{
  "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": 0.7542400501913732,
      "learning_rate": 3.5087719298245616e-07,
      "loss": 1.6751,
      "step": 1
    },
    {
      "epoch": 0.0032,
      "grad_norm": 0.742305937368562,
      "learning_rate": 7.017543859649123e-07,
      "loss": 1.6555,
      "step": 2
    },
    {
      "epoch": 0.0048,
      "grad_norm": 0.6980748485515275,
      "learning_rate": 1.0526315789473685e-06,
      "loss": 1.6802,
      "step": 3
    },
    {
      "epoch": 0.0064,
      "grad_norm": 0.8118757008619297,
      "learning_rate": 1.4035087719298246e-06,
      "loss": 1.7025,
      "step": 4
    },
    {
      "epoch": 0.008,
      "grad_norm": 0.7050523685798155,
      "learning_rate": 1.7543859649122807e-06,
      "loss": 1.6962,
      "step": 5
    },
    {
      "epoch": 0.0096,
      "grad_norm": 0.711727455005183,
      "learning_rate": 2.105263157894737e-06,
      "loss": 1.697,
      "step": 6
    },
    {
      "epoch": 0.0112,
      "grad_norm": 0.7160671118334767,
      "learning_rate": 2.456140350877193e-06,
      "loss": 1.6868,
      "step": 7
    },
    {
      "epoch": 0.0128,
      "grad_norm": 0.7821002979119394,
      "learning_rate": 2.8070175438596493e-06,
      "loss": 1.6639,
      "step": 8
    },
    {
      "epoch": 0.0144,
      "grad_norm": 0.7451098951855825,
      "learning_rate": 3.157894736842105e-06,
      "loss": 1.6825,
      "step": 9
    },
    {
      "epoch": 0.016,
      "grad_norm": 0.7363068641172336,
      "learning_rate": 3.5087719298245615e-06,
      "loss": 1.6668,
      "step": 10
    },
    {
      "epoch": 0.0176,
      "grad_norm": 0.7116123257106387,
      "learning_rate": 3.859649122807018e-06,
      "loss": 1.665,
      "step": 11
    },
    {
      "epoch": 0.0192,
      "grad_norm": 0.692542211487462,
      "learning_rate": 4.210526315789474e-06,
      "loss": 1.6172,
      "step": 12
    },
    {
      "epoch": 0.0208,
      "grad_norm": 0.7049109164414815,
      "learning_rate": 4.56140350877193e-06,
      "loss": 1.6232,
      "step": 13
    },
    {
      "epoch": 0.0224,
      "grad_norm": 0.8451468378901278,
      "learning_rate": 4.912280701754386e-06,
      "loss": 1.6017,
      "step": 14
    },
    {
      "epoch": 0.024,
      "grad_norm": 0.8451468378901278,
      "learning_rate": 4.912280701754386e-06,
      "loss": 1.6056,
      "step": 15
    },
    {
      "epoch": 0.0256,
      "grad_norm": 0.7169606297094776,
      "learning_rate": 5.263157894736842e-06,
      "loss": 1.6057,
      "step": 16
    },
    {
      "epoch": 0.0272,
      "grad_norm": 0.7135705482908589,
      "learning_rate": 5.6140350877192985e-06,
      "loss": 1.6443,
      "step": 17
    },
    {
      "epoch": 0.0288,
      "grad_norm": 0.7167917758950816,
      "learning_rate": 5.964912280701755e-06,
      "loss": 1.5457,
      "step": 18
    },
    {
      "epoch": 0.0304,
      "grad_norm": 0.7764053685064782,
      "learning_rate": 6.31578947368421e-06,
      "loss": 1.5726,
      "step": 19
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.9473759206115546,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.5574,
      "step": 20
    },
    {
      "epoch": 0.0336,
      "grad_norm": 0.7523094343855393,
      "learning_rate": 7.017543859649123e-06,
      "loss": 1.5088,
      "step": 21
    },
    {
      "epoch": 0.0352,
      "grad_norm": 0.7105882407110944,
      "learning_rate": 7.368421052631579e-06,
      "loss": 1.5274,
      "step": 22
    },
    {
      "epoch": 0.0368,
      "grad_norm": 0.6879803795744842,
      "learning_rate": 7.719298245614036e-06,
      "loss": 1.5775,
      "step": 23
    },
    {
      "epoch": 0.0384,
      "grad_norm": 0.6570506808253213,
      "learning_rate": 8.070175438596492e-06,
      "loss": 1.5383,
      "step": 24
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6790633921201122,
      "learning_rate": 8.421052631578948e-06,
      "loss": 1.4843,
      "step": 25
    },
    {
      "epoch": 0.0416,
      "grad_norm": 0.7507734448320031,
      "learning_rate": 8.771929824561405e-06,
      "loss": 1.4724,
      "step": 26
    },
    {
      "epoch": 0.0432,
      "grad_norm": 0.6260435863513739,
      "learning_rate": 9.12280701754386e-06,
      "loss": 1.5053,
      "step": 27
    },
    {
      "epoch": 0.0448,
      "grad_norm": 0.6344279227660959,
      "learning_rate": 9.473684210526315e-06,
      "loss": 1.4264,
      "step": 28
    },
    {
      "epoch": 0.0464,
      "grad_norm": 0.6373882465568406,
      "learning_rate": 9.824561403508772e-06,
      "loss": 1.4379,
      "step": 29
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.7332703557822935,
      "learning_rate": 1.017543859649123e-05,
      "loss": 1.3561,
      "step": 30
    },
    {
      "epoch": 0.0496,
      "grad_norm": 0.6037687269257381,
      "learning_rate": 1.0526315789473684e-05,
      "loss": 1.3818,
      "step": 31
    },
    {
      "epoch": 0.0512,
      "grad_norm": 0.6152316972590014,
      "learning_rate": 1.0877192982456142e-05,
      "loss": 1.3158,
      "step": 32
    },
    {
      "epoch": 0.0528,
      "grad_norm": 0.6574478270840337,
      "learning_rate": 1.1228070175438597e-05,
      "loss": 1.3933,
      "step": 33
    },
    {
      "epoch": 0.0544,
      "grad_norm": 0.5990701570969205,
      "learning_rate": 1.1578947368421053e-05,
      "loss": 1.3143,
      "step": 34
    },
    {
      "epoch": 0.056,
      "grad_norm": 0.6450886782633305,
      "learning_rate": 1.192982456140351e-05,
      "loss": 1.3073,
      "step": 35
    },
    {
      "epoch": 0.0576,
      "grad_norm": 0.5773418755615032,
      "learning_rate": 1.2280701754385966e-05,
      "loss": 1.2652,
      "step": 36
    },
    {
      "epoch": 0.0592,
      "grad_norm": 0.6757255156889946,
      "learning_rate": 1.263157894736842e-05,
      "loss": 1.2634,
      "step": 37
    },
    {
      "epoch": 0.0608,
      "grad_norm": 0.6363505050303085,
      "learning_rate": 1.2982456140350879e-05,
      "loss": 1.2551,
      "step": 38
    },
    {
      "epoch": 0.0624,
      "grad_norm": 0.5900310001990254,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 1.2004,
      "step": 39
    },
    {
      "epoch": 0.064,
      "grad_norm": 0.5947705278540787,
      "learning_rate": 1.3684210526315791e-05,
      "loss": 1.1485,
      "step": 40
    },
    {
      "epoch": 0.0656,
      "grad_norm": 0.6511869039207653,
      "learning_rate": 1.4035087719298246e-05,
      "loss": 1.1496,
      "step": 41
    },
    {
      "epoch": 0.0672,
      "grad_norm": 0.5968807693876552,
      "learning_rate": 1.4385964912280704e-05,
      "loss": 1.1163,
      "step": 42
    },
    {
      "epoch": 0.0688,
      "grad_norm": 0.6768227590842741,
      "learning_rate": 1.4736842105263159e-05,
      "loss": 1.1251,
      "step": 43
    },
    {
      "epoch": 0.0704,
      "grad_norm": 1.331031279479116,
      "learning_rate": 1.5087719298245615e-05,
      "loss": 1.072,
      "step": 44
    },
    {
      "epoch": 0.072,
      "grad_norm": 0.6420589576746323,
      "learning_rate": 1.543859649122807e-05,
      "loss": 1.0612,
      "step": 45
    },
    {
      "epoch": 0.0736,
      "grad_norm": 0.637279753966264,
      "learning_rate": 1.578947368421053e-05,
      "loss": 0.9885,
      "step": 46
    },
    {
      "epoch": 0.0752,
      "grad_norm": 0.6285715183385797,
      "learning_rate": 1.6140350877192984e-05,
      "loss": 0.9694,
      "step": 47
    },
    {
      "epoch": 0.0768,
      "grad_norm": 0.6861440187647094,
      "learning_rate": 1.649122807017544e-05,
      "loss": 0.9652,
      "step": 48
    },
    {
      "epoch": 0.0784,
      "grad_norm": 0.7538552449135,
      "learning_rate": 1.6842105263157896e-05,
      "loss": 0.9091,
      "step": 49
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7139497393950953,
      "learning_rate": 1.719298245614035e-05,
      "loss": 0.8914,
      "step": 50
    },
    {
      "epoch": 0.0816,
      "grad_norm": 0.6987917832417296,
      "learning_rate": 1.754385964912281e-05,
      "loss": 0.8326,
      "step": 51
    },
    {
      "epoch": 0.0832,
      "grad_norm": 0.6871197658295636,
      "learning_rate": 1.7894736842105264e-05,
      "loss": 0.7873,
      "step": 52
    },
    {
      "epoch": 0.0848,
      "grad_norm": 0.7292306074949482,
      "learning_rate": 1.824561403508772e-05,
      "loss": 0.7752,
      "step": 53
    },
    {
      "epoch": 0.0864,
      "grad_norm": 0.7750209058554545,
      "learning_rate": 1.8596491228070176e-05,
      "loss": 0.7384,
      "step": 54
    },
    {
      "epoch": 0.088,
      "grad_norm": 0.8563870585217198,
      "learning_rate": 1.894736842105263e-05,
      "loss": 0.6838,
      "step": 55
    },
    {
      "epoch": 0.0896,
      "grad_norm": 0.8700867511623959,
      "learning_rate": 1.929824561403509e-05,
      "loss": 0.6253,
      "step": 56
    },
    {
      "epoch": 0.0912,
      "grad_norm": 0.8715974869514921,
      "learning_rate": 1.9649122807017544e-05,
      "loss": 0.577,
      "step": 57
    },
    {
      "epoch": 0.0928,
      "grad_norm": 0.8853465018763553,
      "learning_rate": 2e-05,
      "loss": 0.5009,
      "step": 58
    },
    {
      "epoch": 0.0944,
      "grad_norm": 1.0984224579924668,
      "learning_rate": 1.9999985069241058e-05,
      "loss": 0.5193,
      "step": 59
    },
    {
      "epoch": 0.096,
      "grad_norm": 1.4561780667877906,
      "learning_rate": 1.9999940277008807e-05,
      "loss": 0.4672,
      "step": 60
    },
    {
      "epoch": 0.0976,
      "grad_norm": 1.4370299342188482,
      "learning_rate": 1.9999865623437014e-05,
      "loss": 0.404,
      "step": 61
    },
    {
      "epoch": 0.0992,
      "grad_norm": 1.0143477041895286,
      "learning_rate": 1.99997611087486e-05,
      "loss": 0.3898,
      "step": 62
    },
    {
      "epoch": 0.1008,
      "grad_norm": 0.8105824534934927,
      "learning_rate": 1.9999626733255662e-05,
      "loss": 0.3629,
      "step": 63
    },
    {
      "epoch": 0.1024,
      "grad_norm": 0.7390096173444612,
      "learning_rate": 1.9999462497359468e-05,
      "loss": 0.3122,
      "step": 64
    },
    {
      "epoch": 0.104,
      "grad_norm": 0.7115518497609002,
      "learning_rate": 1.9999268401550445e-05,
      "loss": 0.3047,
      "step": 65
    },
    {
      "epoch": 0.1056,
      "grad_norm": 0.7207830753076017,
      "learning_rate": 1.9999044446408203e-05,
      "loss": 0.2692,
      "step": 66
    },
    {
      "epoch": 0.1072,
      "grad_norm": 0.7856619231156039,
      "learning_rate": 1.9998790632601496e-05,
      "loss": 0.2416,
      "step": 67
    },
    {
      "epoch": 0.1088,
      "grad_norm": 0.6475630242734959,
      "learning_rate": 1.9998506960888258e-05,
      "loss": 0.2092,
      "step": 68
    },
    {
      "epoch": 0.1104,
      "grad_norm": 0.5711599547520134,
      "learning_rate": 1.999819343211557e-05,
      "loss": 0.2103,
      "step": 69
    },
    {
      "epoch": 0.112,
      "grad_norm": 0.7075381593815743,
      "learning_rate": 1.999785004721968e-05,
      "loss": 0.2024,
      "step": 70
    },
    {
      "epoch": 0.1136,
      "grad_norm": 0.7278916815227956,
      "learning_rate": 1.9997476807225987e-05,
      "loss": 0.1923,
      "step": 71
    },
    {
      "epoch": 0.1152,
      "grad_norm": 0.7400146380176278,
      "learning_rate": 1.999707371324904e-05,
      "loss": 0.1719,
      "step": 72
    },
    {
      "epoch": 0.1168,
      "grad_norm": 0.731122455083543,
      "learning_rate": 1.9996640766492542e-05,
      "loss": 0.1562,
      "step": 73
    },
    {
      "epoch": 0.1184,
      "grad_norm": 0.6656914514197745,
      "learning_rate": 1.9996177968249336e-05,
      "loss": 0.1392,
      "step": 74
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6552350171809493,
      "learning_rate": 1.999568531990141e-05,
      "loss": 0.147,
      "step": 75
    },
    {
      "epoch": 0.1216,
      "grad_norm": 0.5189005591063778,
      "learning_rate": 1.999516282291988e-05,
      "loss": 0.1282,
      "step": 76
    },
    {
      "epoch": 0.1232,
      "grad_norm": 0.4733125299687125,
      "learning_rate": 1.9994610478865012e-05,
      "loss": 0.1088,
      "step": 77
    },
    {
      "epoch": 0.1248,
      "grad_norm": 0.5579587554542267,
      "learning_rate": 1.999402828938618e-05,
      "loss": 0.1233,
      "step": 78
    },
    {
      "epoch": 0.1264,
      "grad_norm": 0.5092352653843509,
      "learning_rate": 1.9993416256221894e-05,
      "loss": 0.0934,
      "step": 79
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.519808016741284,
      "learning_rate": 1.999277438119978e-05,
      "loss": 0.0914,
      "step": 80
    },
    {
      "epoch": 0.1296,
      "grad_norm": 0.6408642013945355,
      "learning_rate": 1.9992102666236567e-05,
      "loss": 0.1004,
      "step": 81
    },
    {
      "epoch": 0.1312,
      "grad_norm": 0.6438889392476188,
      "learning_rate": 1.9991401113338103e-05,
      "loss": 0.0912,
      "step": 82
    },
    {
      "epoch": 0.1328,
      "grad_norm": 0.5376033248254236,
      "learning_rate": 1.9990669724599336e-05,
      "loss": 0.0876,
      "step": 83
    },
    {
      "epoch": 0.1344,
      "grad_norm": 0.470270008076486,
      "learning_rate": 1.9989908502204295e-05,
      "loss": 0.0856,
      "step": 84
    },
    {
      "epoch": 0.136,
      "grad_norm": 0.5326255229735967,
      "learning_rate": 1.998911744842611e-05,
      "loss": 0.0812,
      "step": 85
    },
    {
      "epoch": 0.1376,
      "grad_norm": 0.9375418947984391,
      "learning_rate": 1.9988296565626988e-05,
      "loss": 0.0864,
      "step": 86
    },
    {
      "epoch": 0.1392,
      "grad_norm": 0.9653364692897405,
      "learning_rate": 1.9987445856258208e-05,
      "loss": 0.0775,
      "step": 87
    },
    {
