e-07,
      "loss": 1.271,
      "step": 1554
    },
    {
      "epoch": 2.488,
      "grad_norm": 0.5777922059144108,
      "learning_rate": 8.670585326818493e-07,
      "loss": 1.2049,
      "step": 1555
    },
    {
      "epoch": 2.4896,
      "grad_norm": 0.5678851585906104,
      "learning_rate": 8.618253033089768e-07,
      "loss": 1.2433,
      "step": 1556
    },
    {
      "epoch": 2.4912,
      "grad_norm": 0.5706069376649251,
      "learning_rate": 8.566064248076001e-07,
      "loss": 1.256,
      "step": 1557
    },
    {
      "epoch": 2.4928,
      "grad_norm": 0.5995073050373722,
      "learning_rate": 8.514019152763852e-07,
      "loss": 1.2848,
      "step": 1558
    },
    {
      "epoch": 2.4944,
      "grad_norm": 0.5726815422351208,
      "learning_rate": 8.462117927641733e-07,
      "loss": 1.319,
      "step": 1559
    },
    {
      "epoch": 2.496,
      "grad_norm": 0.5693047025782862,
      "learning_rate": 8.410360752699099e-07,
      "loss": 1.2044,
      "step": 1560
    },
    {
      "epoch": 2.4976,
      "grad_norm": 0.5683858492256759,
      "learning_rate": 8.358747807425827e-07,
      "loss": 1.3086,
      "step": 1561
    },
    {
      "epoch": 2.4992,
      "grad_norm": 0.5771284278447932,
      "learning_rate": 8.307279270811675e-07,
      "loss": 1.2646,
      "step": 1562
    },
    {
      "epoch": 2.5008,
      "grad_norm": 0.5674305645087373,
      "learning_rate": 8.255955321345533e-07,
      "loss": 1.2724,
      "step": 1563
    },
    {
      "epoch": 2.5023999999999997,
      "grad_norm": 0.562654481710765,
      "learning_rate": 8.20477613701493e-07,
      "loss": 1.3096,
      "step": 1564
    },
    {
      "epoch": 2.504,
      "grad_norm": 0.5612401024063585,
      "learning_rate": 8.153741895305351e-07,
      "loss": 1.2561,
      "step": 1565
    },
    {
      "epoch": 2.5056000000000003,
      "grad_norm": 0.574898145297722,
      "learning_rate": 8.102852773199588e-07,
      "loss": 1.2478,
      "step": 1566
    },
    {
      "epoch": 2.5072,
      "grad_norm": 0.5828204950706997,
      "learning_rate": 8.052108947177234e-07,
      "loss": 1.2618,
      "step": 1567
    },
    {
      "epoch": 2.5088,
      "grad_norm": 0.574137960054832,
      "learning_rate": 8.001510593213946e-07,
      "loss": 1.2409,
      "step": 1568
    },
    {
      "epoch": 2.5103999999999997,
      "grad_norm": 0.5821540567523107,
      "learning_rate": 7.951057886780939e-07,
      "loss": 1.3258,
      "step": 1569
    },
    {
      "epoch": 2.512,
      "grad_norm": 1.1728604710698605,
      "learning_rate": 7.900751002844326e-07,
      "loss": 1.2944,
      "step": 1570
    },
    {
      "epoch": 2.5136,
      "grad_norm": 0.5878645674319394,
      "learning_rate": 7.850590115864481e-07,
      "loss": 1.2749,
      "step": 1571
    },
    {
      "epoch": 2.5152,
      "grad_norm": 0.5987041677221565,
      "learning_rate": 7.80057539979554e-07,
      "loss": 1.2072,
      "step": 1572
    },
    {
      "epoch": 2.5168,
      "grad_norm": 0.5808464262431403,
      "learning_rate": 7.750707028084653e-07,
      "loss": 1.3058,
      "step": 1573
    },
    {
      "epoch": 2.5183999999999997,
      "grad_norm": 0.5575087712350061,
      "learning_rate": 7.70098517367151e-07,
      "loss": 1.2971,
      "step": 1574
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.5783137797365073,
      "learning_rate": 7.651410008987698e-07,
      "loss": 1.2931,
      "step": 1575
    },
    {
      "epoch": 2.5216,
      "grad_norm": 0.5712141862878561,
      "learning_rate": 7.601981705956041e-07,
      "loss": 1.235,
      "step": 1576
    },
    {
      "epoch": 2.5232,
      "grad_norm": 0.5493098603486967,
      "learning_rate": 7.552700435990123e-07,
      "loss": 1.2295,
      "step": 1577
    },
    {
      "epoch": 2.5248,
      "grad_norm": 0.5610422024093158,
      "learning_rate": 7.503566369993564e-07,
      "loss": 1.204,
      "step": 1578
    },
    {
      "epoch": 2.5263999999999998,
      "grad_norm": 0.5928744244056257,
      "learning_rate": 7.454579678359547e-07,
      "loss": 1.29,
      "step": 1579
    },
    {
      "epoch": 2.528,
      "grad_norm": 0.579071835290819,
      "learning_rate": 7.405740530970157e-07,
      "loss": 1.2875,
      "step": 1580
    },
    {
      "epoch": 2.5296,
      "grad_norm": 0.5565694167410128,
      "learning_rate": 7.357049097195773e-07,
      "loss": 1.2091,
      "step": 1581
    },
    {
      "epoch": 2.5312,
      "grad_norm": 0.5672805504407683,
      "learning_rate": 7.308505545894567e-07,
      "loss": 1.3081,
      "step": 1582
    },
    {
      "epoch": 2.5328,
      "grad_norm": 0.5776012469491723,
      "learning_rate": 7.260110045411816e-07,
      "loss": 1.3051,
      "step": 1583
    },
    {
      "epoch": 2.5343999999999998,
      "grad_norm": 0.5841423827083205,
      "learning_rate": 7.211862763579414e-07,
      "loss": 1.2314,
      "step": 1584
    },
    {
      "epoch": 2.536,
      "grad_norm": 0.5598586494647468,
      "learning_rate": 7.163763867715218e-07,
      "loss": 1.2352,
      "step": 1585
    },
    {
      "epoch": 2.5376,
      "grad_norm": 0.5602620595885501,
      "learning_rate": 7.115813524622489e-07,
      "loss": 1.2182,
      "step": 1586
    },
    {
      "epoch": 2.5392,
      "grad_norm": 0.557363659984121,
      "learning_rate": 7.068011900589333e-07,
      "loss": 1.2593,
      "step": 1587
    },
    {
      "epoch": 2.5408,
      "grad_norm": 0.5879422414473574,
      "learning_rate": 7.020359161388108e-07,
      "loss": 1.251,
      "step": 1588
    },
    {
      "epoch": 2.5423999999999998,
      "grad_norm": 0.5624483427693214,
      "learning_rate": 6.972855472274853e-07,
      "loss": 1.2893,
      "step": 1589
    },
    {
      "epoch": 2.544,
      "grad_norm": 0.5560200356668247,
      "learning_rate": 6.925500997988694e-07,
      "loss": 1.2701,
      "step": 1590
    },
    {
      "epoch": 2.5456,
      "grad_norm": 0.5851809310679791,
      "learning_rate": 6.87829590275132e-07,
      "loss": 1.2877,
      "step": 1591
    },
    {
      "epoch": 2.5472,
      "grad_norm": 0.5747395307929473,
      "learning_rate": 6.83124035026635e-07,
      "loss": 1.2612,
      "step": 1592
    },
    {
      "epoch": 2.5488,
      "grad_norm": 0.5657171479902344,
      "learning_rate": 6.784334503718826e-07,
      "loss": 1.2702,
      "step": 1593
    },
    {
      "epoch": 2.5504,
      "grad_norm": 0.5549242521440277,
      "learning_rate": 6.737578525774636e-07,
      "loss": 1.1815,
      "step": 1594
    },
    {
      "epoch": 2.552,
      "grad_norm": 0.6004742530549338,
      "learning_rate": 6.690972578579886e-07,
      "loss": 1.2684,
      "step": 1595
    },
    {
      "epoch": 2.5536,
      "grad_norm": 0.5822295631148511,
      "learning_rate": 6.644516823760439e-07,
      "loss": 1.2984,
      "step": 1596
    },
    {
      "epoch": 2.5552,
      "grad_norm": 0.5860866214411039,
      "learning_rate": 6.598211422421258e-07,
      "loss": 1.2927,
      "step": 1597
    },
    {
      "epoch": 2.5568,
      "grad_norm": 0.5831674489606492,
      "learning_rate": 6.552056535145917e-07,
      "loss": 1.2707,
      "step": 1598
    },
    {
      "epoch": 2.5584,
      "grad_norm": 0.5872752633799103,
      "learning_rate": 6.506052321996037e-07,
      "loss": 1.2795,
      "step": 1599
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.5810429581288002,
      "learning_rate": 6.46019894251066e-07,
      "loss": 1.2235,
      "step": 1600
    },
    {
      "epoch": 2.5616,
      "grad_norm": 0.56991419917998,
      "learning_rate": 6.414496555705802e-07,
      "loss": 1.2201,
      "step": 1601
    },
    {
      "epoch": 2.5632,
      "grad_norm": 0.6092038741703663,
      "learning_rate": 6.368945320073799e-07,
      "loss": 1.3101,
      "step": 1602
    },
    {
      "epoch": 2.5648,
      "grad_norm": 0.5678285050058818,
      "learning_rate": 6.323545393582847e-07,
      "loss": 1.2413,
      "step": 1603
    },
    {
      "epoch": 2.5664,
      "grad_norm": 0.5827396792066857,
      "learning_rate": 6.278296933676414e-07,
      "loss": 1.3003,
      "step": 1604
    },
    {
      "epoch": 2.568,
      "grad_norm": 0.571433480403834,
      "learning_rate": 6.233200097272646e-07,
      "loss": 1.2464,
      "step": 1605
    },
    {
      "epoch": 2.5696,
      "grad_norm": 0.5820761084086447,
      "learning_rate": 6.188255040763929e-07,
      "loss": 1.2064,
      "step": 1606
    },
    {
      "epoch": 2.5712,
      "grad_norm": 0.5652495685159182,
      "learning_rate": 6.143461920016247e-07,
      "loss": 1.2415,
      "step": 1607
    },
    {
      "epoch": 2.5728,
      "grad_norm": 0.5634732989419206,
      "learning_rate": 6.098820890368696e-07,
      "loss": 1.2832,
      "step": 1608
    },
    {
      "epoch": 2.5744,
      "grad_norm": 0.666957518114739,
      "learning_rate": 6.054332106632943e-07,
      "loss": 1.2177,
      "step": 1609
    },
    {
      "epoch": 2.576,
      "grad_norm": 0.6222044887124286,
      "learning_rate": 6.009995723092655e-07,
      "loss": 1.2674,
      "step": 1610
    },
    {
      "epoch": 2.5776,
      "grad_norm": 0.5509431708145212,
      "learning_rate": 5.965811893503015e-07,
      "loss": 1.222,
      "step": 1611
    },
    {
      "epoch": 2.5792,
      "grad_norm": 0.5832129819118231,
      "learning_rate": 5.921780771090124e-07,
      "loss": 1.3018,
      "step": 1612
    },
    {
      "epoch": 2.5808,
      "grad_norm": 0.5632224001598489,
      "learning_rate": 5.877902508550542e-07,
      "loss": 1.2623,
      "step": 1613
    },
    {
      "epoch": 2.5824,
      "grad_norm": 0.5459341281789456,
      "learning_rate": 5.834177258050711e-07,
      "loss": 1.237,
      "step": 1614
    },
    {
      "epoch": 2.584,
      "grad_norm": 0.5536519444648579,
      "learning_rate": 5.790605171226421e-07,
      "loss": 1.2285,
      "step": 1615
    },
    {
      "epoch": 2.5856,
      "grad_norm": 0.559192174763866,
      "learning_rate": 5.747186399182336e-07,
      "loss": 1.2311,
      "step": 1616
    },
    {
      "epoch": 2.5872,
      "grad_norm": 0.8931994716737411,
      "learning_rate": 5.703921092491393e-07,
      "loss": 1.2505,
      "step": 1617
    },
    {
      "epoch": 2.5888,
      "grad_norm": 0.5627657404312993,
      "learning_rate": 5.660809401194362e-07,
      "loss": 1.2486,
      "step": 1618
    },
    {
      "epoch": 2.5904,
      "grad_norm": 0.5862160343534385,
      "learning_rate": 5.617851474799285e-07,
      "loss": 1.2514,
      "step": 1619
    },
    {
      "epoch": 2.592,
      "grad_norm": 0.5542149026742876,
      "learning_rate": 5.575047462280919e-07,
      "loss": 1.2995,
      "step": 1620
    },
    {
      "epoch": 2.5936,
      "grad_norm": 0.5871562970609224,
      "learning_rate": 5.532397512080306e-07,
      "loss": 1.2755,
      "step": 1621
    },
    {
      "epoch": 2.5952,
      "grad_norm": 0.5645173020235477,
      "learning_rate": 5.489901772104178e-07,
      "loss": 1.285,
      "step": 1622
    },
    {
      "epoch": 2.5968,
      "grad_norm": 0.5870737023400517,
      "learning_rate": 5.447560389724499e-07,
      "loss": 1.2475,
      "step": 1623
    },
    {
      "epoch": 2.5984,
      "grad_norm": 0.5839991063563083,
      "learning_rate": 5.405373511777939e-07,
      "loss": 1.2822,
      "step": 1624
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.5802256692057104,
      "learning_rate": 5.363341284565316e-07,
      "loss": 1.2913,
      "step": 1625
    },
    {
      "epoch": 2.6016,
      "grad_norm": 1.2074462823929517,
      "learning_rate": 5.321463853851189e-07,
      "loss": 1.2491,
      "step": 1626
    },
    {
      "epoch": 2.6032,
      "grad_norm": 0.5984986583100685,
      "learning_rate": 5.279741364863244e-07,
      "loss": 1.2502,
      "step": 1627
    },
    {
      "epoch": 2.6048,
      "grad_norm": 0.587120282341044,
      "learning_rate": 5.238173962291881e-07,
      "loss": 1.2675,
      "step": 1628
    },
    {
      "epoch": 2.6064,
      "grad_norm": 0.5864668430348661,
      "learning_rate": 5.196761790289639e-07,
      "loss": 1.2496,
      "step": 1629
    },
    {
      "epoch": 2.608,
      "grad_norm": 0.5867273962760853,
      "learning_rate": 5.155504992470751e-07,
      "loss": 1.2359,
      "step": 1630
    },
    {
      "epoch": 2.6096,
      "grad_norm": 0.5559421444031886,
      "learning_rate": 5.114403711910631e-07,
      "loss": 1.2598,
      "step": 1631
    },
    {
      "epoch": 2.6112,
      "grad_norm": 0.5707270739039474,
      "learning_rate": 5.073458091145328e-07,
      "loss": 1.2348,
      "step": 1632
    },
    {
      "epoch": 2.6128,
      "grad_norm": 0.5576123223263122,
      "learning_rate": 5.032668272171138e-07,
      "loss": 1.2483,
      "step": 1633
    },
    {
      "epoch": 2.6144,
      "grad_norm": 0.5733813019383482,
      "learning_rate": 4.99203439644399e-07,
      "loss": 1.2724,
      "step": 1634
    },
    {
      "epoch": 2.616,
      "grad_norm": 0.5732459049425761,