      "epoch": 0.1408,
      "grad_norm": 0.510837102874891,
      "learning_rate": 1.9986565322860117e-05,
      "loss": 0.0758,
      "step": 88
    },
    {
      "epoch": 0.1424,
      "grad_norm": 0.46503877822117407,
      "learning_rate": 1.9985654968062122e-05,
      "loss": 0.0754,
      "step": 89
    },
    {
      "epoch": 0.144,
      "grad_norm": 0.4771621635835423,
      "learning_rate": 1.9984714794582682e-05,
      "loss": 0.0739,
      "step": 90
    },
    {
      "epoch": 0.1456,
      "grad_norm": 0.3851199281048143,
      "learning_rate": 1.9983744805229296e-05,
      "loss": 0.0741,
      "step": 91
    },
    {
      "epoch": 0.1472,
      "grad_norm": 0.5702319110120412,
      "learning_rate": 1.99827450028985e-05,
      "loss": 0.076,
      "step": 92
    },
    {
      "epoch": 0.1488,
      "grad_norm": 0.5111184671166494,
      "learning_rate": 1.998171539057586e-05,
      "loss": 0.0741,
      "step": 93
    },
    {
      "epoch": 0.1504,
      "grad_norm": 0.41546583111467167,
      "learning_rate": 1.9980655971335944e-05,
      "loss": 0.0609,
      "step": 94
    },
    {
      "epoch": 0.152,
      "grad_norm": 0.49956020283473984,
      "learning_rate": 1.9979566748342348e-05,
      "loss": 0.0638,
      "step": 95
    },
    {
      "epoch": 0.1536,
      "grad_norm": 0.3594293136690318,
      "learning_rate": 1.9978447724847655e-05,
      "loss": 0.0659,
      "step": 96
    },
    {
      "epoch": 0.1552,
      "grad_norm": 0.31456158706187765,
      "learning_rate": 1.9977298904193438e-05,
      "loss": 0.0688,
      "step": 97
    },
    {
      "epoch": 0.1568,
      "grad_norm": 0.31372590519120896,
      "learning_rate": 1.9976120289810247e-05,
      "loss": 0.0612,
      "step": 98
    },
    {
      "epoch": 0.1584,
      "grad_norm": 0.35164770270724616,
      "learning_rate": 1.997491188521761e-05,
      "loss": 0.0623,
      "step": 99
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3678653054533639,
      "learning_rate": 1.9973673694024002e-05,
      "loss": 0.0591,
      "step": 100
    },
    {
      "epoch": 0.1616,
      "grad_norm": 0.2830216452554222,
      "learning_rate": 1.997240571992685e-05,
      "loss": 0.0554,
      "step": 101
    },
    {
      "epoch": 0.1632,
      "grad_norm": 0.31115685625855377,
      "learning_rate": 1.9971107966712518e-05,
      "loss": 0.0581,
      "step": 102
    },
    {
      "epoch": 0.1648,
      "grad_norm": 0.6027402944866155,
      "learning_rate": 1.9969780438256295e-05,
      "loss": 0.0561,
      "step": 103
    },
    {
      "epoch": 0.1664,
      "grad_norm": 0.3062620836300149,
      "learning_rate": 1.9968423138522382e-05,
      "loss": 0.0561,
      "step": 104
    },
    {
      "epoch": 0.168,
      "grad_norm": 0.30374482823732385,
      "learning_rate": 1.9967036071563878e-05,
      "loss": 0.054,
      "step": 105
    },
    {
      "epoch": 0.1696,
      "grad_norm": 0.3215386574382714,
      "learning_rate": 1.996561924152278e-05,
      "loss": 0.055,
      "step": 106
    },
    {
      "epoch": 0.1712,
      "grad_norm": 0.3179166891991341,
      "learning_rate": 1.996417265262996e-05,
      "loss": 0.0526,
      "step": 107
    },
    {
      "epoch": 0.1728,
      "grad_norm": 0.2819886680887226,
      "learning_rate": 1.9962696309205146e-05,
      "loss": 0.0509,
      "step": 108
    },
    {
      "epoch": 0.1744,
      "grad_norm": 0.3863087049989975,
      "learning_rate": 1.996119021565693e-05,
      "loss": 0.0564,
      "step": 109
    },
    {
      "epoch": 0.176,
      "grad_norm": 0.38183766944400965,
      "learning_rate": 1.995965437648273e-05,
      "loss": 0.0541,
      "step": 110
    },
    {
      "epoch": 0.1776,
      "grad_norm": 0.3150541173369872,
      "learning_rate": 1.9958088796268794e-05,
      "loss": 0.054,
      "step": 111
    },
    {
      "epoch": 0.1792,
      "grad_norm": 0.3142985896656338,
      "learning_rate": 1.995649347969019e-05,
      "loss": 0.0493,
      "step": 112
    },
    {
      "epoch": 0.1808,
      "grad_norm": 0.2863614451714051,
      "learning_rate": 1.9954868431510764e-05,
      "loss": 0.0465,
      "step": 113
    },
    {
      "epoch": 0.1824,
      "grad_norm": 0.33168986501926395,
      "learning_rate": 1.995321365658317e-05,
      "loss": 0.0569,
      "step": 114
    },
    {
      "epoch": 0.184,
      "grad_norm": 0.2936366801022514,
      "learning_rate": 1.9951529159848805e-05,
      "loss": 0.0504,
      "step": 115
    },
    {
      "epoch": 0.1856,
      "grad_norm": 0.35989897726851766,
      "learning_rate": 1.994981494633784e-05,
      "loss": 0.0474,
      "step": 116
    },
    {
      "epoch": 0.1872,
      "grad_norm": 0.3118496134874041,
      "learning_rate": 1.9948071021169176e-05,
      "loss": 0.0471,
      "step": 117
    },
    {
      "epoch": 0.1888,
      "grad_norm": 0.3301667250792525,
      "learning_rate": 1.9946297389550433e-05,
      "loss": 0.0506,
      "step": 118
    },
    {
      "epoch": 0.1904,
      "grad_norm": 0.29854367027467466,
      "learning_rate": 1.9944494056777945e-05,
      "loss": 0.0491,
      "step": 119
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.2639233963987313,
      "learning_rate": 1.9942661028236746e-05,
      "loss": 0.0478,
      "step": 120
    },
    {
      "epoch": 0.1936,
      "grad_norm": 0.31310526044955456,
      "learning_rate": 1.9940798309400527e-05,
      "loss": 0.0454,
      "step": 121
    },
    {
      "epoch": 0.1952,
      "grad_norm": 0.30036459286759426,
      "learning_rate": 1.9938905905831657e-05,
      "loss": 0.0512,
      "step": 122
    },
    {
      "epoch": 0.1968,
      "grad_norm": 0.274927744652201,
      "learning_rate": 1.9936983823181132e-05,
      "loss": 0.049,
      "step": 123
    },
    {
      "epoch": 0.1984,
      "grad_norm": 0.3077251417327128,
      "learning_rate": 1.993503206718859e-05,
      "loss": 0.0448,
      "step": 124
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2984318065517011,
      "learning_rate": 1.993305064368227e-05,
      "loss": 0.0447,
      "step": 125
    },
    {
      "epoch": 0.2016,
      "grad_norm": 0.2562044510094792,
      "learning_rate": 1.9931039558578997e-05,
      "loss": 0.0422,
      "step": 126
    },
    {
      "epoch": 0.2032,
      "grad_norm": 0.22049955284817202,
      "learning_rate": 1.9928998817884185e-05,
      "loss": 0.0462,
      "step": 127
    },
    {
      "epoch": 0.2048,
      "grad_norm": 0.2402367528446714,
      "learning_rate": 1.9926928427691788e-05,
      "loss": 0.0416,
      "step": 128
    },
    {
      "epoch": 0.2064,
      "grad_norm": 0.2530328053587009,
      "learning_rate": 1.9924828394184308e-05,
      "loss": 0.0444,
      "step": 129
    },
    {
      "epoch": 0.208,
      "grad_norm": 0.23402020878255644,
      "learning_rate": 1.992269872363277e-05,
      "loss": 0.0454,
      "step": 130
    },
    {
      "epoch": 0.2096,
      "grad_norm": 0.27231943284727395,
      "learning_rate": 1.992053942239668e-05,
      "loss": 0.0432,
      "step": 131
    },
    {
      "epoch": 0.2112,
      "grad_norm": 0.30162603611918526,
      "learning_rate": 1.991835049692405e-05,
      "loss": 0.0461,
      "step": 132
    },
    {
      "epoch": 0.2128,
      "grad_norm": 0.5875657609066812,
      "learning_rate": 1.9916131953751342e-05,
      "loss": 0.0496,
      "step": 133
    },
    {
      "epoch": 0.2144,
      "grad_norm": 0.23444266640024533,
      "learning_rate": 1.991388379950346e-05,
      "loss": 0.0448,
      "step": 134
    },
    {
      "epoch": 0.216,
      "grad_norm": 0.24100730008249682,
      "learning_rate": 1.9911606040893742e-05,
      "loss": 0.0449,
      "step": 135
    },
    {
      "epoch": 0.2176,
      "grad_norm": 0.27409831637006027,
      "learning_rate": 1.9909298684723905e-05,
      "loss": 0.0429,
      "step": 136
    },
    {
      "epoch": 0.2192,
      "grad_norm": 0.21807084455312617,
      "learning_rate": 1.990696173788408e-05,
      "loss": 0.042,
      "step": 137
    },
    {
      "epoch": 0.2208,
      "grad_norm": 0.2859994449145259,
      "learning_rate": 1.9904595207352736e-05,
      "loss": 0.0431,
      "step": 138
    },
    {
      "epoch": 0.2224,
      "grad_norm": 0.20183362855681072,
      "learning_rate": 1.9902199100196697e-05,
      "loss": 0.0385,
      "step": 139
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.3453138382705126,
      "learning_rate": 1.9899773423571102e-05,
      "loss": 0.0424,
      "step": 140
    },
    {
      "epoch": 0.2256,
      "grad_norm": 0.5953531319319878,
      "learning_rate": 1.9897318184719386e-05,
      "loss": 0.0466,
      "step": 141
    },
    {
      "epoch": 0.2272,
      "grad_norm": 0.2458303919799798,
      "learning_rate": 1.9894833390973266e-05,
      "loss": 0.042,
      "step": 142
    },
    {
      "epoch": 0.2288,
      "grad_norm": 0.242002979674024,
      "learning_rate": 1.989231904975272e-05,
      "loss": 0.045,
      "step": 143
    },
    {
      "epoch": 0.2304,
      "grad_norm": 0.2312660284492851,
      "learning_rate": 1.9889775168565942e-05,
      "loss": 0.0409,
      "step": 144
    },
    {
      "epoch": 0.232,
      "grad_norm": 0.24766915847032975,
      "learning_rate": 1.9887201755009358e-05,
      "loss": 0.0407,
      "step": 145
    },
    {
      "epoch": 0.2336,
      "grad_norm": 0.3333321837420313,
      "learning_rate": 1.9884598816767563e-05,
      "loss": 0.0421,
      "step": 146
    },
    {
      "epoch": 0.2352,
      "grad_norm": 0.25937507599864257,
      "learning_rate": 1.988196636161333e-05,
      "loss": 0.0382,
      "step": 147
    },
    {
      "epoch": 0.2368,
      "grad_norm": 0.2644808268622511,
      "learning_rate": 1.987930439740757e-05,
      "loss": 0.0382,
      "step": 148
    },
    {
      "epoch": 0.2384,
      "grad_norm": 0.4004716853916845,
      "learning_rate": 1.987661293209931e-05,
      "loss": 0.0396,
      "step": 149
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.28747417262335284,
      "learning_rate": 1.9873891973725673e-05,
      "loss": 0.0379,
      "step": 150
    },
    {
      "epoch": 0.2416,
      "grad_norm": 1.1173736301665167,
      "learning_rate": 1.9871141530411854e-05,
      "loss": 0.0459,
      "step": 151
    },
    {
      "epoch": 0.2432,
      "grad_norm": 0.3332516879243943,
      "learning_rate": 1.98683616103711e-05,
      "loss": 0.0393,
      "step": 152
    },
    {
      "epoch": 0.2448,
      "grad_norm": 0.2922136524879718,
      "learning_rate": 1.986555222190467e-05,
      "loss": 0.04,
      "step": 153
    },
    {
      "epoch": 0.2464,
      "grad_norm": 0.29721043057775826,
      "learning_rate": 1.986271337340182e-05,
      "loss": 0.041,
      "step": 154
    },
    {
      "epoch": 0.248,
      "grad_norm": 0.29914012628358855,
      "learning_rate": 1.9859845073339788e-05,
      "loss": 0.0428,
      "step": 155
    },
    {
      "epoch": 0.2496,
      "grad_norm": 0.4076615698415316,
      "learning_rate": 1.9856947330283752e-05,
      "loss": 0.0418,
      "step": 156
    },
    {
      "epoch": 0.2512,
      "grad_norm": 0.48998089711242176,
      "learning_rate": 1.9854020152886816e-05,
      "loss": 0.046,
      "step": 157
    },
    {
      "epoch": 0.2528,
      "grad_norm": 0.3520972303139779,
      "learning_rate": 1.985106354988997e-05,
      "loss": 0.0383,
      "step": 158
    },
    {
      "epoch": 0.2544,
      "grad_norm": 0.2416225857243467,
      "learning_rate": 1.9848077530122083e-05,
      "loss": 0.0406,
      "step": 159
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.3351339395051045,
      "learning_rate": 1.984506210249986e-05,
      "loss": 0.0443,
      "step": 160
    },
    {
      "epoch": 0.2576,
      "grad_norm": 0.4373154457475244,
      "learning_rate": 1.984201727602783e-05,
      "loss": 0.0377,
      "step": 161
    },
    {
      "epoch": 0.2592,
      "grad_norm": 0.32347937035340885,
      "learning_rate": 1.9838943059798305e-05,
      "loss": 0.0404,
      "step": 162
    },
    {
      "epoch": 0.2608,
      "grad_norm": 0.1931293350449502,
      "learning_rate": 1.983583946299136e-05,
      "loss": 0.0361,
      "step": 163
    },
    {
      "epoch": 0.2624,
      "grad_norm": 0.23801575923761545,
      "learning_rate": 1.9832706494874812e-05,
      "loss": 0.0359,
      "step": 164
    },
    {
      "epoch": 0.264,
      "grad_norm": 0.2732645051219492,
      "learning_rate": 1.9829544164804172e-05,
      "loss": 0.0369,
      "step": 165
    },
    {
      "epoch": 0.2656,
      "grad_norm": 0.27126891539245734,
      "learning_rate": 1.982635248222264e-05,
      "loss": 0.0341,
      "step": 166
    },
    {
      "epoch": 0.2672,
      "grad_norm": 0.21752650221622016,
      "learning_rate": 1.9823131456661064e-05,
      "loss": 0.0318,
      "step": 167
    },
    {
      "epoch": 0.2688,
      "grad_norm": 0.2349754020387302,
      "learning_rate": 1.9819881097737917e-05,
      "loss": 0.0323,
      "step": 168
    },
    {
      "epoch": 0.2704,
      "grad_norm": 0.4203489034506969,
      "learning_rate": 1.9816601415159266e-05,
      "loss": 0.0431,
      "step": 169
    },
    {
      "epoch": 0.272,
      "grad_norm": 0.4203489034506969,
      "learning_rate": 1.9816601415159266e-05,
      "loss": 0.0348,
      "step": 170
    },
    {
      "epoch": 0.2736,
      "grad_norm": 0.44413902157418056,
      "learning_rate": 1.9813292418718734e-05,
      "loss": 0.0363,
      "step": 171
    },
    {
      "epoch": 0.2752,
      "grad_norm": 0.30897582190180134,
      "learning_rate": 1.980995411829749e-05,
      "loss": 0.0333,
      "step": 172
    },
    {
      "epoch": 0.2768,
      "grad_norm": 0.2564689775545095,
      "learning_rate": 1.9806586523864212e-05,
      "loss": 0.0341,
      "step": 173
    },
    {
      "epoch": 0.2784,