      "learning_rate": 4.951556604879049e-07,
      "loss": 1.2868,
      "step": 1635
    },
    {
      "epoch": 2.6176,
      "grad_norm": 0.5643912241668058,
      "learning_rate": 4.911235037850187e-07,
      "loss": 1.2677,
      "step": 1636
    },
    {
      "epoch": 2.6192,
      "grad_norm": 0.5865066144164197,
      "learning_rate": 4.871069835189485e-07,
      "loss": 1.2919,
      "step": 1637
    },
    {
      "epoch": 2.6208,
      "grad_norm": 0.5662109805646789,
      "learning_rate": 4.831061136186787e-07,
      "loss": 1.1685,
      "step": 1638
    },
    {
      "epoch": 2.6224,
      "grad_norm": 0.5551216702842707,
      "learning_rate": 4.791209079589165e-07,
      "loss": 1.2778,
      "step": 1639
    },
    {
      "epoch": 2.624,
      "grad_norm": 0.5708860551955073,
      "learning_rate": 4.7515138036005157e-07,
      "loss": 1.2661,
      "step": 1640
    },
    {
      "epoch": 2.6256,
      "grad_norm": 0.5811670934541747,
      "learning_rate": 4.7119754458809727e-07,
      "loss": 1.3375,
      "step": 1641
    },
    {
      "epoch": 2.6272,
      "grad_norm": 0.5682491306060832,
      "learning_rate": 4.672594143546538e-07,
      "loss": 1.2239,
      "step": 1642
    },
    {
      "epoch": 2.6288,
      "grad_norm": 0.5962951070876662,
      "learning_rate": 4.6333700331685385e-07,
      "loss": 1.2669,
      "step": 1643
    },
    {
      "epoch": 2.6304,
      "grad_norm": 0.5711165390078929,
      "learning_rate": 4.594303250773152e-07,
      "loss": 1.3147,
      "step": 1644
    },
    {
      "epoch": 2.632,
      "grad_norm": 0.5874152209842891,
      "learning_rate": 4.555393931841001e-07,
      "loss": 1.2807,
      "step": 1645
    },
    {
      "epoch": 2.6336,
      "grad_norm": 0.5882847808261812,
      "learning_rate": 4.5166422113065877e-07,
      "loss": 1.2751,
      "step": 1646
    },
    {
      "epoch": 2.6352,
      "grad_norm": 0.5648061955984504,
      "learning_rate": 4.478048223557907e-07,
      "loss": 1.2905,
      "step": 1647
    },
    {
      "epoch": 2.6368,
      "grad_norm": 0.5412842721674558,
      "learning_rate": 4.439612102435942e-07,
      "loss": 1.2288,
      "step": 1648
    },
    {
      "epoch": 2.6384,
      "grad_norm": 0.552898410060892,
      "learning_rate": 4.401333981234196e-07,
      "loss": 1.2822,
      "step": 1649
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.552246147049269,
      "learning_rate": 4.3632139926982676e-07,
      "loss": 1.2533,
      "step": 1650
    },
    {
      "epoch": 2.6416,
      "grad_norm": 0.5601183398680893,
      "learning_rate": 4.325252269025315e-07,
      "loss": 1.2583,
      "step": 1651
    },
    {
      "epoch": 2.6432,
      "grad_norm": 0.5956624516072417,
      "learning_rate": 4.287448941863692e-07,
      "loss": 1.2704,
      "step": 1652
    },
    {
      "epoch": 2.6448,
      "grad_norm": 0.5876926408456438,
      "learning_rate": 4.249804142312436e-07,
      "loss": 1.3105,
      "step": 1653
    },
    {
      "epoch": 2.6464,
      "grad_norm": 0.5507143336439376,
      "learning_rate": 4.2123180009207956e-07,
      "loss": 1.2677,
      "step": 1654
    },
    {
      "epoch": 2.648,
      "grad_norm": 0.5469065231044178,
      "learning_rate": 4.1749906476878486e-07,
      "loss": 1.2286,
      "step": 1655
    },
    {
      "epoch": 2.6496,
      "grad_norm": 0.5763311040568029,
      "learning_rate": 4.137822212061965e-07,
      "loss": 1.2944,
      "step": 1656
    },
    {
      "epoch": 2.6512000000000002,
      "grad_norm": 0.5527799275325929,
      "learning_rate": 4.100812822940431e-07,
      "loss": 1.2481,
      "step": 1657
    },
    {
      "epoch": 2.6528,
      "grad_norm": 0.5519231442964638,
      "learning_rate": 4.063962608668959e-07,
      "loss": 1.274,
      "step": 1658
    },
    {
      "epoch": 2.6544,
      "grad_norm": 0.574365653023745,
      "learning_rate": 4.0272716970412516e-07,
      "loss": 1.3407,
      "step": 1659
    },
    {
      "epoch": 2.656,
      "grad_norm": 0.5553200900789331,
      "learning_rate": 3.990740215298583e-07,
      "loss": 1.2553,
      "step": 1660
    },
    {
      "epoch": 2.6576,
      "grad_norm": 0.5625604583117277,
      "learning_rate": 3.954368290129301e-07,
      "loss": 1.2068,
      "step": 1661
    },
    {
      "epoch": 2.6592000000000002,
      "grad_norm": 0.5668384326815948,
      "learning_rate": 3.918156047668453e-07,
      "loss": 1.2723,
      "step": 1662
    },
    {
      "epoch": 2.6608,
      "grad_norm": 0.5697170884385802,
      "learning_rate": 3.882103613497318e-07,
      "loss": 1.3314,
      "step": 1663
    },
    {
      "epoch": 2.6624,
      "grad_norm": 0.5789392960420602,
      "learning_rate": 3.84621111264295e-07,
      "loss": 1.2671,
      "step": 1664
    },
    {
      "epoch": 2.664,
      "grad_norm": 0.575361956644775,
      "learning_rate": 3.810478669577794e-07,
      "loss": 1.282,
      "step": 1665
    },
    {
      "epoch": 2.6656,
      "grad_norm": 0.5605846348265893,
      "learning_rate": 3.7749064082191976e-07,
      "loss": 1.2646,
      "step": 1666
    },
    {
      "epoch": 2.6672000000000002,
      "grad_norm": 0.5647974023767571,
      "learning_rate": 3.739494451929049e-07,
      "loss": 1.2092,
      "step": 1667
    },
    {
      "epoch": 2.6688,
      "grad_norm": 0.5552916411247354,
      "learning_rate": 3.7042429235132625e-07,
      "loss": 1.2308,
      "step": 1668
    },
    {
      "epoch": 2.6704,
      "grad_norm": 0.5428390277202814,
      "learning_rate": 3.6691519452214387e-07,
      "loss": 1.2503,
      "step": 1669
    },
    {
      "epoch": 2.672,
      "grad_norm": 0.5631331084664957,
      "learning_rate": 3.6342216387464047e-07,
      "loss": 1.3196,
      "step": 1670
    },
    {
      "epoch": 2.6736,
      "grad_norm": 0.5774268895579237,
      "learning_rate": 3.5994521252237516e-07,
      "loss": 1.2659,
      "step": 1671
    },
    {
      "epoch": 2.6752000000000002,
      "grad_norm": 0.5483083627550193,
      "learning_rate": 3.564843525231498e-07,
      "loss": 1.2023,
      "step": 1672
    },
    {
      "epoch": 2.6768,
      "grad_norm": 0.5658439286411029,
      "learning_rate": 3.53039595878959e-07,
      "loss": 1.3257,
      "step": 1673
    },
    {
      "epoch": 2.6784,
      "grad_norm": 0.5718307645632935,
      "learning_rate": 3.496109545359544e-07,
      "loss": 1.2167,
      "step": 1674
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.5519930983826912,
      "learning_rate": 3.461984403844015e-07,
      "loss": 1.2286,
      "step": 1675
    },
    {
      "epoch": 2.6816,
      "grad_norm": 0.5433896968883223,
      "learning_rate": 3.42802065258635e-07,
      "loss": 1.2341,
      "step": 1676
    },
    {
      "epoch": 2.6832000000000003,
      "grad_norm": 0.580487840713298,
      "learning_rate": 3.394218409370242e-07,
      "loss": 1.2705,
      "step": 1677
    },
    {
      "epoch": 2.6848,
      "grad_norm": 0.5579893247221943,
      "learning_rate": 3.360577791419256e-07,
      "loss": 1.2469,
      "step": 1678
    },
    {
      "epoch": 2.6864,
      "grad_norm": 0.5633246606372273,
      "learning_rate": 3.3270989153964707e-07,
      "loss": 1.3014,
      "step": 1679
    },
    {
      "epoch": 2.6879999999999997,
      "grad_norm": 0.5662956338010805,
      "learning_rate": 3.2937818974040637e-07,
      "loss": 1.2344,
      "step": 1680
    },
    {
      "epoch": 2.6896,
      "grad_norm": 0.7573685694519351,
      "learning_rate": 3.260626852982873e-07,
      "loss": 1.2521,
      "step": 1681
    },
    {
      "epoch": 2.6912000000000003,
      "grad_norm": 0.5556622878337115,
      "learning_rate": 3.227633897112059e-07,
      "loss": 1.2573,
      "step": 1682
    },
    {
      "epoch": 2.6928,
      "grad_norm": 0.5554179197115993,
      "learning_rate": 3.194803144208636e-07,
      "loss": 1.2267,
      "step": 1683
    },
    {
      "epoch": 2.6944,
      "grad_norm": 0.5367089556383667,
      "learning_rate": 3.16213470812714e-07,
      "loss": 1.1966,
      "step": 1684
    },
    {
      "epoch": 2.6959999999999997,
      "grad_norm": 0.5648303648884356,
      "learning_rate": 3.129628702159204e-07,
      "loss": 1.2434,
      "step": 1685
    },
    {
      "epoch": 2.6976,
      "grad_norm": 0.5491437923897278,
      "learning_rate": 3.097285239033138e-07,
      "loss": 1.2732,
      "step": 1686
    },
    {
      "epoch": 2.6992000000000003,
      "grad_norm": 0.5785176316387552,
      "learning_rate": 3.0651044309136016e-07,
      "loss": 1.2652,
      "step": 1687
    },
    {
      "epoch": 2.7008,
      "grad_norm": 0.5701671068994607,
      "learning_rate": 3.033086389401141e-07,
      "loss": 1.2698,
      "step": 1688
    },
    {
      "epoch": 2.7024,
      "grad_norm": 0.5717653276488596,
      "learning_rate": 3.0012312255318696e-07,
      "loss": 1.3355,
      "step": 1689
    },
    {
      "epoch": 2.7039999999999997,
      "grad_norm": 0.5465817035541323,
      "learning_rate": 2.9695390497770535e-07,
      "loss": 1.2156,
      "step": 1690
    },
    {
      "epoch": 2.7056,
      "grad_norm": 0.5494048165728074,
      "learning_rate": 2.93800997204271e-07,
      "loss": 1.3107,
      "step": 1691
    },
    {
      "epoch": 2.7072000000000003,
      "grad_norm": 0.5720085061449189,
      "learning_rate": 2.9066441016692594e-07,
      "loss": 1.2551,
      "step": 1692
    },
    {
      "epoch": 2.7088,
      "grad_norm": 0.548621790664392,
      "learning_rate": 2.8754415474311235e-07,
      "loss": 1.2362,
      "step": 1693
    },
    {
      "epoch": 2.7104,
      "grad_norm": 0.5735756358210508,
      "learning_rate": 2.844402417536374e-07,
      "loss": 1.2783,
      "step": 1694
    },
    {
      "epoch": 2.7119999999999997,
      "grad_norm": 0.5434770319514175,
      "learning_rate": 2.8135268196263055e-07,
      "loss": 1.28,
      "step": 1695
    },
    {
      "epoch": 2.7136,
      "grad_norm": 0.576062139312047,
      "learning_rate": 2.782814860775124e-07,
      "loss": 1.3153,
      "step": 1696
    },
    {
      "epoch": 2.7152,
      "grad_norm": 0.5450779996652965,
      "learning_rate": 2.752266647489549e-07,
      "loss": 1.2326,
      "step": 1697
    },
    {
      "epoch": 2.7168,
      "grad_norm": 0.5759278383761401,
      "learning_rate": 2.7218822857084217e-07,
      "loss": 1.2707,
      "step": 1698
    },
    {
      "epoch": 2.7184,
      "grad_norm": 0.5707254998837294,
      "learning_rate": 2.691661880802382e-07,
      "loss": 1.2907,
      "step": 1699
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 0.542082094311642,
      "learning_rate": 2.661605537573453e-07,
      "loss": 1.2493,
      "step": 1700
    },
    {
      "epoch": 2.7216,
      "grad_norm": 0.5423809812096312,
      "learning_rate": 2.631713360254734e-07,
      "loss": 1.2198,
      "step": 1701
    },
    {
      "epoch": 2.7232,
      "grad_norm": 0.5547231467147336,
      "learning_rate": 2.6019854525099977e-07,
      "loss": 1.2955,
      "step": 1702
    },
    {
      "epoch": 2.7248,
      "grad_norm": 0.5625323995472893,
      "learning_rate": 2.572421917433332e-07,
      "loss": 1.2665,
      "step": 1703
    },
    {
      "epoch": 2.7264,
      "grad_norm": 0.5528688557609134,
      "learning_rate": 2.5430228575488156e-07,
      "loss": 1.2726,
      "step": 1704
    },
    {
      "epoch": 2.7279999999999998,
      "grad_norm": 0.5516910005492746,
      "learning_rate": 2.513788374810111e-07,
      "loss": 1.2998,
      "step": 1705
    },
    {
      "epoch": 2.7296,
      "grad_norm": 0.5557198332723496,
      "learning_rate": 2.4847185706001643e-07,
      "loss": 1.2751,
      "step": 1706
    },
    {
      "epoch": 2.7312,
      "grad_norm": 0.5638962599940763,
      "learning_rate": 2.455813545730812e-07,
      "loss": 1.2476,
      "step": 1707
    },
    {
      "epoch": 2.7328,
      "grad_norm": 0.57305300986568,
      "learning_rate": 2.4270734004424643e-07,
      "loss": 1.2791,
      "step": 1708
    },
    {
      "epoch": 2.7344,
      "grad_norm": 0.5685452609137315,
      "learning_rate": 2.39849823440374e-07,
      "loss": 1.2381,
      "step": 1709
    },
    {
      "epoch": 2.7359999999999998,
      "grad_norm": 0.5695026814372859,
      "learning_rate": 2.3700881467111025e-07,
      "loss": 1.2634,
      "step": 1710
    },
    {
      "epoch": 2.7376,
      "grad_norm": 0.5719150936876053,
      "learning_rate": 2.3418432358885633e-07,
      "loss": 1.2536,
      "step": 1711
    },
    {
      "epoch": 2.7392,
      "grad_norm": 0.5619500117524342,
      "learning_rate": 2.3137635998872808e-07,
      "loss": 1.3051,
      "step": 1712
    },
    {
      "epoch": 2.7408,
      "grad_norm": 0.553560502762873,
      "learning_rate": 2.285849336085294e-07,
      "loss": 1.297,
      "step": 1713
    },
    {
      "epoch": 2.7424,
      "grad_norm": 0.5622257024568822,
      "learning_rate": 2.258100541287117e-07,
      "loss": 1.273,
      "step": 1714
    },
    {
      "epoch": 2.7439999999999998,
      "grad_norm": 0.5711241682628831,