      "grad_norm": 0.2359324424976115,
      "learning_rate": 1.980318964547504e-05,
      "loss": 0.0371,
      "step": 174
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38324646806929435,
      "learning_rate": 1.9799763493273572e-05,
      "loss": 0.0337,
      "step": 175
    },
    {
      "epoch": 0.2816,
      "grad_norm": 0.2989848276810454,
      "learning_rate": 1.9796308077490817e-05,
      "loss": 0.0356,
      "step": 176
    },
    {
      "epoch": 0.2832,
      "grad_norm": 0.26092774592942874,
      "learning_rate": 1.9792823408445173e-05,
      "loss": 0.0337,
      "step": 177
    },
    {
      "epoch": 0.2848,
      "grad_norm": 0.24958450769694013,
      "learning_rate": 1.978930949654239e-05,
      "loss": 0.0335,
      "step": 178
    },
    {
      "epoch": 0.2864,
      "grad_norm": 0.27590569127811576,
      "learning_rate": 1.978576635227554e-05,
      "loss": 0.0348,
      "step": 179
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.3213869696488523,
      "learning_rate": 1.9782193986224997e-05,
      "loss": 0.0325,
      "step": 180
    },
    {
      "epoch": 0.2896,
      "grad_norm": 0.3128042058548295,
      "learning_rate": 1.9778592409058376e-05,
      "loss": 0.0375,
      "step": 181
    },
    {
      "epoch": 0.2912,
      "grad_norm": 1.0075076172124091,
      "learning_rate": 1.9774961631530543e-05,
      "loss": 0.0349,
      "step": 182
    },
    {
      "epoch": 0.2928,
      "grad_norm": 0.2446969181592618,
      "learning_rate": 1.9771301664483548e-05,
      "loss": 0.0323,
      "step": 183
    },
    {
      "epoch": 0.2944,
      "grad_norm": 0.2617986459213487,
      "learning_rate": 1.976761251884661e-05,
      "loss": 0.0351,
      "step": 184
    },
    {
      "epoch": 0.296,
      "grad_norm": 0.45657092656538206,
      "learning_rate": 1.976389420563607e-05,
      "loss": 0.0365,
      "step": 185
    },
    {
      "epoch": 0.2976,
      "grad_norm": 0.24512426907244425,
      "learning_rate": 1.9760146735955388e-05,
      "loss": 0.0278,
      "step": 186
    },
    {
      "epoch": 0.2992,
      "grad_norm": 0.24913183105659079,
      "learning_rate": 1.975637012099507e-05,
      "loss": 0.0342,
      "step": 187
    },
    {
      "epoch": 0.3008,
      "grad_norm": 0.2824975570969569,
      "learning_rate": 1.9752564372032655e-05,
      "loss": 0.0298,
      "step": 188
    },
    {
      "epoch": 0.3024,
      "grad_norm": 0.31008595066638983,
      "learning_rate": 1.97487295004327e-05,
      "loss": 0.0318,
      "step": 189
    },
    {
      "epoch": 0.304,
      "grad_norm": 0.3387799766405644,
      "learning_rate": 1.974486551764671e-05,
      "loss": 0.0339,
      "step": 190
    },
    {
      "epoch": 0.3056,
      "grad_norm": 0.4995885536634404,
      "learning_rate": 1.9740972435213114e-05,
      "loss": 0.0359,
      "step": 191
    },
    {
      "epoch": 0.3072,
      "grad_norm": 0.25349932173180195,
      "learning_rate": 1.973705026475726e-05,
      "loss": 0.0312,
      "step": 192
    },
    {
      "epoch": 0.3088,
      "grad_norm": 0.32648409714588955,
      "learning_rate": 1.9733099017991342e-05,
      "loss": 0.0315,
      "step": 193
    },
    {
      "epoch": 0.3104,
      "grad_norm": 0.3440565315834147,
      "learning_rate": 1.9729118706714377e-05,
      "loss": 0.0342,
      "step": 194
    },
    {
      "epoch": 0.312,
      "grad_norm": 0.2677942023605695,
      "learning_rate": 1.972510934281218e-05,
      "loss": 0.0326,
      "step": 195
    },
    {
      "epoch": 0.3136,
      "grad_norm": 0.386774474399475,
      "learning_rate": 1.9721070938257326e-05,
      "loss": 0.0361,
      "step": 196
    },
    {
      "epoch": 0.3152,
      "grad_norm": 0.2530848402763779,
      "learning_rate": 1.9717003505109097e-05,
      "loss": 0.0315,
      "step": 197
    },
    {
      "epoch": 0.3168,
      "grad_norm": 0.4090054430468855,
      "learning_rate": 1.971290705551347e-05,
      "loss": 0.0331,
      "step": 198
    },
    {
      "epoch": 0.3184,
      "grad_norm": 0.4367545289940036,
      "learning_rate": 1.9708781601703066e-05,
      "loss": 0.0316,
      "step": 199
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.19063531216095272,
      "learning_rate": 1.970462715599711e-05,
      "loss": 0.0325,
      "step": 200
    },
    {
      "epoch": 0.3216,
      "grad_norm": 0.36599672109503767,
      "learning_rate": 1.9700443730801412e-05,
      "loss": 0.0372,
      "step": 201
    },
    {
      "epoch": 0.3232,
      "grad_norm": 0.250894690133702,
      "learning_rate": 1.9696231338608317e-05,
      "loss": 0.0319,
      "step": 202
    },
    {
      "epoch": 0.3248,
      "grad_norm": 0.35665383810945905,
      "learning_rate": 1.9691989991996663e-05,
      "loss": 0.0299,
      "step": 203
    },
    {
      "epoch": 0.3264,
      "grad_norm": 0.26890697010330017,
      "learning_rate": 1.9687719703631757e-05,
      "loss": 0.0325,
      "step": 204
    },
    {
      "epoch": 0.328,
      "grad_norm": 0.25896028779850616,
      "learning_rate": 1.9683420486265328e-05,
      "loss": 0.0304,
      "step": 205
    },
    {
      "epoch": 0.3296,
      "grad_norm": 0.3009648507964905,
      "learning_rate": 1.967909235273549e-05,
      "loss": 0.032,
      "step": 206
    },
    {
      "epoch": 0.3312,
      "grad_norm": 0.20628455614138974,
      "learning_rate": 1.967473531596671e-05,
      "loss": 0.0294,
      "step": 207
    },
    {
      "epoch": 0.3328,
      "grad_norm": 0.3014924152402286,
      "learning_rate": 1.9670349388969758e-05,
      "loss": 0.0287,
      "step": 208
    },
    {
      "epoch": 0.3344,
      "grad_norm": 0.23648053994541607,
      "learning_rate": 1.966593458484168e-05,
      "loss": 0.0304,
      "step": 209
    },
    {
      "epoch": 0.336,
      "grad_norm": 0.3306354577588259,
      "learning_rate": 1.9661490916765752e-05,
      "loss": 0.029,
      "step": 210
    },
    {
      "epoch": 0.3376,
      "grad_norm": 0.3438378127543805,
      "learning_rate": 1.9657018398011435e-05,
      "loss": 0.0278,
      "step": 211
    },
    {
      "epoch": 0.3392,
      "grad_norm": 0.2168801651038601,
      "learning_rate": 1.9652517041934357e-05,
      "loss": 0.0258,
      "step": 212
    },
    {
      "epoch": 0.3408,
      "grad_norm": 0.26209554081420433,
      "learning_rate": 1.9647986861976246e-05,
      "loss": 0.028,
      "step": 213
    },
    {
      "epoch": 0.3424,
      "grad_norm": 0.29361599815797396,
      "learning_rate": 1.9643427871664912e-05,
      "loss": 0.0309,
      "step": 214
    },
    {
      "epoch": 0.344,
      "grad_norm": 0.2362773483998967,
      "learning_rate": 1.9638840084614182e-05,
      "loss": 0.0259,
      "step": 215
    },
    {
      "epoch": 0.3456,
      "grad_norm": 0.35454665238094957,
      "learning_rate": 1.963422351452389e-05,
      "loss": 0.0323,
      "step": 216
    },
    {
      "epoch": 0.3472,
      "grad_norm": 0.21029650807600475,
      "learning_rate": 1.9629578175179823e-05,
      "loss": 0.0271,
      "step": 217
    },
    {
      "epoch": 0.3488,
      "grad_norm": 0.26629914086279055,
      "learning_rate": 1.9624904080453656e-05,
      "loss": 0.0284,
      "step": 218
    },
    {
      "epoch": 0.3504,
      "grad_norm": 0.18941616218847454,
      "learning_rate": 1.9620201244302952e-05,
      "loss": 0.0276,
      "step": 219
    },
    {
      "epoch": 0.352,
      "grad_norm": 0.19426874809023298,
      "learning_rate": 1.9615469680771097e-05,
      "loss": 0.0282,
      "step": 220
    },
    {
      "epoch": 0.3536,
      "grad_norm": 0.2624611759558848,
      "learning_rate": 1.9610709403987248e-05,
      "loss": 0.029,
      "step": 221
    },
    {
      "epoch": 0.3552,
      "grad_norm": 0.325503693832948,
      "learning_rate": 1.960592042816632e-05,
      "loss": 0.026,
      "step": 222
    },
    {
      "epoch": 0.3568,
      "grad_norm": 0.20446996096247685,
      "learning_rate": 1.9601102767608924e-05,
      "loss": 0.0271,
      "step": 223
    },
    {
      "epoch": 0.3584,
      "grad_norm": 0.26119031688067684,
      "learning_rate": 1.9596256436701324e-05,
      "loss": 0.0261,
      "step": 224
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.17955547518671727,
      "learning_rate": 1.95913814499154e-05,
      "loss": 0.0276,
      "step": 225
    },
    {
      "epoch": 0.3616,
      "grad_norm": 0.24063273200380345,
      "learning_rate": 1.9586477821808597e-05,
      "loss": 0.0265,
      "step": 226
    },
    {
      "epoch": 0.3632,
      "grad_norm": 0.2557128126599869,
      "learning_rate": 1.95815455670239e-05,
      "loss": 0.0277,
      "step": 227
    },
    {
      "epoch": 0.3648,
      "grad_norm": 0.16440851052959135,
      "learning_rate": 1.957658470028977e-05,
      "loss": 0.0254,
      "step": 228
    },
    {
      "epoch": 0.3664,
      "grad_norm": 0.37790528555642366,
      "learning_rate": 1.9571595236420103e-05,
      "loss": 0.0251,
      "step": 229
    },
    {
      "epoch": 0.368,
      "grad_norm": 0.17331729000725077,
      "learning_rate": 1.95665771903142e-05,
      "loss": 0.025,
      "step": 230
    },
    {
      "epoch": 0.3696,
      "grad_norm": 0.22255498342727273,
      "learning_rate": 1.9561530576956703e-05,
      "loss": 0.026,
      "step": 231
    },
    {
      "epoch": 0.3712,
      "grad_norm": 0.13851311457314286,
      "learning_rate": 1.9556455411417575e-05,
      "loss": 0.0257,
      "step": 232
    },
    {
      "epoch": 0.3728,
      "grad_norm": 0.2180209443564617,
      "learning_rate": 1.955135170885202e-05,
      "loss": 0.0266,
      "step": 233
    },
    {
      "epoch": 0.3744,
      "grad_norm": 0.18936063079669935,
      "learning_rate": 1.9546219484500475e-05,
      "loss": 0.0254,
      "step": 234
    },
    {
      "epoch": 0.376,
      "grad_norm": 0.3323779097048608,
      "learning_rate": 1.9541058753688538e-05,
      "loss": 0.0302,
      "step": 235
    },
    {
      "epoch": 0.3776,
      "grad_norm": 0.3245826088945582,
      "learning_rate": 1.9535869531826938e-05,
      "loss": 0.028,
      "step": 236
    },
    {
      "epoch": 0.3792,
      "grad_norm": 0.19574917790356397,
      "learning_rate": 1.9530651834411477e-05,
      "loss": 0.0253,
      "step": 237
    },
    {
      "epoch": 0.3808,
      "grad_norm": 0.2537992886761661,
      "learning_rate": 1.952540567702299e-05,
      "loss": 0.0292,
      "step": 238
    },
    {
      "epoch": 0.3824,
      "grad_norm": 0.4225817760415451,
      "learning_rate": 1.95201310753273e-05,
      "loss": 0.0309,
      "step": 239
    },
    {
      "epoch": 0.384,
      "grad_norm": 0.2038343453442293,
      "learning_rate": 1.951482804507517e-05,
      "loss": 0.0251,
      "step": 240
    },
    {
      "epoch": 0.3856,
      "grad_norm": 0.49544587227207304,
      "learning_rate": 1.9509496602102253e-05,
      "loss": 0.0239,
      "step": 241
    },
    {
      "epoch": 0.3872,
      "grad_norm": 0.3482229630924854,
      "learning_rate": 1.9504136762329046e-05,
      "loss": 0.0273,
      "step": 242
    },
    {
      "epoch": 0.3888,
      "grad_norm": 0.42426491159244356,
      "learning_rate": 1.9498748541760845e-05,
      "loss": 0.0256,
      "step": 243
    },
    {
      "epoch": 0.3904,
      "grad_norm": 0.2564993321314424,
      "learning_rate": 1.949333195648769e-05,
      "loss": 0.0265,
      "step": 244
    },
    {
      "epoch": 0.392,
      "grad_norm": 0.2147695548639086,
      "learning_rate": 1.9487887022684336e-05,
      "loss": 0.0256,
      "step": 245
    },
    {
      "epoch": 0.3936,
      "grad_norm": 0.16683905641566563,
      "learning_rate": 1.9482413756610175e-05,
      "loss": 0.0237,
      "step": 246
    },
    {
      "epoch": 0.3952,
      "grad_norm": 0.18124224889405904,
      "learning_rate": 1.947691217460921e-05,
      "loss": 0.0255,
      "step": 247
    },
    {
      "epoch": 0.3968,
      "grad_norm": 0.22738342310219078,
      "learning_rate": 1.9471382293110004e-05,
      "loss": 0.0282,
      "step": 248
    },
    {
      "epoch": 0.3984,
      "grad_norm": 0.15315087071959668,
      "learning_rate": 1.946582412862562e-05,
      "loss": 0.0256,
      "step": 249
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.17131894088568825,
      "learning_rate": 1.9460237697753577e-05,
      "loss": 0.0253,
      "step": 250
    },
    {
      "epoch": 0.4016,
      "grad_norm": 0.2736617070066449,
      "learning_rate": 1.9454623017175814e-05,
      "loss": 0.0267,
      "step": 251
    },
    {
      "epoch": 0.4032,
      "grad_norm": 0.3657647944090099,
      "learning_rate": 1.9448980103658613e-05,
      "loss": 0.026,
      "step": 252
    },
    {
      "epoch": 0.4048,
      "grad_norm": 0.2876716836389149,
      "learning_rate": 1.9443308974052574e-05,
      "loss": 0.0251,
      "step": 253
    },
    {
      "epoch": 0.4064,
      "grad_norm": 0.25445616948155586,
      "learning_rate": 1.943760964529255e-05,
      "loss": 0.0271,
      "step": 254
    },
    {
      "epoch": 0.408,
      "grad_norm": 0.1854945634725827,
      "learning_rate": 1.9431882134397596e-05,
      "loss": 0.0237,
      "step": 255
    },
    {
      "epoch": 0.4096,
      "grad_norm": 0.20873308410010488,
      "learning_rate": 1.9426126458470936e-05,
      "loss": 0.0249,
      "step": 256
    },
    {
      "epoch": 0.4112,
      "grad_norm": 0.23299332351744328,
      "learning_rate": 1.9420342634699893e-05,
      "loss": 0.0279,
      "step": 257
    },
    {
      "epoch": 0.4128,
      "grad_norm": 0.19319678676101346,
      "learning_rate": 1.9414530680355837e-05,
      "loss": 0.0261,
      "step": 258
    },
    {
      "epoch": 0.4144,
      "grad_norm": 0.18150742033137932,
      "learning_rate": 1.9408690612794146e-05,
      "loss": 0.0257,
      "step": 259
    },
    {