      "learning_rate": 2.2305173117234236e-07,
      "loss": 1.2702,
      "step": 1715
    },
    {
      "epoch": 2.7456,
      "grad_norm": 0.5528461753219661,
      "learning_rate": 2.2030997430507462e-07,
      "loss": 1.2547,
      "step": 1716
    },
    {
      "epoch": 2.7472,
      "grad_norm": 0.5494950706266473,
      "learning_rate": 2.1758479303510937e-07,
      "loss": 1.263,
      "step": 1717
    },
    {
      "epoch": 2.7488,
      "grad_norm": 0.5481935662426995,
      "learning_rate": 2.148761968131663e-07,
      "loss": 1.3522,
      "step": 1718
    },
    {
      "epoch": 2.7504,
      "grad_norm": 0.5533102522944502,
      "learning_rate": 2.121841950324488e-07,
      "loss": 1.2095,
      "step": 1719
    },
    {
      "epoch": 2.752,
      "grad_norm": 0.5506125046200556,
      "learning_rate": 2.0950879702861082e-07,
      "loss": 1.2405,
      "step": 1720
    },
    {
      "epoch": 2.7536,
      "grad_norm": 0.5630761116413253,
      "learning_rate": 2.0685001207972843e-07,
      "loss": 1.2675,
      "step": 1721
    },
    {
      "epoch": 2.7552,
      "grad_norm": 6.485228256380767,
      "learning_rate": 2.042078494062616e-07,
      "loss": 1.207,
      "step": 1722
    },
    {
      "epoch": 2.7568,
      "grad_norm": 0.5577211045098692,
      "learning_rate": 2.0158231817102858e-07,
      "loss": 1.2844,
      "step": 1723
    },
    {
      "epoch": 2.7584,
      "grad_norm": 0.5698573114141362,
      "learning_rate": 1.9897342747916938e-07,
      "loss": 1.3135,
      "step": 1724
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.5636628906819656,
      "learning_rate": 1.9638118637811564e-07,
      "loss": 1.1863,
      "step": 1725
    },
    {
      "epoch": 2.7616,
      "grad_norm": 0.558490117654026,
      "learning_rate": 1.9380560385756088e-07,
      "loss": 1.2313,
      "step": 1726
    },
    {
      "epoch": 2.7632,
      "grad_norm": 0.5469805277345128,
      "learning_rate": 1.9124668884942632e-07,
      "loss": 1.271,
      "step": 1727
    },
    {
      "epoch": 2.7648,
      "grad_norm": 0.5543830521407147,
      "learning_rate": 1.8870445022783234e-07,
      "loss": 1.2858,
      "step": 1728
    },
    {
      "epoch": 2.7664,
      "grad_norm": 0.5753203750245102,
      "learning_rate": 1.861788968090683e-07,
      "loss": 1.283,
      "step": 1729
    },
    {
      "epoch": 2.768,
      "grad_norm": 0.5331845200127951,
      "learning_rate": 1.8367003735155764e-07,
      "loss": 1.2333,
      "step": 1730
    },
    {
      "epoch": 2.7696,
      "grad_norm": 0.5479470701760915,
      "learning_rate": 1.8117788055583286e-07,
      "loss": 1.2281,
      "step": 1731
    },
    {
      "epoch": 2.7712,
      "grad_norm": 0.5490307844997762,
      "learning_rate": 1.7870243506450113e-07,
      "loss": 1.2806,
      "step": 1732
    },
    {
      "epoch": 2.7728,
      "grad_norm": 0.5554602292063848,
      "learning_rate": 1.762437094622177e-07,
      "loss": 1.2518,
      "step": 1733
    },
    {
      "epoch": 2.7744,
      "grad_norm": 0.5629745025455414,
      "learning_rate": 1.738017122756541e-07,
      "loss": 1.2724,
      "step": 1734
    },
    {
      "epoch": 2.776,
      "grad_norm": 0.5487622371683619,
      "learning_rate": 1.713764519734673e-07,
      "loss": 1.2623,
      "step": 1735
    },
    {
      "epoch": 2.7776,
      "grad_norm": 0.5546181497981874,
      "learning_rate": 1.68967936966275e-07,
      "loss": 1.2512,
      "step": 1736
    },
    {
      "epoch": 2.7792,
      "grad_norm": 0.572219319753532,
      "learning_rate": 1.6657617560662088e-07,
      "loss": 1.286,
      "step": 1737
    },
    {
      "epoch": 2.7808,
      "grad_norm": 0.5625356009975033,
      "learning_rate": 1.6420117618895003e-07,
      "loss": 1.2981,
      "step": 1738
    },
    {
      "epoch": 2.7824,
      "grad_norm": 0.5521299101268624,
      "learning_rate": 1.6184294694957747e-07,
      "loss": 1.2014,
      "step": 1739
    },
    {
      "epoch": 2.784,
      "grad_norm": 0.5638914642529425,
      "learning_rate": 1.5950149606666077e-07,
      "loss": 1.2985,
      "step": 1740
    },
    {
      "epoch": 2.7856,
      "grad_norm": 0.5621321293848132,
      "learning_rate": 1.5717683166017184e-07,
      "loss": 1.3247,
      "step": 1741
    },
    {
      "epoch": 2.7872,
      "grad_norm": 0.5526528785660573,
      "learning_rate": 1.5486896179186693e-07,
      "loss": 1.251,
      "step": 1742
    },
    {
      "epoch": 2.7888,
      "grad_norm": 0.563273043521601,
      "learning_rate": 1.5257789446526172e-07,
      "loss": 1.214,
      "step": 1743
    },
    {
      "epoch": 2.7904,
      "grad_norm": 0.5524851072332969,
      "learning_rate": 1.5030363762560228e-07,
      "loss": 1.2495,
      "step": 1744
    },
    {
      "epoch": 2.792,
      "grad_norm": 0.5806087574581205,
      "learning_rate": 1.480461991598353e-07,
      "loss": 1.2649,
      "step": 1745
    },
    {
      "epoch": 2.7936,
      "grad_norm": 0.5669463826052761,
      "learning_rate": 1.458055868965841e-07,
      "loss": 1.2388,
      "step": 1746
    },
    {
      "epoch": 2.7952,
      "grad_norm": 0.5604999174093179,
      "learning_rate": 1.4358180860611913e-07,
      "loss": 1.2591,
      "step": 1747
    },
    {
      "epoch": 2.7968,
      "grad_norm": 0.583478000443546,
      "learning_rate": 1.4137487200033383e-07,
      "loss": 1.2052,
      "step": 1748
    },
    {
      "epoch": 2.7984,
      "grad_norm": 0.5637211568595342,
      "learning_rate": 1.3918478473271325e-07,
      "loss": 1.3364,
      "step": 1749
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.5756786767008422,
      "learning_rate": 1.3701155439831249e-07,
      "loss": 1.2723,
      "step": 1750
    },
    {
      "epoch": 2.8016,
      "grad_norm": 0.5733696136773114,
      "learning_rate": 1.3485518853372625e-07,
      "loss": 1.2945,
      "step": 1751
    },
    {
      "epoch": 2.8032,
      "grad_norm": 0.5567044228808886,
      "learning_rate": 1.3271569461706547e-07,
      "loss": 1.2539,
      "step": 1752
    },
    {
      "epoch": 2.8048,
      "grad_norm": 0.5823460897259464,
      "learning_rate": 1.305930800679317e-07,
      "loss": 1.2696,
      "step": 1753
    },
    {
      "epoch": 2.8064,
      "grad_norm": 0.5633148119120308,
      "learning_rate": 1.2848735224738729e-07,
      "loss": 1.252,
      "step": 1754
    },
    {
      "epoch": 2.808,
      "grad_norm": 0.5384790222475053,
      "learning_rate": 1.2639851845793583e-07,
      "loss": 1.2783,
      "step": 1755
    },
    {
      "epoch": 2.8096,
      "grad_norm": 0.5471358901681584,
      "learning_rate": 1.2432658594349113e-07,
      "loss": 1.2781,
      "step": 1756
    },
    {
      "epoch": 2.8112,
      "grad_norm": 0.5512803760524161,
      "learning_rate": 1.2227156188935552e-07,
      "loss": 1.2617,
      "step": 1757
    },
    {
      "epoch": 2.8128,
      "grad_norm": 0.5480493171167334,
      "learning_rate": 1.202334534221955e-07,
      "loss": 1.2445,
      "step": 1758
    },
    {
      "epoch": 2.8144,
      "grad_norm": 0.5566149740836109,
      "learning_rate": 1.1821226761001391e-07,
      "loss": 1.3023,
      "step": 1759
    },
    {
      "epoch": 2.816,
      "grad_norm": 0.5577083756122834,
      "learning_rate": 1.1620801146212723e-07,
      "loss": 1.2534,
      "step": 1760
    },
    {
      "epoch": 2.8176,
      "grad_norm": 0.543445149417826,
      "learning_rate": 1.1422069192914221e-07,
      "loss": 1.3149,
      "step": 1761
    },
    {
      "epoch": 2.8192,
      "grad_norm": 0.5533514187160694,
      "learning_rate": 1.1225031590292923e-07,
      "loss": 1.2499,
      "step": 1762
    },
    {
      "epoch": 2.8208,
      "grad_norm": 0.558031937102035,
      "learning_rate": 1.1029689021660183e-07,
      "loss": 1.3679,
      "step": 1763
    },
    {
      "epoch": 2.8224,
      "grad_norm": 0.5566192564618486,
      "learning_rate": 1.0836042164448945e-07,
      "loss": 1.2137,
      "step": 1764
    },
    {
      "epoch": 2.824,
      "grad_norm": 0.5829423747609419,
      "learning_rate": 1.0644091690211633e-07,
      "loss": 1.2436,
      "step": 1765
    },
    {
      "epoch": 2.8256,
      "grad_norm": 0.5400393700872067,
      "learning_rate": 1.0453838264617711e-07,
      "loss": 1.3105,
      "step": 1766
    },
    {
      "epoch": 2.8272,
      "grad_norm": 0.568906930526538,
      "learning_rate": 1.0265282547451405e-07,
      "loss": 1.278,
      "step": 1767
    },
    {
      "epoch": 2.8288,
      "grad_norm": 0.5723840992556024,
      "learning_rate": 1.0078425192609487e-07,
      "loss": 1.3283,
      "step": 1768
    },
    {
      "epoch": 2.8304,
      "grad_norm": 0.5609867332729649,
      "learning_rate": 9.893266848098826e-08,
      "loss": 1.2923,
      "step": 1769
    },
    {
      "epoch": 2.832,
      "grad_norm": 0.5575687227061882,
      "learning_rate": 9.709808156034394e-08,
      "loss": 1.216,
      "step": 1770
    },
    {
      "epoch": 2.8336,
      "grad_norm": 0.546186497963765,
      "learning_rate": 9.528049752636714e-08,
      "loss": 1.236,
      "step": 1771
    },
    {
      "epoch": 2.8352,
      "grad_norm": 0.566280877534487,
      "learning_rate": 9.347992268230022e-08,
      "loss": 1.3236,
      "step": 1772
    },
    {
      "epoch": 2.8368,
      "grad_norm": 0.5429987763939276,
      "learning_rate": 9.169636327239883e-08,
      "loss": 1.2626,
      "step": 1773
    },
    {
      "epoch": 2.8384,
      "grad_norm": 0.5551936278078753,
      "learning_rate": 8.992982548190809e-08,
      "loss": 1.2783,
      "step": 1774
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.5581005368904405,
      "learning_rate": 8.818031543704641e-08,
      "loss": 1.1827,
      "step": 1775
    },
    {
      "epoch": 2.8416,
      "grad_norm": 0.5562215361967292,
      "learning_rate": 8.644783920498001e-08,
      "loss": 1.2595,
      "step": 1776
    },
    {
      "epoch": 2.8432,
      "grad_norm": 0.5611765362568665,
      "learning_rate": 8.473240279380235e-08,
      "loss": 1.2469,
      "step": 1777
    },
    {
      "epoch": 2.8448,
      "grad_norm": 0.5668827747547003,
      "learning_rate": 8.303401215251583e-08,
      "loss": 1.211,
      "step": 1778
    },
    {
      "epoch": 2.8464,
      "grad_norm": 0.566017628055969,
      "learning_rate": 8.135267317100792e-08,
      "loss": 1.3043,
      "step": 1779
    },
    {
      "epoch": 2.848,
      "grad_norm": 0.5549775012646712,
      "learning_rate": 7.968839168003395e-08,
      "loss": 1.2545,
      "step": 1780
    },
    {
      "epoch": 2.8496,
      "grad_norm": 6.3820726568433415,
      "learning_rate": 7.804117345119266e-08,
      "loss": 1.294,
      "step": 1781
    },
    {
      "epoch": 2.8512,
      "grad_norm": 0.5402478945972932,
      "learning_rate": 7.64110241969107e-08,
      "loss": 1.2625,
      "step": 1782
    },
    {
      "epoch": 2.8528000000000002,
      "grad_norm": 0.5541939930159807,
      "learning_rate": 7.479794957042041e-08,
      "loss": 1.2345,
      "step": 1783
    },
    {
      "epoch": 2.8544,
      "grad_norm": 0.5492766795074051,
      "learning_rate": 7.320195516574036e-08,
      "loss": 1.2742,
      "step": 1784
    },
    {
      "epoch": 2.856,
      "grad_norm": 0.5426417187390493,
      "learning_rate": 7.16230465176565e-08,
      "loss": 1.2217,
      "step": 1785
    },
    {
      "epoch": 2.8576,
      "grad_norm": 0.5539553154403377,
      "learning_rate": 7.00612291017022e-08,
      "loss": 1.285,
      "step": 1786
    },
    {
      "epoch": 2.8592,
      "grad_norm": 0.5472860921974896,
      "learning_rate": 6.851650833414103e-08,
      "loss": 1.2952,
      "step": 1787
    },
    {
      "epoch": 2.8608000000000002,
      "grad_norm": 0.5806367778429027,
      "learning_rate": 6.698888957194505e-08,
      "loss": 1.2726,
      "step": 1788
    },
    {
      "epoch": 2.8624,
      "grad_norm": 0.5587488333197712,
      "learning_rate": 6.547837811277824e-08,
      "loss": 1.2396,
      "step": 1789
    },
    {
      "epoch": 2.864,
      "grad_norm": 0.5541428274066975,
      "learning_rate": 6.39849791949787e-08,
      "loss": 1.2446,
      "step": 1790
    },
    {
      "epoch": 2.8656,
      "grad_norm": 0.554033054584323,
      "learning_rate": 6.250869799753866e-08,
      "loss": 1.2301,
      "step": 1791
    },
    {
      "epoch": 2.8672,
      "grad_norm": 0.552770731748147,
      "learning_rate": 6.104953964008897e-08,
      "loss": 1.3161,
      "step": 1792
    },
    {
      "epoch": 2.8688000000000002,
      "grad_norm": 0.5244521147888241,
      "learning_rate": 5.960750918287627e-08,
      "loss": 1.2271,
      "step": 1793
    },
    {
      "epoch": 2.8704,
      "grad_norm": 0.5697836550573574,
      "learning_rate": 5.818261162675309e-08,
      "loss": 1.3051,
      "step": 1794
    },
    {
      "epoch": 2.872,
      "grad_norm": 0.558172935724767,
      "learning_rate": 5.677485191315391e-08,
      "loss": 1.2103,
      "step": 1795
    },
    {
      "epoch": 2.8736,