      "epoch": 0.416,
      "grad_norm": 0.26277783543834804,
      "learning_rate": 1.9402822449454154e-05,
      "loss": 0.0278,
      "step": 260
    },
    {
      "epoch": 0.4176,
      "grad_norm": 0.24540077426443008,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 0.0255,
      "step": 261
    },
    {
      "epoch": 0.4192,
      "grad_norm": 0.24639773880399232,
      "learning_rate": 1.939100190561601e-05,
      "loss": 0.0255,
      "step": 262
    },
    {
      "epoch": 0.4208,
      "grad_norm": 0.18519198438379944,
      "learning_rate": 1.9385049560415794e-05,
      "loss": 0.024,
      "step": 263
    },
    {
      "epoch": 0.4224,
      "grad_norm": 0.17894539537653406,
      "learning_rate": 1.9379069190033042e-05,
      "loss": 0.0249,
      "step": 264
    },
    {
      "epoch": 0.424,
      "grad_norm": 0.32960028213367215,
      "learning_rate": 1.9373060812326053e-05,
      "loss": 0.0264,
      "step": 265
    },
    {
      "epoch": 0.4256,
      "grad_norm": 0.18565514288989918,
      "learning_rate": 1.936702444523675e-05,
      "loss": 0.0258,
      "step": 266
    },
    {
      "epoch": 0.4272,
      "grad_norm": 0.21328083113489418,
      "learning_rate": 1.9360960106790645e-05,
      "loss": 0.0267,
      "step": 267
    },
    {
      "epoch": 0.4288,
      "grad_norm": 0.7772355439669246,
      "learning_rate": 1.9354867815096772e-05,
      "loss": 0.0254,
      "step": 268
    },
    {
      "epoch": 0.4304,
      "grad_norm": 0.2870524330818712,
      "learning_rate": 1.9348747588347637e-05,
      "loss": 0.0237,
      "step": 269
    },
    {
      "epoch": 0.432,
      "grad_norm": 0.3739445599859368,
      "learning_rate": 1.9342599444819167e-05,
      "loss": 0.0273,
      "step": 270
    },
    {
      "epoch": 0.4336,
      "grad_norm": 0.2511413022176609,
      "learning_rate": 1.9336423402870655e-05,
      "loss": 0.025,
      "step": 271
    },
    {
      "epoch": 0.4352,
      "grad_norm": 0.2720251884441693,
      "learning_rate": 1.9330219480944693e-05,
      "loss": 0.0237,
      "step": 272
    },
    {
      "epoch": 0.4368,
      "grad_norm": 0.19155647163622136,
      "learning_rate": 1.932398769756714e-05,
      "loss": 0.0225,
      "step": 273
    },
    {
      "epoch": 0.4384,
      "grad_norm": 0.22998262921006735,
      "learning_rate": 1.931772807134704e-05,
      "loss": 0.0248,
      "step": 274
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.26222229165931915,
      "learning_rate": 1.9311440620976597e-05,
      "loss": 0.0249,
      "step": 275
    },
    {
      "epoch": 0.4416,
      "grad_norm": 0.2409259694452907,
      "learning_rate": 1.9305125365231087e-05,
      "loss": 0.0254,
      "step": 276
    },
    {
      "epoch": 0.4432,
      "grad_norm": 0.18412201144034487,
      "learning_rate": 1.9298782322968817e-05,
      "loss": 0.0242,
      "step": 277
    },
    {
      "epoch": 0.4448,
      "grad_norm": 0.23927348676565244,
      "learning_rate": 1.929241151313108e-05,
      "loss": 0.0243,
      "step": 278
    },
    {
      "epoch": 0.4464,
      "grad_norm": 0.3135936679260672,
      "learning_rate": 1.9286012954742078e-05,
      "loss": 0.0264,
      "step": 279
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.32007604599511913,
      "learning_rate": 1.9279586666908886e-05,
      "loss": 0.0252,
      "step": 280
    },
    {
      "epoch": 0.4496,
      "grad_norm": 0.3096904351422991,
      "learning_rate": 1.9273132668821363e-05,
      "loss": 0.0253,
      "step": 281
    },
    {
      "epoch": 0.4512,
      "grad_norm": 0.23490286212024764,
      "learning_rate": 1.9266650979752137e-05,
      "loss": 0.0257,
      "step": 282
    },
    {
      "epoch": 0.4528,
      "grad_norm": 0.3190306995921017,
      "learning_rate": 1.9260141619056507e-05,
      "loss": 0.0242,
      "step": 283
    },
    {
      "epoch": 0.4544,
      "grad_norm": 0.1460060583985359,
      "learning_rate": 1.925360460617242e-05,
      "loss": 0.0244,
      "step": 284
    },
    {
      "epoch": 0.456,
      "grad_norm": 0.40457351297947347,
      "learning_rate": 1.924703996062038e-05,
      "loss": 0.0273,
      "step": 285
    },
    {
      "epoch": 0.4576,
      "grad_norm": 0.21777864726658375,
      "learning_rate": 1.9240447702003422e-05,
      "loss": 0.0249,
      "step": 286
    },
    {
      "epoch": 0.4592,
      "grad_norm": 0.13886632445514155,
      "learning_rate": 1.9233827850007028e-05,
      "loss": 0.0231,
      "step": 287
    },
    {
      "epoch": 0.4608,
      "grad_norm": 0.23268263869417316,
      "learning_rate": 1.9227180424399082e-05,
      "loss": 0.0253,
      "step": 288
    },
    {
      "epoch": 0.4624,
      "grad_norm": 0.24274828301682744,
      "learning_rate": 1.9220505445029803e-05,
      "loss": 0.0268,
      "step": 289
    },
    {
      "epoch": 0.464,
      "grad_norm": 0.20450417450718378,
      "learning_rate": 1.9213802931831697e-05,
      "loss": 0.0232,
      "step": 290
    },
    {
      "epoch": 0.4656,
      "grad_norm": 0.16078185374649792,
      "learning_rate": 1.9207072904819484e-05,
      "loss": 0.0232,
      "step": 291
    },
    {
      "epoch": 0.4672,
      "grad_norm": 0.3118234085979711,
      "learning_rate": 1.9200315384090045e-05,
      "loss": 0.0268,
      "step": 292
    },
    {
      "epoch": 0.4688,
      "grad_norm": 0.1657397801331871,
      "learning_rate": 1.9193530389822364e-05,
      "loss": 0.0234,
      "step": 293
    },
    {
      "epoch": 0.4704,
      "grad_norm": 0.24790087220099571,
      "learning_rate": 1.9186717942277466e-05,
      "loss": 0.0249,
      "step": 294
    },
    {
      "epoch": 0.472,
      "grad_norm": 0.18698610147821218,
      "learning_rate": 1.9179878061798347e-05,
      "loss": 0.0242,
      "step": 295
    },
    {
      "epoch": 0.4736,
      "grad_norm": 0.20122110461948367,
      "learning_rate": 1.9173010768809934e-05,
      "loss": 0.0253,
      "step": 296
    },
    {
      "epoch": 0.4752,
      "grad_norm": 0.16444390761754338,
      "learning_rate": 1.9166116083819002e-05,
      "loss": 0.0239,
      "step": 297
    },
    {
      "epoch": 0.4768,
      "grad_norm": 0.2096031816916971,
      "learning_rate": 1.915919402741413e-05,
      "loss": 0.0237,
      "step": 298
    },
    {
      "epoch": 0.4784,
      "grad_norm": 0.2871382876650298,
      "learning_rate": 1.915224462026563e-05,
      "loss": 0.0247,
      "step": 299
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23396068054992797,
      "learning_rate": 1.9145267883125483e-05,
      "loss": 0.0226,
      "step": 300
    },
    {
      "epoch": 0.4816,
      "grad_norm": 0.17806677822199216,
      "learning_rate": 1.913826383682729e-05,
      "loss": 0.0228,
      "step": 301
    },
    {
      "epoch": 0.4832,
      "grad_norm": 0.22421810228175576,
      "learning_rate": 1.913123250228619e-05,
      "loss": 0.0222,
      "step": 302
    },
    {
      "epoch": 0.4848,
      "grad_norm": 0.2671737148491569,
      "learning_rate": 1.912417390049882e-05,
      "loss": 0.0258,
      "step": 303
    },
    {
      "epoch": 0.4864,
      "grad_norm": 0.22521421563326308,
      "learning_rate": 1.9117088052543233e-05,
      "loss": 0.0259,
      "step": 304
    },
    {
      "epoch": 0.488,
      "grad_norm": 0.2512719570632976,
      "learning_rate": 1.9109974979578852e-05,
      "loss": 0.0223,
      "step": 305
    },
    {
      "epoch": 0.4896,
      "grad_norm": 0.200102823046399,
      "learning_rate": 1.9102834702846387e-05,
      "loss": 0.0244,
      "step": 306
    },
    {
      "epoch": 0.4912,
      "grad_norm": 0.2322391139619468,
      "learning_rate": 1.909566724366779e-05,
      "loss": 0.0248,
      "step": 307
    },
    {
      "epoch": 0.4928,
      "grad_norm": 0.33073866828097187,
      "learning_rate": 1.9088472623446182e-05,
      "loss": 0.026,
      "step": 308
    },
    {
      "epoch": 0.4944,
      "grad_norm": 0.1893349038318749,
      "learning_rate": 1.9081250863665794e-05,
      "loss": 0.0235,
      "step": 309
    },
    {
      "epoch": 0.496,
      "grad_norm": 0.29277990880028976,
      "learning_rate": 1.9074001985891893e-05,
      "loss": 0.0265,
      "step": 310
    },
    {
      "epoch": 0.4976,
      "grad_norm": 0.19477738061872754,
      "learning_rate": 1.9066726011770725e-05,
      "loss": 0.0228,
      "step": 311
    },
    {
      "epoch": 0.4992,
      "grad_norm": 0.22915444168212584,
      "learning_rate": 1.9059422963029464e-05,
      "loss": 0.0256,
      "step": 312
    },
    {
      "epoch": 0.5008,
      "grad_norm": 0.13982238568398425,
      "learning_rate": 1.905209286147611e-05,
      "loss": 0.022,
      "step": 313
    },
    {
      "epoch": 0.5024,
      "grad_norm": 0.30311950923241815,
      "learning_rate": 1.9044735728999472e-05,
      "loss": 0.0261,
      "step": 314
    },
    {
      "epoch": 0.504,
      "grad_norm": 0.2124747140890154,
      "learning_rate": 1.903735158756905e-05,
      "loss": 0.0205,
      "step": 315
    },
    {
      "epoch": 0.5056,
      "grad_norm": 0.31912333300382856,
      "learning_rate": 1.902994045923502e-05,
      "loss": 0.0236,
      "step": 316
    },
    {
      "epoch": 0.5072,
      "grad_norm": 0.205258171466095,
      "learning_rate": 1.9022502366128136e-05,
      "loss": 0.0249,
      "step": 317
    },
    {
      "epoch": 0.5088,
      "grad_norm": 0.25082631668722966,
      "learning_rate": 1.901503733045967e-05,
      "loss": 0.0249,
      "step": 318
    },
    {
      "epoch": 0.5104,
      "grad_norm": 0.293085694954146,
      "learning_rate": 1.9007545374521354e-05,
      "loss": 0.0238,
      "step": 319
    },
    {
      "epoch": 0.512,
      "grad_norm": 0.2181476235450802,
      "learning_rate": 1.90000265206853e-05,
      "loss": 0.0233,
      "step": 320
    },
    {
      "epoch": 0.5136,
      "grad_norm": 0.16868250926306455,
      "learning_rate": 1.8992480791403957e-05,
      "loss": 0.0211,
      "step": 321
    },
    {
      "epoch": 0.5152,
      "grad_norm": 0.1452284105599233,
      "learning_rate": 1.898490820921001e-05,
      "loss": 0.0235,
      "step": 322
    },
    {
      "epoch": 0.5168,
      "grad_norm": 0.16961367231195754,
      "learning_rate": 1.897730879671634e-05,
      "loss": 0.0217,
      "step": 323
    },
    {
      "epoch": 0.5184,
      "grad_norm": 0.2427685545953266,
      "learning_rate": 1.8969682576615947e-05,
      "loss": 0.0238,
      "step": 324
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2560872454036068,
      "learning_rate": 1.8962029571681887e-05,
      "loss": 0.024,
      "step": 325
    },
    {
      "epoch": 0.5216,
      "grad_norm": 0.42672805528036134,
      "learning_rate": 1.8954349804767185e-05,
      "loss": 0.0227,
      "step": 326
    },
    {
      "epoch": 0.5232,
      "grad_norm": 0.25995213776532555,
      "learning_rate": 1.8946643298804794e-05,
      "loss": 0.0221,
      "step": 327
    },
    {
      "epoch": 0.5248,
      "grad_norm": 0.2995897648125354,
      "learning_rate": 1.8938910076807514e-05,
      "loss": 0.0262,
      "step": 328
    },
    {
      "epoch": 0.5264,
      "grad_norm": 0.2088551905545114,
      "learning_rate": 1.8931150161867917e-05,
      "loss": 0.0206,
      "step": 329
    },
    {
      "epoch": 0.528,
      "grad_norm": 0.43085085565275305,
      "learning_rate": 1.892336357715829e-05,
      "loss": 0.0252,
      "step": 330
    },
    {
      "epoch": 0.5296,
      "grad_norm": 0.36257688507162905,
      "learning_rate": 1.891555034593055e-05,
      "loss": 0.0235,
      "step": 331
    },
    {
      "epoch": 0.5312,
      "grad_norm": 0.20771270465771838,
      "learning_rate": 1.8907710491516197e-05,
      "loss": 0.022,
      "step": 332
    },
    {
      "epoch": 0.5328,
      "grad_norm": 0.17509810477386462,
      "learning_rate": 1.8899844037326227e-05,
      "loss": 0.0211,
      "step": 333
    },
    {
      "epoch": 0.5344,
      "grad_norm": 0.12538653447060122,
      "learning_rate": 1.889195100685106e-05,
      "loss": 0.0193,
      "step": 334
    },
    {
      "epoch": 0.536,
      "grad_norm": 0.1980666975815131,
      "learning_rate": 1.8884031423660492e-05,
      "loss": 0.0209,
      "step": 335
    },
    {
      "epoch": 0.5376,
      "grad_norm": 0.26669488594459173,
      "learning_rate": 1.8876085311403592e-05,
      "loss": 0.0223,
      "step": 336
    },
    {
      "epoch": 0.5392,
      "grad_norm": 0.21213789151153697,
      "learning_rate": 1.8868112693808664e-05,
      "loss": 0.0227,
      "step": 337
    },
    {
      "epoch": 0.5408,
      "grad_norm": 0.3009564209037192,
      "learning_rate": 1.8860113594683148e-05,
      "loss": 0.0243,
      "step": 338
    },
    {
      "epoch": 0.5424,
      "grad_norm": 0.16434507191527442,
      "learning_rate": 1.8852088037913577e-05,
      "loss": 0.0193,
      "step": 339
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.25856557410366665,
      "learning_rate": 1.884403604746547e-05,
      "loss": 0.0235,
      "step": 340
    },
    {
      "epoch": 0.5456,
      "grad_norm": 0.2753428966537922,
      "learning_rate": 1.8835957647383304e-05,
      "loss": 0.0243,
      "step": 341
    },
    {
      "epoch": 0.5472,
      "grad_norm": 0.23000290993852823,
      "learning_rate": 1.8827852861790398e-05,
      "loss": 0.0225,
      "step": 342
    },
    {
      "epoch": 0.5488,
      "grad_norm": 0.18518777898446165,
      "learning_rate": 1.8819721714888878e-05,
      "loss": 0.0226,
      "step": 343
    },
    {
      "epoch": 0.5504,
      "grad_norm": 0.22708845342130995,
      "learning_rate": 1.8811564230959585e-05,
      "loss": 0.0222,
      "step": 344
    },
    {
      "epoch": 0.552,
      "grad_norm": 0.22327313120164785,
      "learning_rate": 1.8803380434362e-05,
      "loss": 0.0223,
      "step": 345
    },
    {
      "epoch": 0.5536,
      "grad_norm": 0.23016725124648443,
      "learning_rate": 1.879517034953418e-05,
      "loss": 0.0219,
      "step": 346
    },
    {
      "epoch": 0.5552,
      "grad_norm": 0.12311146499918242,
      "learning_rate": 1.878693400099269e-05,
      "loss": 0.0188,
      "step": 347
    },
    {
      "epoch": 0.5568,
      "grad_norm": 0.25520263166702184,
      "learning_rate": 1.8778671413332513e-05,
      "loss": 0.0217,
      "step": 348
    },
    {
      "epoch": 0.5584,
      "grad_norm": 0.17375942859806887,
      "learning_rate": 1.877038261122699e-05,
      "loss": 0.0227,
      "step": 349
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2868786417041136,
      "learning_rate": 1.8762067619427745e-05,
      "loss": 0.0232,
      "step": 350
    },
    {
      "epoch": 0.5616,
      "grad_norm": 0.22263963887687188,
      "learning_rate": 1.87537264627646e-05,
      "loss": 0.0205,
      "step": 351
    },
    {
      "epoch": 0.5632,
      "grad_norm": 0.3216611897383196,
      "learning_rate": 1.8745359166145526e-05,
      "loss": 0.0222,
      "step": 352
    },
    {
      "epoch": 0.5648,
      "grad_norm": 0.348533541354947,
      "learning_rate": 1.8736965754556527e-05,
      "loss": 0.024,
      "step": 353
    },
    {
      "epoch": 0.5664,
      "grad_norm": 0.14815375140096462,
      "learning_rate": 1.8728546253061614e-05,
      "loss": 0.0189,
      "step": 354
    },
    {
      "epoch": 0.568,
      "grad_norm": 0.276884259305941,
      "learning_rate": 1.8720100686802693e-05,
      "loss": 0.0196,
      "step": 355
    },
    {
      "epoch": 0.5696,
      "grad_norm": 0.24410384498418092,
      "learning_rate": 1.8711629080999506e-05,
      "loss": 0.0222,
      "step": 356
    },
    {
      "epoch": 0.5712,
      "grad_norm": 0.33067569013923137,
      "learning_rate": 1.8703131460949555e-05,
      "loss": 0.022,
      "step": 357
    },
    {
      "epoch": 0.5728,
      "grad_norm": 0.22661205766856518,
      "learning_rate": 1.869460785202802e-05,
      "loss": 0.0185,
      "step": 358
    },
    {
      "epoch": 0.5744,
      "grad_norm": 0.503638421558941,
      "learning_rate": 1.86860582796877e-05,
      "loss": 0.0263,
      "step": 359
    },
    {
      "epoch": 0.576,
      "grad_norm": 0.30972726099984216,
      "learning_rate": 1.8677482769458905e-05,
      "loss": 0.022,
      "step": 360
    },
    {
      "epoch": 0.5776,
      "grad_norm": 0.3154407238669633,
      "learning_rate": 1.866888134694942e-05,
      "loss": 0.0214,
      "step": 361
    },
    {
      "epoch": 0.5792,
      "grad_norm": 0.3588250103478784,
      "learning_rate": 1.866025403784439e-05,
      "loss": 0.024,
      "step": 362
    },
    {
      "epoch": 0.5808,
      "grad_norm": 0.2103162608999855,
      "learning_rate": 1.865160086790627e-05,
      "loss": 0.0201,
      "step": 363
    },
    {
      "epoch": 0.5824,
      "grad_norm": 0.24144473483802972,
      "learning_rate": 1.8642921862974742e-05,
      "loss": 0.0197,
      "step": 364
    },
    {
      "epoch": 0.584,
      "grad_norm": 0.8072550367254895,
      "learning_rate": 1.8634217048966638e-05,
      "loss": 0.0248,
      "step": 365
    },
    {
      "epoch": 0.5856,
      "grad_norm": 0.27443717495137954,
      "learning_rate": 1.8625486451875843e-05,
      "loss": 0.0229,
      "step": 366
    },
    {
      "epoch": 0.5872,
      "grad_norm": 0.3238915337976546,
      "learning_rate": 1.861673009777325e-05,
      "loss": 0.021,
      "step": 367
    },
    {
      "epoch": 0.5888,
      "grad_norm": 0.2769206307667872,
      "learning_rate": 1.8607948012806664e-05,
      "loss": 0.0254,
      "step": 368
    },
    {
      "epoch": 0.5904,
      "grad_norm": 0.35901812150739465,
      "learning_rate": 1.8599140223200716e-05,
      "loss": 0.0235,
      "step": 369
    },
    {
      "epoch": 0.592,
      "grad_norm": 0.4209880875878923,
      "learning_rate": 1.859030675525681e-05,
      "loss": 0.0228,
      "step": 370
    },
    {
      "epoch": 0.5936,
      "grad_norm": 0.2743855954190781,
      "learning_rate": 1.858144763535302e-05,
      "loss": 0.0213,
      "step": 371
    },
    {
      "epoch": 0.5952,
      "grad_norm": 0.4891783966635455,
      "learning_rate": 1.857256288994402e-05,
      "loss": 0.0239,
      "step": 372
    },
    {
      "epoch": 0.5968,
      "grad_norm": 0.3334065452478207,
      "learning_rate": 1.8563652545561014e-05,
      "loss": 0.0226,
      "step": 373
    },
    {
      "epoch": 0.5984,
      "grad_norm": 0.22921939143541975,
      "learning_rate": 1.855471662881164e-05,
      "loss": 0.0211,
      "step": 374
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2083502675972401,
      "learning_rate": 1.8545755166379898e-05,
      "loss": 0.0188,
      "step": 375
    },
    {
      "epoch": 0.6016,
      "grad_norm": 0.1738970990934248,
      "learning_rate": 1.8536768185026085e-05,
      "loss": 0.0208,
      "step": 376
    },
    {
      "epoch": 0.6032,
      "grad_norm": 0.2288134807094265,
      "learning_rate": 1.852775571158668e-05,
      "loss": 0.0226,
      "step": 377
    },
    {
      "epoch": 0.6048,
      "grad_norm": 0.21350233078025635,
      "learning_rate": 1.85187177729743e-05,
      "loss": 0.0231,
      "step": 378
    },
    {
      "epoch": 0.6064,
      "grad_norm": 0.24932344963951814,
      "learning_rate": 1.850965439617761e-05,
      "loss": 0.0232,
      "step": 379
    },
    {
      "epoch": 0.608,
      "grad_norm": 0.2973275026012781,
      "learning_rate": 1.8500565608261215e-05,
      "loss": 0.022,
      "step": 380
    },
    {
      "epoch": 0.6096,
      "grad_norm": 0.31199357328500743,
      "learning_rate": 1.8491451436365628e-05,
      "loss": 0.022,
      "step": 381
    },
    {
      "epoch": 0.6112,
      "grad_norm": 0.19071566443042118,
      "learning_rate": 1.848231190770714e-05,
      "loss": 0.02,
      "step": 382
    },
    {
      "epoch": 0.6128,
      "grad_norm": 0.13039451500974714,
      "learning_rate": 1.8473147049577777e-05,
      "loss": 0.0194,
      "step": 383
    },
    {
      "epoch": 0.6144,
      "grad_norm": 0.1709393072560162,
      "learning_rate": 1.8463956889345195e-05,
      "loss": 0.0209,
      "step": 384
    },
    {
      "epoch": 0.616,
      "grad_norm": 0.4326244410727375,
      "learning_rate": 1.8454741454452604e-05,
      "loss": 0.0243,
      "step": 385
    },
    {
      "epoch": 0.6176,
      "grad_norm": 0.29733583141209047,
      "learning_rate": 1.8445500772418697e-05,
      "loss": 0.0229,
      "step": 386
    },
    {
      "epoch": 0.6192,
      "grad_norm": 0.18634249572348707,
      "learning_rate": 1.843623487083755e-05,
      "loss": 0.0198,
      "step": 387
    },
    {
      "epoch": 0.6208,
      "grad_norm": 0.17262948208342496,
      "learning_rate": 1.842694377737855e-05,
      "loss": 0.0169,
      "step": 388
    },
    {
      "epoch": 0.6224,
      "grad_norm": 0.21827100172380481,
      "learning_rate": 1.8417627519786317e-05,
      "loss": 0.0212,
      "step": 389
    },
    {
      "epoch": 0.624,
      "grad_norm": 0.25942130447905226,
      "learning_rate": 1.8408286125880605e-05,
      "loss": 0.0224,
      "step": 390
    },
    {
      "epoch": 0.6256,
      "grad_norm": 0.40627656684977626,
      "learning_rate": 1.839891962355624e-05,
      "loss": 0.0196,
      "step": 391
    },
    {
      "epoch": 0.6272,
      "grad_norm": 0.16973658798713487,
      "learning_rate": 1.8389528040783014e-05,
      "loss": 0.0207,
      "step": 392
    },
    {
      "epoch": 0.6288,
      "grad_norm": 0.21438004340509942,
      "learning_rate": 1.838011140560562e-05,
      "loss": 0.0198,
      "step": 393
    },
    {
      "epoch": 0.6304,
      "grad_norm": 0.31232945857171035,
      "learning_rate": 1.8370669746143566e-05,
      "loss": 0.0203,
      "step": 394
    },
    {
      "epoch": 0.632,
      "grad_norm": 0.2146851485289993,
      "learning_rate": 1.836120309059107e-05,
      "loss": 0.0195,
      "step": 395
    },
    {
      "epoch": 0.6336,
      "grad_norm": 0.24251036655589733,
      "learning_rate": 1.835171146721701e-05,
      "loss": 0.0172,
      "step": 396
    },
    {
      "epoch": 0.6352,
      "grad_norm": 0.3042791151359387,
      "learning_rate": 1.8342194904364815e-05,
      "loss": 0.0183,
      "step": 397
    },
    {
      "epoch": 0.6368,
      "grad_norm": 0.26113605341230206,
      "learning_rate": 1.8332653430452375e-05,
      "loss": 0.0202,
      "step": 398
    },
    {
      "epoch": 0.6384,
      "grad_norm": 0.34152736814393164,
      "learning_rate": 1.8323087073971996e-05,
      "loss": 0.0221,
      "step": 399
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.22308120700113318,
      "learning_rate": 1.831349586349026e-05,
      "loss": 0.0183,
      "step": 400
    },
    {
      "epoch": 0.6416,
      "grad_norm": 0.3146848923252582,
      "learning_rate": 1.8303879827647977e-05,
      "loss": 0.019,
      "step": 401
    },
    {
      "epoch": 0.6432,
      "grad_norm": 0.3138495171429804,
      "learning_rate": 1.8294238995160093e-05,
      "loss": 0.0179,
      "step": 402
    },
    {
      "epoch": 0.6448,
      "grad_norm": 0.34576002225133945,
      "learning_rate": 1.8284573394815596e-05,
      "loss": 0.0146,
      "step": 403
    },
    {
      "epoch": 0.6464,
      "grad_norm": 0.32694901216997213,
      "learning_rate": 1.8274883055477436e-05,
      "loss": 0.0212,
      "step": 404
    },
    {
      "epoch": 0.648,
      "grad_norm": 0.35767715776140246,
      "learning_rate": 1.826516800608244e-05,
      "loss": 0.0222,
      "step": 405
    },
    {
      "epoch": 0.6496,
      "grad_norm": 0.2689394197658266,
      "learning_rate": 1.8255428275641212e-05,
      "loss": 0.017,
      "step": 406
    },
    {
      "epoch": 0.6512,
      "grad_norm": 0.3106575036482928,
      "learning_rate": 1.8245663893238075e-05,
      "loss": 0.0253,
      "step": 407
    },
    {
      "epoch": 0.6528,
      "grad_norm": 0.27604524063461516,
      "learning_rate": 1.823587488803095e-05,
      "loss": 0.0152,
      "step": 408
    },
    {
      "epoch": 0.6544,
      "grad_norm": 0.47681734881426474,
      "learning_rate": 1.8226061289251297e-05,
      "loss": 0.0205,
      "step": 409
    },
    {
      "epoch": 0.656,
      "grad_norm": 0.21430013633514908,
      "learning_rate": 1.821622312620401e-05,
      "loss": 0.0182,
      "step": 410
    },
    {
      "epoch": 0.6576,
      "grad_norm": 0.30350010790826815,
      "learning_rate": 1.8206360428267332e-05,
      "loss": 0.0191,
      "step": 411
    },
    {
      "epoch": 0.6592,
      "grad_norm": 0.2860070509652396,
      "learning_rate": 1.8196473224892784e-05,
      "loss": 0.019,
      "step": 412
    },
    {
      "epoch": 0.6608,
      "grad_norm": 0.28196191886430455,
      "learning_rate": 1.8186561545605055e-05,
      "loss": 0.0206,
      "step": 413
    },
    {
      "epoch": 0.6624,
      "grad_norm": 0.3079688364968603,
      "learning_rate": 1.817662542000192e-05,
      "loss": 0.0184,
      "step": 414
    },
    {
      "epoch": 0.664,
      "grad_norm": 0.3122700677645408,
      "learning_rate": 1.816666487775416e-05,
      "loss": 0.0178,
      "step": 415
    },
    {
      "epoch": 0.6656,
      "grad_norm": 0.2487569400300341,
      "learning_rate": 1.815667994860547e-05,
      "loss": 0.0194,
      "step": 416
    },
    {
      "epoch": 0.6672,
      "grad_norm": 0.2155913058486151,
      "learning_rate": 1.8146670662372353e-05,
      "loss": 0.0186,
      "step": 417
    },
    {
      "epoch": 0.6688,
      "grad_norm": 0.3581498362161358,
      "learning_rate": 1.813663704894407e-05,
      "loss": 0.0194,
      "step": 418
    },
    {
      "epoch": 0.6704,
      "grad_norm": 0.25611597657771756,
      "learning_rate": 1.8126579138282502e-05,
      "loss": 0.0173,
      "step": 419
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.33612713879076267,
      "learning_rate": 1.8116496960422108e-05,
      "loss": 0.0171,
      "step": 420
    },
    {
      "epoch": 0.6736,
      "grad_norm": 0.26310467780127467,
      "learning_rate": 1.8106390545469797e-05,
      "loss": 0.0195,
      "step": 421
    },
    {
      "epoch": 0.6752,
      "grad_norm": 0.28730353462201863,
      "learning_rate": 1.809625992360485e-05,
      "loss": 0.0195,
      "step": 422
    },
    {
      "epoch": 0.6768,
      "grad_norm": 0.35988954674299883,
      "learning_rate": 1.8086105125078858e-05,
      "loss": 0.0195,
      "step": 423
    },
    {
      "epoch": 0.6784,
      "grad_norm": 0.2730538724372587,
      "learning_rate": 1.8075926180215576e-05,
      "loss": 0.0175,
      "step": 424
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23158565001568174,
      "learning_rate": 1.8065723119410885e-05,
      "loss": 0.0156,
      "step": 425
    },
    {
      "epoch": 0.6816,
      "grad_norm": 0.34307170718940355,
      "learning_rate": 1.805549597313267e-05,
      "loss": 0.0169,
      "step": 426
    },
    {
      "epoch": 0.6832,
      "grad_norm": 0.3044701690296964,
      "learning_rate": 1.804524477192075e-05,
      "loss": 0.021,
      "step": 427
    },
    {
      "epoch": 0.6848,
      "grad_norm": 0.3284070104407129,
      "learning_rate": 1.803496954638676e-05,
      "loss": 0.0196,
      "step": 428
    },
    {
      "epoch": 0.6864,
      "grad_norm": 0.5072657066644937,
      "learning_rate": 1.8024670327214084e-05,
      "loss": 0.0225,
      "step": 429
    },
    {
      "epoch": 0.688,
      "grad_norm": 0.2941198460671475,
      "learning_rate": 1.8014347145157757e-05,
      "loss": 0.0188,
      "step": 430
    },
    {
      "epoch": 0.6896,
      "grad_norm": 0.25643271150318425,
      "learning_rate": 1.8004000031044363e-05,
      "loss": 0.0159,
      "step": 431