      "grad_norm": 0.5660792080866764,
      "learning_rate": 5.538423492408129e-08,
      "loss": 1.2274,
      "step": 1796
    },
    {
      "epoch": 2.8752,
      "grad_norm": 0.5487817166239571,
      "learning_rate": 5.401076548208761e-08,
      "loss": 1.2788,
      "step": 1797
    },
    {
      "epoch": 2.8768000000000002,
      "grad_norm": 0.5856207295839101,
      "learning_rate": 5.265444835025946e-08,
      "loss": 1.3512,
      "step": 1798
    },
    {
      "epoch": 2.8784,
      "grad_norm": 0.5644880564862201,
      "learning_rate": 5.1315288232201e-08,
      "loss": 1.3283,
      "step": 1799
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.5538014345024743,
      "learning_rate": 4.9993289772015116e-08,
      "loss": 1.2938,
      "step": 1800
    },
    {
      "epoch": 2.8816,
      "grad_norm": 0.5493350424923675,
      "learning_rate": 4.8688457554291746e-08,
      "loss": 1.2232,
      "step": 1801
    },
    {
      "epoch": 2.8832,
      "grad_norm": 0.5895281084017707,
      "learning_rate": 4.7400796104088434e-08,
      "loss": 1.3033,
      "step": 1802
    },
    {
      "epoch": 2.8848000000000003,
      "grad_norm": 0.5525111520355052,
      "learning_rate": 4.613030988691536e-08,
      "loss": 1.3229,
      "step": 1803
    },
    {
      "epoch": 2.8864,
      "grad_norm": 0.5493992722969065,
      "learning_rate": 4.4877003308722575e-08,
      "loss": 1.187,
      "step": 1804
    },
    {
      "epoch": 2.888,
      "grad_norm": 0.5467253046710658,
      "learning_rate": 4.364088071587891e-08,
      "loss": 1.2109,
      "step": 1805
    },
    {
      "epoch": 2.8895999999999997,
      "grad_norm": 0.5494904231939686,
      "learning_rate": 4.2421946395164174e-08,
      "loss": 1.2773,
      "step": 1806
    },
    {
      "epoch": 2.8912,
      "grad_norm": 0.565528576490617,
      "learning_rate": 4.1220204573747534e-08,
      "loss": 1.2904,
      "step": 1807
    },
    {
      "epoch": 2.8928000000000003,
      "grad_norm": 0.5432455640597353,
      "learning_rate": 4.0035659419178086e-08,
      "loss": 1.2691,
      "step": 1808
    },
    {
      "epoch": 2.8944,
      "grad_norm": 0.5622561015689554,
      "learning_rate": 3.88683150393665e-08,
      "loss": 1.2497,
      "step": 1809
    },
    {
      "epoch": 2.896,
      "grad_norm": 0.5311586033052118,
      "learning_rate": 3.771817548257395e-08,
      "loss": 1.2482,
      "step": 1810
    },
    {
      "epoch": 2.8975999999999997,
      "grad_norm": 0.5413519400977449,
      "learning_rate": 3.658524473739544e-08,
      "loss": 1.2393,
      "step": 1811
    },
    {
      "epoch": 2.8992,
      "grad_norm": 0.5669633803332041,
      "learning_rate": 3.546952673274817e-08,
      "loss": 1.1989,
      "step": 1812
    },
    {
      "epoch": 2.9008000000000003,
      "grad_norm": 0.6085035378763393,
      "learning_rate": 3.437102533785541e-08,
      "loss": 1.2349,
      "step": 1813
    },
    {
      "epoch": 2.9024,
      "grad_norm": 0.5649246662413893,
      "learning_rate": 3.328974436223709e-08,
      "loss": 1.3193,
      "step": 1814
    },
    {
      "epoch": 2.904,
      "grad_norm": 0.5543204713231675,
      "learning_rate": 3.2225687555690886e-08,
      "loss": 1.2382,
      "step": 1815
    },
    {
      "epoch": 2.9055999999999997,
      "grad_norm": 0.5631023693118996,
      "learning_rate": 3.117885860828396e-08,
      "loss": 1.2818,
      "step": 1816
    },
    {
      "epoch": 2.9072,
      "grad_norm": 0.5353915898825748,
      "learning_rate": 3.014926115034012e-08,
      "loss": 1.2322,
      "step": 1817
    },
    {
      "epoch": 2.9088000000000003,
      "grad_norm": 0.5866828575031897,
      "learning_rate": 2.9136898752422648e-08,
      "loss": 1.2409,
      "step": 1818
    },
    {
      "epoch": 2.9104,
      "grad_norm": 0.5554945687263335,
      "learning_rate": 2.8141774925327103e-08,
      "loss": 1.2633,
      "step": 1819
    },
    {
      "epoch": 2.912,
      "grad_norm": 0.5552972497442331,
      "learning_rate": 2.7163893120066288e-08,
      "loss": 1.2618,
      "step": 1820
    },
    {
      "epoch": 2.9135999999999997,
      "grad_norm": 0.5666243832567139,
      "learning_rate": 2.6203256727859172e-08,
      "loss": 1.2658,
      "step": 1821
    },
    {
      "epoch": 2.9152,
      "grad_norm": 0.5797165679243611,
      "learning_rate": 2.5259869080118127e-08,
      "loss": 1.2788,
      "step": 1822
    },
    {
      "epoch": 2.9168,
      "grad_norm": 0.5434415774831625,
      "learning_rate": 2.4333733448440033e-08,
      "loss": 1.18,
      "step": 1823
    },
    {
      "epoch": 2.9184,
      "grad_norm": 0.5556051392415596,
      "learning_rate": 2.34248530445913e-08,
      "loss": 1.2235,
      "step": 1824
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.5571591022622593,
      "learning_rate": 2.2533231020499536e-08,
      "loss": 1.2391,
      "step": 1825
    },
    {
      "epoch": 2.9215999999999998,
      "grad_norm": 0.5438941776887378,
      "learning_rate": 2.1658870468241332e-08,
      "loss": 1.2518,
      "step": 1826
    },
    {
      "epoch": 2.9232,
      "grad_norm": 0.561781609906814,
      "learning_rate": 2.0801774420031172e-08,
      "loss": 1.2658,
      "step": 1827
    },
    {
      "epoch": 2.9248,
      "grad_norm": 0.5807589904595962,
      "learning_rate": 1.9961945848213092e-08,
      "loss": 1.2941,
      "step": 1828
    },
    {
      "epoch": 2.9264,
      "grad_norm": 0.5442600920696243,
      "learning_rate": 1.9139387665247922e-08,
      "loss": 1.3221,
      "step": 1829
    },
    {
      "epoch": 2.928,
      "grad_norm": 0.5550128395237249,
      "learning_rate": 1.8334102723703286e-08,
      "loss": 1.3474,
      "step": 1830
    },
    {
      "epoch": 2.9295999999999998,
      "grad_norm": 0.5426490006842143,
      "learning_rate": 1.754609381624639e-08,
      "loss": 1.2343,
      "step": 1831
    },
    {
      "epoch": 2.9312,
      "grad_norm": 0.5465873783624812,
      "learning_rate": 1.677536367563126e-08,
      "loss": 1.3296,
      "step": 1832
    },
    {
      "epoch": 2.9328,
      "grad_norm": 0.5470090459739341,
      "learning_rate": 1.6021914974690413e-08,
      "loss": 1.2411,
      "step": 1833
    },
    {
      "epoch": 2.9344,
      "grad_norm": 0.5585156695850074,
      "learning_rate": 1.5285750326325953e-08,
      "loss": 1.2675,
      "step": 1834
    },
    {
      "epoch": 2.936,
      "grad_norm": 0.5737608035647862,
      "learning_rate": 1.4566872283500733e-08,
      "loss": 1.2718,