    },
    {
      "epoch": 0.6912,
      "grad_norm": 0.372899223911401,
      "learning_rate": 1.799362901577196e-05,
      "loss": 0.0183,
      "step": 432
    },
    {
      "epoch": 0.6928,
      "grad_norm": 0.335477210653575,
      "learning_rate": 1.798323413030997e-05,
      "loss": 0.019,
      "step": 433
    },
    {
      "epoch": 0.6944,
      "grad_norm": 0.34141551328982883,
      "learning_rate": 1.7972815405699105e-05,
      "loss": 0.0168,
      "step": 434
    },
    {
      "epoch": 0.696,
      "grad_norm": 0.3393022535549739,
      "learning_rate": 1.796237287305125e-05,
      "loss": 0.0204,
      "step": 435
    },
    {
      "epoch": 0.6976,
      "grad_norm": 0.2215319109318955,
      "learning_rate": 1.7951906563549397e-05,
      "loss": 0.0177,
      "step": 436
    },
    {
      "epoch": 0.6992,
      "grad_norm": 0.33171031082369407,
      "learning_rate": 1.7941416508447537e-05,
      "loss": 0.0173,
      "step": 437
    },
    {
      "epoch": 0.7008,
      "grad_norm": 0.47259061929176016,
      "learning_rate": 1.793090273907056e-05,
      "loss": 0.0162,
      "step": 438
    },
    {
      "epoch": 0.7024,
      "grad_norm": 0.17497052475464156,
      "learning_rate": 1.792036528681418e-05,
      "loss": 0.0158,
      "step": 439
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.3207435066493352,
      "learning_rate": 1.7909804183144837e-05,
      "loss": 0.0189,
      "step": 440
    },
    {
      "epoch": 0.7056,
      "grad_norm": 0.3768624074051605,
      "learning_rate": 1.789921945959958e-05,
      "loss": 0.0166,
      "step": 441
    },
    {
      "epoch": 0.7072,
      "grad_norm": 0.29665002817002356,
      "learning_rate": 1.7888611147786003e-05,
      "loss": 0.0159,
      "step": 442
    },
    {
      "epoch": 0.7088,
      "grad_norm": 0.3207389922464664,
      "learning_rate": 1.7877979279382135e-05,
      "loss": 0.0203,
      "step": 443
    },
    {
      "epoch": 0.7104,
      "grad_norm": 0.18890591065624907,
      "learning_rate": 1.786732388613635e-05,
      "loss": 0.0165,
      "step": 444
    },
    {
      "epoch": 0.712,
      "grad_norm": 0.23594625568492894,
      "learning_rate": 1.7856644999867264e-05,
      "loss": 0.0204,
      "step": 445
    },
    {
      "epoch": 0.7136,
      "grad_norm": 0.2552483310094428,
      "learning_rate": 1.784594265246366e-05,
      "loss": 0.0168,
      "step": 446
    },
    {
      "epoch": 0.7152,
      "grad_norm": 0.24151614992913603,
      "learning_rate": 1.783521687588437e-05,
      "loss": 0.0167,
      "step": 447
    },
    {
      "epoch": 0.7168,
      "grad_norm": 0.3604766090360732,
      "learning_rate": 1.782446770215819e-05,
      "loss": 0.022,
      "step": 448
    },
    {
      "epoch": 0.7184,
      "grad_norm": 0.24693074827496292,
      "learning_rate": 1.781369516338378e-05,
      "loss": 0.0188,
      "step": 449
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.51301858757728,
      "learning_rate": 1.7802899291729585e-05,
      "loss": 0.0188,
      "step": 450
    },
    {
      "epoch": 0.7216,
      "grad_norm": 0.44328837920573166,
      "learning_rate": 1.779208011943371e-05,
      "loss": 0.0171,
      "step": 451
    },
    {
      "epoch": 0.7232,
      "grad_norm": 0.3028936714557896,
      "learning_rate": 1.7781237678803845e-05,
      "loss": 0.0187,
      "step": 452
    },
    {
      "epoch": 0.7248,
      "grad_norm": 0.19178441580412345,
      "learning_rate": 1.777037200221717e-05,
      "loss": 0.0165,
      "step": 453
    },
    {
      "epoch": 0.7264,
      "grad_norm": 0.1723409498307265,
      "learning_rate": 1.775948312212024e-05,
      "loss": 0.0164,
      "step": 454
    },
    {
      "epoch": 0.728,
      "grad_norm": 0.5766940297412076,
      "learning_rate": 1.77485710710289e-05,
      "loss": 0.0152,
      "step": 455
    },
    {
      "epoch": 0.7296,
      "grad_norm": 0.3736278036944696,
      "learning_rate": 1.7737635881528198e-05,
      "loss": 0.0192,
      "step": 456
    },
    {
      "epoch": 0.7312,
      "grad_norm": 0.37918511783155534,
      "learning_rate": 1.7726677586272263e-05,
      "loss": 0.0185,
      "step": 457
    },
    {
      "epoch": 0.7328,
      "grad_norm": 0.31748017727601047,
      "learning_rate": 1.7715696217984233e-05,
      "loss": 0.0174,
      "step": 458
    },
    {
      "epoch": 0.7344,
      "grad_norm": 0.28283710719929595,
      "learning_rate": 1.7704691809456142e-05,
      "loss": 0.017,
      "step": 459
    },
    {
      "epoch": 0.736,
      "grad_norm": 0.21831749033244646,
      "learning_rate": 1.7693664393548822e-05,
      "loss": 0.0136,
      "step": 460
    },
    {
      "epoch": 0.7376,
      "grad_norm": 0.3209663077765436,
      "learning_rate": 1.7682614003191807e-05,
      "loss": 0.0164,
      "step": 461
    },
    {
      "epoch": 0.7392,
      "grad_norm": 0.3648279664943445,
      "learning_rate": 1.7671540671383245e-05,
      "loss": 0.0222,
      "step": 462
    },
    {
      "epoch": 0.7408,
      "grad_norm": 0.3650059458944533,
      "learning_rate": 1.766044443118978e-05,
      "loss": 0.0178,
      "step": 463
    },
    {
      "epoch": 0.7424,
      "grad_norm": 0.23141197197187555,
      "learning_rate": 1.764932531574648e-05,
      "loss": 0.0177,
      "step": 464
    },
    {
      "epoch": 0.744,
      "grad_norm": 0.24848492713296702,
      "learning_rate": 1.76381833582567e-05,
      "loss": 0.0152,
      "step": 465
    },
    {
      "epoch": 0.7456,
      "grad_norm": 0.212058059906063,
      "learning_rate": 1.762701859199202e-05,
      "loss": 0.016,
      "step": 466
    },
    {
      "epoch": 0.7472,
      "grad_norm": 0.22424171520312197,
      "learning_rate": 1.761583105029213e-05,
      "loss": 0.0151,
      "step": 467
    },
    {
      "epoch": 0.7488,
      "grad_norm": 0.3352178009928058,
      "learning_rate": 1.7604620766564725e-05,
      "loss": 0.0196,
      "step": 468
    },
    {
      "epoch": 0.7504,
      "grad_norm": 0.26322572301360064,
      "learning_rate": 1.7593387774285412e-05,
      "loss": 0.015,
      "step": 469
    },
    {
      "epoch": 0.752,
      "grad_norm": 0.2814098104445483,
      "learning_rate": 1.7582132106997615e-05,
      "loss": 0.0184,
      "step": 470
    },
    {
      "epoch": 0.7536,
      "grad_norm": 0.307432575270684,
      "learning_rate": 1.7570853798312462e-05,
      "loss": 0.0164,
      "step": 471
    },
    {
      "epoch": 0.7552,
      "grad_norm": 0.2966570504116292,
      "learning_rate": 1.7559552881908698e-05,
      "loss": 0.0167,
      "step": 472
    },
    {
      "epoch": 0.7568,
      "grad_norm": 0.4866919307321116,
      "learning_rate": 1.7548229391532572e-05,
      "loss": 0.0201,
      "step": 473
    },
    {
      "epoch": 0.7584,
      "grad_norm": 0.2717271876657307,
      "learning_rate": 1.7536883360997743e-05,
      "loss": 0.0149,
      "step": 474
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.23215274345701958,
      "learning_rate": 1.7525514824185187e-05,
      "loss": 0.0157,
      "step": 475
    },
    {
      "epoch": 0.7616,
      "grad_norm": 0.2733866519645903,
      "learning_rate": 1.7514123815043073e-05,
      "loss": 0.0155,
      "step": 476
    },
    {
      "epoch": 0.7632,
      "grad_norm": 0.26453774848889056,
      "learning_rate": 1.750271036758669e-05,
      "loss": 0.0174,
      "step": 477
    },
    {
      "epoch": 0.7648,
      "grad_norm": 0.23059650092556158,
      "learning_rate": 1.749127451589832e-05,
      "loss": 0.0151,
      "step": 478
    },
    {
      "epoch": 0.7664,
      "grad_norm": 0.3304249529546193,
      "learning_rate": 1.747981629412715e-05,
      "loss": 0.0174,
      "step": 479
    },
    {
      "epoch": 0.768,
      "grad_norm": 0.2270056832127085,
      "learning_rate": 1.7468335736489177e-05,
      "loss": 0.0163,
      "step": 480
    },
    {
      "epoch": 0.7696,
      "grad_norm": 0.29069700663640613,
      "learning_rate": 1.7456832877267083e-05,
      "loss": 0.0158,
      "step": 481
    },
    {
      "epoch": 0.7712,
      "grad_norm": 0.3752070968464264,
      "learning_rate": 1.7445307750810153e-05,
      "loss": 0.0159,
      "step": 482
    },
    {
      "epoch": 0.7728,
      "grad_norm": 0.23327504768427246,
      "learning_rate": 1.7433760391534166e-05,
      "loss": 0.0167,
      "step": 483
    },
    {
      "epoch": 0.7744,
      "grad_norm": 0.2239594613519126,
      "learning_rate": 1.7422190833921284e-05,
      "loss": 0.0156,
      "step": 484
    },
    {
      "epoch": 0.776,
      "grad_norm": 0.1777337204794238,
      "learning_rate": 1.741059911251997e-05,
      "loss": 0.0147,
      "step": 485
    },
    {
      "epoch": 0.7776,
      "grad_norm": 0.30193231609851146,
      "learning_rate": 1.7398985261944857e-05,
      "loss": 0.0155,
      "step": 486
    },
    {
      "epoch": 0.7792,
      "grad_norm": 0.3705440433007421,
      "learning_rate": 1.7387349316876668e-05,
      "loss": 0.016,
      "step": 487
    },
    {
      "epoch": 0.7808,
      "grad_norm": 0.3036805749121437,
      "learning_rate": 1.7375691312062102e-05,
      "loss": 0.0159,
      "step": 488
    },
    {
      "epoch": 0.7824,
      "grad_norm": 0.22418435070009263,
      "learning_rate": 1.7364011282313732e-05,
      "loss": 0.0151,
      "step": 489
    },
    {
      "epoch": 0.784,
      "grad_norm": 0.2447819237603347,
      "learning_rate": 1.7352309262509894e-05,
      "loss": 0.0157,
      "step": 490
    },
    {
      "epoch": 0.7856,
      "grad_norm": 0.2563576164383423,
      "learning_rate": 1.7340585287594605e-05,
      "loss": 0.0167,
      "step": 491
    },
    {
      "epoch": 0.7872,
      "grad_norm": 0.25731657995972784,
      "learning_rate": 1.7328839392577422e-05,
      "loss": 0.0181,
      "step": 492
    },
    {
      "epoch": 0.7888,
      "grad_norm": 0.33470461643316934,
      "learning_rate": 1.731707161253338e-05,
      "loss": 0.0154,
      "step": 493
    },
    {
      "epoch": 0.7904,
      "grad_norm": 0.2771412691165488,
      "learning_rate": 1.730528198260285e-05,
      "loss": 0.0145,
      "step": 494
    },
    {
      "epoch": 0.792,
      "grad_norm": 0.2709908554160519,
      "learning_rate": 1.7293470537991463e-05,
      "loss": 0.0179,
      "step": 495
    },
    {
      "epoch": 0.7936,
      "grad_norm": 0.2558694357144735,
      "learning_rate": 1.728163731396998e-05,
      "loss": 0.0138,
      "step": 496
    },
    {
      "epoch": 0.7952,
      "grad_norm": 0.25559177407084555,
      "learning_rate": 1.7269782345874204e-05,
      "loss": 0.021,
      "step": 497
    },
    {
      "epoch": 0.7968,
      "grad_norm": 0.33197746334294914,
      "learning_rate": 1.7257905669104874e-05,
      "loss": 0.0163,
      "step": 498
    },
    {
      "epoch": 0.7984,
      "grad_norm": 0.24632128756768099,
      "learning_rate": 1.7246007319127547e-05,
      "loss": 0.0144,
      "step": 499
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.32033168801101264,
      "learning_rate": 1.72340873314725e-05,
      "loss": 0.0165,
      "step": 500
    },
    {
      "epoch": 0.8016,
      "grad_norm": 0.3268250789735254,
      "learning_rate": 1.7222145741734625e-05,
      "loss": 0.0164,
      "step": 501
    },
    {
      "epoch": 0.8032,
      "grad_norm": 0.2232688071161065,
      "learning_rate": 1.721018258557333e-05,
      "loss": 0.0153,
      "step": 502
    },
    {
      "epoch": 0.8048,
      "grad_norm": 0.4371826628599994,
      "learning_rate": 1.7198197898712402e-05,
      "loss": 0.0158,
      "step": 503
    },
    {
      "epoch": 0.8064,
      "grad_norm": 0.272064986688738,
      "learning_rate": 1.7186191716939946e-05,
      "loss": 0.016,
      "step": 504
    },
    {
      "epoch": 0.808,
      "grad_norm": 0.2999734487686739,
      "learning_rate": 1.717416407610824e-05,
      "loss": 0.0165,
      "step": 505
    },
    {
      "epoch": 0.8096,
      "grad_norm": 0.20381910185537508,
      "learning_rate": 1.7162115012133643e-05,
      "loss": 0.0122,
      "step": 506
    },
    {
      "epoch": 0.8112,
      "grad_norm": 0.2514599572754894,
      "learning_rate": 1.7150044560996488e-05,
      "loss": 0.0165,
      "step": 507
    },
    {
      "epoch": 0.8128,
      "grad_norm": 0.24878584344864693,
      "learning_rate": 1.713795275874098e-05,
      "loss": 0.0141,
      "step": 508
    },
    {
      "epoch": 0.8144,
      "grad_norm": 0.2495924620006895,
      "learning_rate": 1.7125839641475074e-05,
      "loss": 0.0136,
      "step": 509
    },
    {
      "epoch": 0.816,
      "grad_norm": 0.2983231959518722,
      "learning_rate": 1.711370524537037e-05,
      "loss": 0.0153,
      "step": 510
    },
    {
      "epoch": 0.8176,
      "grad_norm": 0.32363209088505135,
      "learning_rate": 1.7101549606662025e-05,
      "loss": 0.0162,
      "step": 511
    },
    {
      "epoch": 0.8192,
      "grad_norm": 0.23024824037626007,
      "learning_rate": 1.7089372761648617e-05,
      "loss": 0.0129,
      "step": 512
    },
    {
      "epoch": 0.8208,
      "grad_norm": 0.3099805526735493,
      "learning_rate": 1.7077174746692054e-05,
      "loss": 0.0146,
      "step": 513
    },
    {
      "epoch": 0.8224,
      "grad_norm": 0.2958790504323679,
      "learning_rate": 1.7064955598217463e-05,
      "loss": 0.0144,
      "step": 514
    },
    {
      "epoch": 0.824,
      "grad_norm": 0.32180951645652084,
      "learning_rate": 1.7052715352713076e-05,
      "loss": 0.0124,
      "step": 515
    },
    {
      "epoch": 0.8256,
      "grad_norm": 0.3161408048838766,
      "learning_rate": 1.7040454046730118e-05,
      "loss": 0.0122,
      "step": 516
    },
    {
      "epoch": 0.8272,
      "grad_norm": 0.4359972092447916,
      "learning_rate": 1.7028171716882714e-05,
      "loss": 0.016,
      "step": 517
    },
    {
      "epoch": 0.8288,
      "grad_norm": 0.6070067239235338,
      "learning_rate": 1.7015868399847768e-05,
      "loss": 0.0157,
      "step": 518
    },
    {
      "epoch": 0.8304,
      "grad_norm": 0.30921833575080176,
      "learning_rate": 1.7003544132364847e-05,
      "loss": 0.0175,
      "step": 519
    },
    {
      "epoch": 0.832,
      "grad_norm": 0.24855531102712775,
      "learning_rate": 1.6991198951236088e-05,
      "loss": 0.0135,
      "step": 520
    },
    {
      "epoch": 0.8336,
      "grad_norm": 0.3990339989018151,
      "learning_rate": 1.6978832893326074e-05,
      "loss": 0.0156,
      "step": 521
    },
    {
      "epoch": 0.8352,
      "grad_norm": 0.4165012367115773,
      "learning_rate": 1.696644599556173e-05,
      "loss": 0.016,
      "step": 522
    },
    {
      "epoch": 0.8368,
      "grad_norm": 0.43412052801627804,
      "learning_rate": 1.6954038294932215e-05,
      "loss": 0.0201,
      "step": 523
    },
    {
      "epoch": 0.8384,
      "grad_norm": 0.47790373526083657,
      "learning_rate": 1.6941609828488806e-05,
      "loss": 0.0125,
      "step": 524
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2970205406935243,
      "learning_rate": 1.692916063334479e-05,
      "loss": 0.0138,
      "step": 525
    },
    {
      "epoch": 0.8416,
      "grad_norm": 0.4618090240290649,
      "learning_rate": 1.691669074667535e-05,
      "loss": 0.0188,
      "step": 526
    },
    {
      "epoch": 0.8432,
      "grad_norm": 0.34029536024377827,
      "learning_rate": 1.690420020571747e-05,
      "loss": 0.018,
      "step": 527
    },
    {
      "epoch": 0.8448,
      "grad_norm": 0.3439089692946025,
      "learning_rate": 1.689168904776979e-05,
      "loss": 0.017,
      "step": 528
    },
    {
      "epoch": 0.8464,
      "grad_norm": 0.30043383316215877,
      "learning_rate": 1.6879157310192537e-05,
      "loss": 0.0179,
      "step": 529
    },
    {
      "epoch": 0.848,
      "grad_norm": 0.4054893693102873,
      "learning_rate": 1.686660503040737e-05,
      "loss": 0.0159,
      "step": 530
    },
    {
      "epoch": 0.8496,
      "grad_norm": 0.2557418982826781,
      "learning_rate": 1.685403224589731e-05,
      "loss": 0.0159,
      "step": 531
    },
    {
      "epoch": 0.8512,
      "grad_norm": 0.3221299446889479,
      "learning_rate": 1.6841438994206597e-05,
      "loss": 0.0139,
      "step": 532
    },
    {
      "epoch": 0.8528,
      "grad_norm": 0.20671619932779295,
      "learning_rate": 1.6828825312940594e-05,
      "loss": 0.0133,
      "step": 533
    },
    {
      "epoch": 0.8544,
      "grad_norm": 0.395621658771693,
      "learning_rate": 1.6816191239765668e-05,
      "loss": 0.015,
      "step": 534
    },
    {
      "epoch": 0.856,
      "grad_norm": 0.18571913914672067,
      "learning_rate": 1.6803536812409077e-05,
      "loss": 0.0133,
      "step": 535
    },
    {
      "epoch": 0.8576,
      "grad_norm": 0.4219460617356573,
      "learning_rate": 1.6790862068658863e-05,
      "loss": 0.0167,
      "step": 536
    },
    {
      "epoch": 0.8592,
      "grad_norm": 0.4313099517984012,
      "learning_rate": 1.6778167046363735e-05,
      "loss": 0.016,
      "step": 537
    },
    {
      "epoch": 0.8608,
      "grad_norm": 0.24685981400311702,
      "learning_rate": 1.6765451783432953e-05,
      "loss": 0.0129,
      "step": 538
    },
    {
      "epoch": 0.8624,
      "grad_norm": 0.36926713742823264,
      "learning_rate": 1.675271631783623e-05,
      "loss": 0.0176,
      "step": 539
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.27147011307048174,
      "learning_rate": 1.6739960687603592e-05,
      "loss": 0.0193,
      "step": 540
    },
    {
      "epoch": 0.8656,
      "grad_norm": 0.30418888127452126,
      "learning_rate": 1.672718493082529e-05,
      "loss": 0.0187,
      "step": 541
    },
    {
      "epoch": 0.8672,
      "grad_norm": 0.31714375660907473,
      "learning_rate": 1.671438908565167e-05,
      "loss": 0.0139,
      "step": 542
    },
    {
      "epoch": 0.8688,
      "grad_norm": 0.44713073525701746,
      "learning_rate": 1.6701573190293076e-05,
      "loss": 0.0154,
      "step": 543
    },
    {
      "epoch": 0.8704,
      "grad_norm": 0.32977461143617176,
      "learning_rate": 1.6688737283019708e-05,
      "loss": 0.0172,
      "step": 544
    },
    {
      "epoch": 0.872,
      "grad_norm": 0.22999075617827208,
      "learning_rate": 1.667588140216154e-05,
      "loss": 0.0169,
      "step": 545
    },
    {
      "epoch": 0.8736,
      "grad_norm": 0.23997918810025493,
      "learning_rate": 1.6663005586108175e-05,
      "loss": 0.0166,
      "step": 546
    },
    {
      "epoch": 0.8752,
      "grad_norm": 0.43197070145474226,
      "learning_rate": 1.6650109873308763e-05,
      "loss": 0.0156,
      "step": 547
    },
    {
      "epoch": 0.8768,
      "grad_norm": 0.4741914350797042,
      "learning_rate": 1.663719430227186e-05,
      "loss": 0.0191,
      "step": 548
    },
    {
      "epoch": 0.8784,
      "grad_norm": 0.43115846393700075,
      "learning_rate": 1.6624258911565312e-05,
      "loss": 0.0157,
      "step": 549
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2965223549899331,
      "learning_rate": 1.661130373981617e-05,
      "loss": 0.0116,
      "step": 550
    },
    {
      "epoch": 0.8816,
      "grad_norm": 0.24489453090945398,
      "learning_rate": 1.6598328825710536e-05,
      "loss": 0.0141,
      "step": 551
    },
    {
      "epoch": 0.8832,
      "grad_norm": 0.28538146507727113,
      "learning_rate": 1.6585334207993475e-05,
      "loss": 0.0139,
      "step": 552
    },
    {
      "epoch": 0.8848,
      "grad_norm": 0.35788977059608296,
      "learning_rate": 1.6572319925468892e-05,
      "loss": 0.0147,
      "step": 553
    },
    {
      "epoch": 0.8864,
      "grad_norm": 0.28960712882823075,
      "learning_rate": 1.65592860169994e-05,
      "loss": 0.0147,
      "step": 554
    },
    {
      "epoch": 0.888,
      "grad_norm": 0.31950119423116574,
      "learning_rate": 1.654623252150624e-05,
      "loss": 0.0135,
      "step": 555
    },
    {
      "epoch": 0.8896,
      "grad_norm": 0.3807738816010361,
      "learning_rate": 1.6533159477969122e-05,
      "loss": 0.0133,
      "step": 556
    },
    {
      "epoch": 0.8912,
      "grad_norm": 0.31788894397034834,
      "learning_rate": 1.6520066925426146e-05,
      "loss": 0.0163,
      "step": 557
    },
    {
      "epoch": 0.8928,
      "grad_norm": 0.29489370953447575,
      "learning_rate": 1.6506954902973657e-05,
      "loss": 0.0118,
      "step": 558
    },
    {
      "epoch": 0.8944,
      "grad_norm": 0.32915478563644357,
      "learning_rate": 1.6493823449766137e-05,
      "loss": 0.0145,
      "step": 559
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.2496157180299605,
      "learning_rate": 1.648067260501611e-05,
      "loss": 0.0105,
      "step": 560
    },
    {
      "epoch": 0.8976,
      "grad_norm": 0.23022392976252964,
      "learning_rate": 1.6467502407993995e-05,
      "loss": 0.0153,
      "step": 561
    },
    {
      "epoch": 0.8992,
      "grad_norm": 0.2741504250149933,
      "learning_rate": 1.6454312898027992e-05,
      "loss": 0.0129,
      "step": 562
    },
    {
      "epoch": 0.9008,
      "grad_norm": 0.37673992433131875,
      "learning_rate": 1.644110411450398e-05,
      "loss": 0.0145,
      "step": 563
    },
    {
      "epoch": 0.9024,
      "grad_norm": 0.24683345688818692,
      "learning_rate": 1.6427876096865394e-05,
      "loss": 0.0132,
      "step": 564
    },
    {
      "epoch": 0.904,
      "grad_norm": 0.21155010306282332,
      "learning_rate": 1.6414628884613106e-05,
      "loss": 0.0124,
      "step": 565
    },
    {
      "epoch": 0.9056,
      "grad_norm": 0.2282821732438531,
      "learning_rate": 1.6401362517305296e-05,
      "loss": 0.0121,
      "step": 566
    },
    {
      "epoch": 0.9072,
      "grad_norm": 0.5087710551373635,
      "learning_rate": 1.6388077034557355e-05,
      "loss": 0.0172,
      "step": 567
    },
    {
      "epoch": 0.9088,
      "grad_norm": 0.38875972717524937,
      "learning_rate": 1.637477247604175e-05,
      "loss": 0.0143,
      "step": 568
    },
    {
      "epoch": 0.9104,
      "grad_norm": 0.33977547058849117,
      "learning_rate": 1.6361448881487913e-05,
      "loss": 0.0144,
      "step": 569
    },
    {
      "epoch": 0.912,
      "grad_norm": 0.39719043920205616,
      "learning_rate": 1.6348106290682117e-05,
      "loss": 0.0135,
      "step": 570
    },
    {
      "epoch": 0.9136,
      "grad_norm": 0.41966092792866705,
      "learning_rate": 1.6334744743467366e-05,
      "loss": 0.0148,
      "step": 571
    },
    {
      "epoch": 0.9152,
      "grad_norm": 0.5310320819623283,
      "learning_rate": 1.6321364279743267e-05,
      "loss": 0.0148,
      "step": 572
    },
    {
      "epoch": 0.9168,
      "grad_norm": 0.36975329288920533,
      "learning_rate": 1.6307964939465914e-05,
      "loss": 0.0142,
      "step": 573
    },
    {
      "epoch": 0.9184,
      "grad_norm": 0.24157258569983317,
      "learning_rate": 1.6294546762647775e-05,
      "loss": 0.0091,
      "step": 574
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4828289939576407,
      "learning_rate": 1.628110978935756e-05,
      "loss": 0.0141,
      "step": 575
    },
    {
      "epoch": 0.9216,
      "grad_norm": 0.4861418486779714,
      "learning_rate": 1.626765405972011e-05,
      "loss": 0.0159,
      "step": 576
    },
    {
      "epoch": 0.9232,
      "grad_norm": 0.3805881534396026,
      "learning_rate": 1.625417961391628e-05,
      "loss": 0.0143,
      "step": 577
    },
    {
      "epoch": 0.9248,
      "grad_norm": 0.2824880755489354,
      "learning_rate": 1.6240686492182806e-05,
      "loss": 0.0113,
      "step": 578
    },
    {
      "epoch": 0.9264,
      "grad_norm": 0.47408413957904083,
      "learning_rate": 1.62271747348122e-05,
      "loss": 0.0168,
      "step": 579
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.5431515431501727,
      "learning_rate": 1.621364438215262e-05,
      "loss": 0.0177,
      "step": 580
    },
    {
      "epoch": 0.9296,
      "grad_norm": 0.4281155633808909,
      "learning_rate": 1.6200095474607753e-05,
      "loss": 0.0137,
      "step": 581
    },
    {
      "epoch": 0.9312,
      "grad_norm": 0.2893495598259243,
      "learning_rate": 1.6186528052636692e-05,
      "loss": 0.0126,
      "step": 582
    },
    {
      "epoch": 0.9328,
      "grad_norm": 0.31186922217219265,
      "learning_rate": 1.6172942156753822e-05,
      "loss": 0.0138,
      "step": 583
    },
    {
      "epoch": 0.9344,
      "grad_norm": 0.16779597630347912,
      "learning_rate": 1.6159337827528686e-05,
      "loss": 0.0111,
      "step": 584
    },
    {
      "epoch": 0.936,
      "grad_norm": 0.47723171668240166,
      "learning_rate": 1.614571510558588e-05,
      "loss": 0.0156,
      "step": 585
    },
    {
      "epoch": 0.9376,
      "grad_norm": 0.4819625853916828,
      "learning_rate": 1.6132074031604917e-05,
      "loss": 0.0142,
      "step": 586
    },
    {
      "epoch": 0.9392,
      "grad_norm": 0.4558378441295942,
      "learning_rate": 1.6118414646320115e-05,
      "loss": 0.0149,
      "step": 587
    },
    {
      "epoch": 0.9408,
      "grad_norm": 0.3157472452703629,
      "learning_rate": 1.6104736990520468e-05,
      "loss": 0.0116,
      "step": 588
    },
    {
      "epoch": 0.9424,
      "grad_norm": 0.1989816586608992,
      "learning_rate": 1.6091041105049542e-05,
      "loss": 0.0112,
      "step": 589
    },
    {
      "epoch": 0.944,
      "grad_norm": 0.34300136848839013,
      "learning_rate": 1.6077327030805318e-05,
      "loss": 0.0129,
      "step": 590
    },
    {
      "epoch": 0.9456,
      "grad_norm": 0.1805561810136135,
      "learning_rate": 1.6063594808740112e-05,
      "loss": 0.0108,
      "step": 591
    },
    {
      "epoch": 0.9472,
      "grad_norm": 0.18712891674307838,
      "learning_rate": 1.604984447986042e-05,
      "loss": 0.0135,
      "step": 592
    },
    {
      "epoch": 0.9488,
      "grad_norm": 0.3067750461036351,
      "learning_rate": 1.6036076085226813e-05,
      "loss": 0.0133,
      "step": 593
    },
    {
      "epoch": 0.9504,
      "grad_norm": 0.1812977796323941,
      "learning_rate": 1.602228966595381e-05,
      "loss": 0.0099,
      "step": 594
    },
    {
      "epoch": 0.952,
      "grad_norm": 0.2480339618306651,
      "learning_rate": 1.6008485263209742e-05,
      "loss": 0.0102,
      "step": 595
    },
    {
      "epoch": 0.9536,
      "grad_norm": 0.24450667954263358,
      "learning_rate": 1.599466291821666e-05,
      "loss": 0.0134,
      "step": 596
    },
    {
      "epoch": 0.9552,
      "grad_norm": 0.29278728917338376,
      "learning_rate": 1.598082267225018e-05,
      "loss": 0.0125,
      "step": 597
    },
    {
      "epoch": 0.9568,
      "grad_norm": 0.3086905154384183,
      "learning_rate": 1.596696456663938e-05,
      "loss": 0.0121,
      "step": 598
    },
    {
      "epoch": 0.9584,
      "grad_norm": 0.2766246945781701,
      "learning_rate": 1.595308864276666e-05,
      "loss": 0.0107,
      "step": 599
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23271296887264653,
      "learning_rate": 1.5939194942067647e-05,
      "loss": 0.0118,
      "step": 600
    },
    {
      "epoch": 0.9616,
      "grad_norm": 0.21834384775961552,
      "learning_rate": 1.592528350603103e-05,
      "loss": 0.0095,
      "step": 601
    },
    {
      "epoch": 0.9632,
      "grad_norm": 0.31082618893772185,
      "learning_rate": 1.5911354376198468e-05,
      "loss": 0.0139,
      "step": 602
    },
    {
      "epoch": 0.9648,
      "grad_norm": 0.3716195934094933,
      "learning_rate": 1.5897407594164468e-05,
      "loss": 0.0121,