      "step": 1835
    },
    {
      "epoch": 2.9375999999999998,
      "grad_norm": 0.5493876573202087,
      "learning_rate": 1.3865283339228319e-08,
      "loss": 1.2566,
      "step": 1836
    },
    {
      "epoch": 2.9392,
      "grad_norm": 0.5511998161371885,
      "learning_rate": 1.3180985926564693e-08,
      "loss": 1.279,
      "step": 1837
    },
    {
      "epoch": 2.9408,
      "grad_norm": 0.5450961729420604,
      "learning_rate": 1.2513982418601024e-08,
      "loss": 1.2381,
      "step": 1838
    },
    {
      "epoch": 2.9424,
      "grad_norm": 0.5664756371845556,
      "learning_rate": 1.1864275128454783e-08,
      "loss": 1.2865,
      "step": 1839
    },
    {
      "epoch": 2.944,
      "grad_norm": 0.5340468698796819,
      "learning_rate": 1.1231866309259764e-08,
      "loss": 1.273,
      "step": 1840
    },
    {
      "epoch": 2.9455999999999998,
      "grad_norm": 0.542983327824006,
      "learning_rate": 1.0616758154161633e-08,
      "loss": 1.2374,
      "step": 1841
    },
    {
      "epoch": 2.9472,
      "grad_norm": 0.5658780480386415,
      "learning_rate": 1.0018952796307934e-08,
      "loss": 1.2837,
      "step": 1842
    },
    {
      "epoch": 2.9488,
      "grad_norm": 0.5394891439174904,
      "learning_rate": 9.438452308841995e-09,
      "loss": 1.2998,
      "step": 1843
    },
    {
      "epoch": 2.9504,
      "grad_norm": 0.5294189650317123,
      "learning_rate": 8.87525870489514e-09,
      "loss": 1.2184,
      "step": 1844
    },
    {
      "epoch": 2.952,
      "grad_norm": 0.5445959303677909,
      "learning_rate": 8.329373937578378e-09,
      "loss": 1.2931,
      "step": 1845
    },
    {
      "epoch": 2.9536,
      "grad_norm": 0.551925803386912,
      "learning_rate": 7.800799899979061e-09,
      "loss": 1.3072,
      "step": 1846
    },
    {
      "epoch": 2.9552,
      "grad_norm": 0.5530900737103289,
      "learning_rate": 7.289538425150899e-09,
      "loss": 1.3022,
      "step": 1847
    },
    {
      "epoch": 2.9568,
      "grad_norm": 0.5720257078780323,
      "learning_rate": 6.7955912861095155e-09,
      "loss": 1.2395,
      "step": 1848
    },
    {
      "epoch": 2.9584,
      "grad_norm": 0.5425361495515822,
      "learning_rate": 6.31896019582523e-09,
      "loss": 1.273,
      "step": 1849
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.5659720898281938,
      "learning_rate": 5.8596468072180665e-09,
      "loss": 1.1759,
      "step": 1850
    },
    {
      "epoch": 2.9616,
      "grad_norm": 0.5702973270268117,
      "learning_rate": 5.417652713152199e-09,
      "loss": 1.2409,
      "step": 1851
    },
    {
      "epoch": 2.9632,
      "grad_norm": 0.5723472616633926,
      "learning_rate": 4.992979446428736e-09,
      "loss": 1.2516,
      "step": 1852
    },
    {
      "epoch": 2.9648,
      "grad_norm": 0.5559222517976384,
      "learning_rate": 4.585628479781279e-09,
      "loss": 1.2548,
      "step": 1853
    },
    {
      "epoch": 2.9664,
      "grad_norm": 0.5465710026119162,
      "learning_rate": 4.195601225872592e-09,
      "loss": 1.2431,
      "step": 1854
    },
    {
      "epoch": 2.968,
      "grad_norm": 0.5582789315842982,
      "learning_rate": 3.822899037286276e-09,
      "loss": 1.2969,
      "step": 1855
    },
    {
      "epoch": 2.9696,
      "grad_norm": 0.5877639207401307,
      "learning_rate": 3.4675232065256583e-09,
      "loss": 1.2032,
      "step": 1856
    },
    {
      "epoch": 2.9712,
      "grad_norm": 0.5358401331931416,
      "learning_rate": 3.129474966006574e-09,
      "loss": 1.2387,
      "step": 1857
    },
    {
      "epoch": 2.9728,
      "grad_norm": 0.5500186624165608,
      "learning_rate": 2.808755488054038e-09,
      "loss": 1.3358,
      "step": 1858
    },
    {
      "epoch": 2.9744,
      "grad_norm": 0.5771807221414395,
      "learning_rate": 2.5053658848989137e-09,
      "loss": 1.3086,
      "step": 1859
    },
    {
      "epoch": 2.976,
      "grad_norm": 0.5764632864157987,
      "learning_rate": 2.219307208672361e-09,
      "loss": 1.2558,
      "step": 1860
    },
    {
      "epoch": 2.9776,
      "grad_norm": 0.5403778946712635,
      "learning_rate": 1.9505804514047266e-09,
      "loss": 1.198,
      "step": 1861
    },
    {
      "epoch": 2.9792,
      "grad_norm": 0.5642478953827482,
      "learning_rate": 1.6991865450188827e-09,
      "loss": 1.2546,
      "step": 1862
    },
    {
      "epoch": 2.9808,
      "grad_norm": 0.5449461206604768,
      "learning_rate": 1.465126361330227e-09,
      "loss": 1.2631,
      "step": 1863
    },
    {
      "epoch": 2.9824,
      "grad_norm": 0.5491952631718166,
      "learning_rate": 1.2484007120411312e-09,
      "loss": 1.2368,
      "step": 1864
    },
    {
      "epoch": 2.984,
      "grad_norm": 0.5515085333582388,
      "learning_rate": 1.0490103487392766e-09,
      "loss": 1.2826,
      "step": 1865
    },
    {
      "epoch": 2.9856,
      "grad_norm": 0.575214675070689,
      "learning_rate": 8.669559628954327e-10,
      "loss": 1.2634,
      "step": 1866
    },
    {
      "epoch": 2.9872,
      "grad_norm": 0.5996318497136518,
      "learning_rate": 7.02238185860682e-10,
      "loss": 1.3026,
      "step": 1867
    },
    {
      "epoch": 2.9888,
      "grad_norm": 0.5522995252182995,
      "learning_rate": 5.54857588862534e-10,
      "loss": 1.3101,
      "step": 1868
    },
    {
      "epoch": 2.9904,
      "grad_norm": 0.5512786370931705,
      "learning_rate": 4.2481468300603625e-10,
      "loss": 1.2483,
      "step": 1869
    },
    {
      "epoch": 2.992,
      "grad_norm": 0.5636789030695993,
      "learning_rate": 3.1210991927044244e-10,
      "loss": 1.2646,
      "step": 1870
    },
    {
      "epoch": 2.9936,
      "grad_norm": 0.5504579363064919,
      "learning_rate": 2.167436885064378e-10,
      "loss": 1.261,
      "step": 1871
    },
    {
      "epoch": 2.9952,
      "grad_norm": 0.5368713095253141,
      "learning_rate": 1.387163214372489e-10,
      "loss": 1.3009,
      "step": 1872
    },
    {
      "epoch": 2.9968,
      "grad_norm": 0.5492833250352148,
      "learning_rate": 7.80280886558682e-11,
      "loss": 1.3102,
      "step": 1873
    },
    {
      "epoch": 2.9984,
      "grad_norm": 0.5850602583136507,
      "learning_rate": 3.467920062394381e-11,
      "loss": 1.2787,
      "step": 1874
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.551926907229094,
      "learning_rate": 8.669807672334606e-12,
      "loss": 1.293,
      "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
}