      "step": 603
    },
    {
      "epoch": 0.9664,
      "grad_norm": 0.46949516504477684,
      "learning_rate": 1.5883443201576225e-05,
      "loss": 0.0122,
      "step": 604
    },
    {
      "epoch": 0.968,
      "grad_norm": 0.35059424145377427,
      "learning_rate": 1.586946124013354e-05,
      "loss": 0.0116,
      "step": 605
    },
    {
      "epoch": 0.9696,
      "grad_norm": 0.40688113982199503,
      "learning_rate": 1.585546175158868e-05,
      "loss": 0.014,
      "step": 606
    },
    {
      "epoch": 0.9712,
      "grad_norm": 0.4156616285372754,
      "learning_rate": 1.5841444777746232e-05,
      "loss": 0.013,
      "step": 607
    },
    {
      "epoch": 0.9728,
      "grad_norm": 0.21046745034986,
      "learning_rate": 1.582741036046301e-05,
      "loss": 0.0088,
      "step": 608
    },
    {
      "epoch": 0.9744,
      "grad_norm": 0.36941300422007295,
      "learning_rate": 1.5813358541647915e-05,
      "loss": 0.0106,
      "step": 609
    },
    {
      "epoch": 0.976,
      "grad_norm": 0.24297535036589946,
      "learning_rate": 1.5799289363261815e-05,
      "loss": 0.0115,
      "step": 610
    },
    {
      "epoch": 0.9776,
      "grad_norm": 0.303837166125424,
      "learning_rate": 1.578520286731741e-05,
      "loss": 0.0106,
      "step": 611
    },
    {
      "epoch": 0.9792,
      "grad_norm": 0.2803480753538006,
      "learning_rate": 1.5771099095879108e-05,
      "loss": 0.0096,
      "step": 612
    },
    {
      "epoch": 0.9808,
      "grad_norm": 0.31074041433807803,
      "learning_rate": 1.575697809106292e-05,
      "loss": 0.0118,
      "step": 613
    },
    {
      "epoch": 0.9824,
      "grad_norm": 0.49870271318985004,
      "learning_rate": 1.5742839895036305e-05,
      "loss": 0.0165,
      "step": 614
    },
    {
      "epoch": 0.984,
      "grad_norm": 0.3044978450333263,
      "learning_rate": 1.5728684550018066e-05,
      "loss": 0.0126,
      "step": 615
    },
    {
      "epoch": 0.9856,
      "grad_norm": 0.3493681227793422,
      "learning_rate": 1.571451209827821e-05,
      "loss": 0.014,
      "step": 616
    },
    {
      "epoch": 0.9872,
      "grad_norm": 0.37956769866427376,
      "learning_rate": 1.570032258213783e-05,
      "loss": 0.0134,
      "step": 617
    },
    {
      "epoch": 0.9888,
      "grad_norm": 0.37799510952249216,
      "learning_rate": 1.5686116043968975e-05,
      "loss": 0.0111,
      "step": 618
    },
    {
      "epoch": 0.9904,
      "grad_norm": 0.39548477772844104,
      "learning_rate": 1.5671892526194515e-05,
      "loss": 0.0153,
      "step": 619
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.4257260663819077,
      "learning_rate": 1.565765207128805e-05,
      "loss": 0.0131,
      "step": 620
    },
    {
      "epoch": 0.9936,
      "grad_norm": 0.30794139713595914,
      "learning_rate": 1.564339472177373e-05,
      "loss": 0.0117,
      "step": 621
    },
    {
      "epoch": 0.9952,
      "grad_norm": 0.278719510483356,
      "learning_rate": 1.5629120520226163e-05,
      "loss": 0.0119,
      "step": 622
    },
    {
      "epoch": 0.9968,
      "grad_norm": 0.297554543587557,
      "learning_rate": 1.561482950927029e-05,
      "loss": 0.0112,
      "step": 623
    },
    {
      "epoch": 0.9984,
      "grad_norm": 0.24695394521400738,
      "learning_rate": 1.560052173158123e-05,
      "loss": 0.0125,
      "step": 624
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.40345363184405636,
      "learning_rate": 1.5586197229884185e-05,
      "loss": 0.0133,
      "step": 625
    },
    {
      "epoch": 1.0016,
      "grad_norm": 0.417857034308898,
      "learning_rate": 1.5571856046954284e-05,
      "loss": 0.0114,
      "step": 626
    },
    {
      "epoch": 1.0032,
      "grad_norm": 0.21833831910708582,
      "learning_rate": 1.5557498225616488e-05,
      "loss": 0.0125,
      "step": 627
    },
    {
      "epoch": 1.0048,
      "grad_norm": 0.2743191504104528,
      "learning_rate": 1.5543123808745418e-05,
      "loss": 0.0132,
      "step": 628
    },
    {
      "epoch": 1.0064,
      "grad_norm": 0.30618341671798766,
      "learning_rate": 1.5528732839265272e-05,
      "loss": 0.0124,
      "step": 629
    },
    {
      "epoch": 1.008,
      "grad_norm": 0.3689469002067123,
      "learning_rate": 1.5514325360149668e-05,
      "loss": 0.0147,
      "step": 630
    },
    {
      "epoch": 1.0096,
      "grad_norm": 0.3088756615860901,
      "learning_rate": 1.549990141442153e-05,
      "loss": 0.0122,
      "step": 631
    },
    {
      "epoch": 1.0112,
      "grad_norm": 0.19294285505406303,
      "learning_rate": 1.5485461045152937e-05,
      "loss": 0.012,
      "step": 632
    },
    {
      "epoch": 1.0128,
      "grad_norm": 0.3001012276511453,
      "learning_rate": 1.5471004295465034e-05,
      "loss": 0.0132,
      "step": 633
    },
    {
      "epoch": 1.0144,
      "grad_norm": 0.1921528171891681,
      "learning_rate": 1.5456531208527868e-05,
      "loss": 0.0093,
      "step": 634
    },
    {
      "epoch": 1.016,
      "grad_norm": 0.16268529604637008,
      "learning_rate": 1.5442041827560274e-05,
      "loss": 0.009,
      "step": 635
    },
    {
      "epoch": 1.0176,
      "grad_norm": 0.30072233263249754,
      "learning_rate": 1.542753619582974e-05,
      "loss": 0.0101,
      "step": 636
    },
    {
      "epoch": 1.0192,
      "grad_norm": 0.25149533409567965,
      "learning_rate": 1.5413014356652287e-05,
      "loss": 0.0115,
      "step": 637
    },
    {
      "epoch": 1.0208,
      "grad_norm": 0.27926618346757304,
      "learning_rate": 1.5398476353392323e-05,
      "loss": 0.0128,
      "step": 638
    },
    {
      "epoch": 1.0224,
      "grad_norm": 0.23840478861268233,
      "learning_rate": 1.538392222946255e-05,
      "loss": 0.0125,
      "step": 639
    },
    {
      "epoch": 1.024,
      "grad_norm": 0.2748477006078572,
      "learning_rate": 1.5369352028323773e-05,
      "loss": 0.0114,
      "step": 640
    },
    {
      "epoch": 1.0256,
      "grad_norm": 0.36023776554838094,
      "learning_rate": 1.5354765793484834e-05,
      "loss": 0.0097,
      "step": 641
    },
    {
      "epoch": 1.0272,
      "grad_norm": 0.36739400395113647,
      "learning_rate": 1.534016356850244e-05,
      "loss": 0.0119,
      "step": 642
    },
    {
      "epoch": 1.0288,
      "grad_norm": 0.32947192475025666,
      "learning_rate": 1.5325545396981053e-05,
      "loss": 0.0087,
      "step": 643
    },
    {
      "epoch": 1.0304,
      "grad_norm": 0.4587397886694395,
      "learning_rate": 1.531091132257275e-05,
      "loss": 0.0119,
      "step": 644
    },
    {
      "epoch": 1.032,
      "grad_norm": 0.2852871073646235,
      "learning_rate": 1.5296261388977107e-05,
      "loss": 0.0102,
      "step": 645
    },
    {
      "epoch": 1.0336,
      "grad_norm": 0.44917912534081555,
      "learning_rate": 1.528159563994104e-05,
      "loss": 0.0111,
      "step": 646
    },
    {
      "epoch": 1.0352,
      "grad_norm": 0.47157149649612523,
      "learning_rate": 1.52669141192587e-05,
      "loss": 0.0139,
      "step": 647
    },
    {
      "epoch": 1.0368,
      "grad_norm": 0.3528186945026468,
      "learning_rate": 1.5252216870771345e-05,
      "loss": 0.0138,
      "step": 648
    },
    {
      "epoch": 1.0384,
      "grad_norm": 0.3986037607641987,
      "learning_rate": 1.5237503938367186e-05,
      "loss": 0.0144,
      "step": 649
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.5600923250750897,
      "learning_rate": 1.5222775365981272e-05,
      "loss": 0.013,
      "step": 650
    },
    {
      "epoch": 1.0416,
      "grad_norm": 0.421695944273333,
      "learning_rate": 1.5208031197595357e-05,
      "loss": 0.011,
      "step": 651
    },
    {
      "epoch": 1.0432,
      "grad_norm": 0.43277467796608626,
      "learning_rate": 1.5193271477237761e-05,
      "loss": 0.0139,
      "step": 652
    },
    {
      "epoch": 1.0448,
      "grad_norm": 0.33407654281757615,
      "learning_rate": 1.5178496248983254e-05,
      "loss": 0.0121,
      "step": 653
    },
    {
      "epoch": 1.0464,
      "grad_norm": 0.3645641843666089,
      "learning_rate": 1.5163705556952912e-05,
      "loss": 0.0107,
      "step": 654
    },
    {
      "epoch": 1.048,
      "grad_norm": 0.3621632764026968,
      "learning_rate": 1.5148899445313983e-05,
      "loss": 0.0142,
      "step": 655
    },
    {
      "epoch": 1.0496,
      "grad_norm": 0.32624268869617895,
      "learning_rate": 1.5134077958279764e-05,
      "loss": 0.0133,
      "step": 656
    },
    {
      "epoch": 1.0512,
      "grad_norm": 0.29284343589574013,
      "learning_rate": 1.5119241140109466e-05,
      "loss": 0.0095,
      "step": 657
    },
    {
      "epoch": 1.0528,
      "grad_norm": 0.41739711667469,
      "learning_rate": 1.5104389035108078e-05,
      "loss": 0.0121,
      "step": 658
    },
    {
      "epoch": 1.0544,
      "grad_norm": 0.32747130295385996,
      "learning_rate": 1.5089521687626243e-05,
      "loss": 0.0135,
      "step": 659
    },
    {
      "epoch": 1.056,
      "grad_norm": 0.3291067637296192,
      "learning_rate": 1.5074639142060119e-05,
      "loss": 0.0094,
      "step": 660
    },
    {
      "epoch": 1.0576,
      "grad_norm": 0.31460572785375496,
      "learning_rate": 1.505974144285124e-05,
      "loss": 0.0104,
      "step": 661
    },
    {
      "epoch": 1.0592,
      "grad_norm": 0.5394369508399102,
      "learning_rate": 1.50448286344864e-05,
      "loss": 0.0126,
      "step": 662
    },
    {
      "epoch": 1.0608,
      "grad_norm": 0.6173037121141223,
      "learning_rate": 1.5029900761497507e-05,
      "loss": 0.0158,
      "step": 663
    },
    {
      "epoch": 1.0624,
      "grad_norm": 0.3429474313068254,
      "learning_rate": 1.501495786846146e-05,
      "loss": 0.0134,
      "step": 664
    },
    {
      "epoch": 1.064,
      "grad_norm": 0.2944391342591949,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.0085,
      "step": 665
    },
    {
      "epoch": 1.0656,
      "grad_norm": 0.4287258380040374,
      "learning_rate": 1.4985027200779599e-05,
      "loss": 0.0144,
      "step": 666
    },
    {
      "epoch": 1.0672,
      "grad_norm": 0.3770767536307347,
      "learning_rate": 1.4970039515511303e-05,
      "loss": 0.0125,
      "step": 667
    },
    {
      "epoch": 1.0688,
      "grad_norm": 0.5615733955806762,
      "learning_rate": 1.4955036988950617e-05,
      "loss": 0.015,
      "step": 668
    },
    {
      "epoch": 1.0704,
      "grad_norm": 0.23511834985253832,
      "learning_rate": 1.4940019665897363e-05,
      "loss": 0.0106,
      "step": 669
    },
    {
      "epoch": 1.072,
      "grad_norm": 0.30497553069081007,
      "learning_rate": 1.4924987591195548e-05,
      "loss": 0.0113,
      "step": 670
    },
    {
      "epoch": 1.0735999999999999,
      "grad_norm": 0.35650652604214367,
      "learning_rate": 1.4909940809733223e-05,
      "loss": 0.0122,
      "step": 671
    },
    {
      "epoch": 1.0752,
      "grad_norm": 0.3308671401210887,
      "learning_rate": 1.489487936644237e-05,
      "loss": 0.0118,
      "step": 672
    },
    {
      "epoch": 1.0768,
      "grad_norm": 0.5047129564910963,
      "learning_rate": 1.4879803306298736e-05,
      "loss": 0.0105,
      "step": 673
    },
    {
      "epoch": 1.0784,
      "grad_norm": 0.2888826727326239,
      "learning_rate": 1.4864712674321733e-05,
      "loss": 0.0107,
      "step": 674
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.22439723192462555,
      "learning_rate": 1.4849607515574276e-05,
      "loss": 0.0097,
      "step": 675
    },
    {
      "epoch": 1.0816,
      "grad_norm": 0.351953190025847,
      "learning_rate": 1.4834487875162657e-05,
      "loss": 0.01,
      "step": 676
    },
    {
      "epoch": 1.0832,
      "grad_norm": 0.29334503002848533,
      "learning_rate": 1.4819353798236427e-05,
      "loss": 0.0097,
      "step": 677
    },
    {
      "epoch": 1.0848,
      "grad_norm": 0.35188988523350234,
      "learning_rate": 1.4804205329988226e-05,
      "loss": 0.0122,
      "step": 678
    },
    {
      "epoch": 1.0864,
      "grad_norm": 0.24505321145577785,
      "learning_rate": 1.4789042515653687e-05,
      "loss": 0.0129,
      "step": 679
    },
    {
      "epoch": 1.088,
      "grad_norm": 0.25008455950445413,
      "learning_rate": 1.477386540051127e-05,
      "loss": 0.0102,
      "step": 680
    },
    {
      "epoch": 1.0896,
      "grad_norm": 0.35576816544038325,
      "learning_rate": 1.4758674029882152e-05,
      "loss": 0.0116,
      "step": 681
    },
    {
      "epoch": 1.0912,
      "grad_norm": 0.33614223269913607,
      "learning_rate": 1.4743468449130065e-05,
      "loss": 0.0133,
      "step": 682
    },
    {
      "epoch": 1.0928,
      "grad_norm": 0.22442180749880328,
      "learning_rate": 1.4728248703661183e-05,
      "loss": 0.01,
      "step": 683
    },
    {
      "epoch": 1.0944,
      "grad_norm": 0.28219807599355523,
      "learning_rate": 1.4713014838923975e-05,
      "loss": 0.0117,
      "step": 684
    },
    {
      "epoch": 1.096,
      "grad_norm": 0.2650287871574999,
      "learning_rate": 1.4697766900409076e-05,
      "loss": 0.0088,
      "step": 685
    },
    {
      "epoch": 1.0976,
      "grad_norm": 0.2766787747844056,
      "learning_rate": 1.4682504933649144e-05,
      "loss": 0.0086,
      "step": 686
    },
    {
      "epoch": 1.0992,
      "grad_norm": 0.3620818365197564,
      "learning_rate": 1.466722898421873e-05,
      "loss": 0.0079,
      "step": 687
    },
    {
      "epoch": 1.1008,
      "grad_norm": 0.37691673416789456,
      "learning_rate": 1.4651939097734132e-05,
      "loss": 0.0144,
      "step": 688
    },
    {
      "epoch": 1.1024,
      "grad_norm": 0.29790859515349194,
      "learning_rate": 1.4636635319853274e-05,
      "loss": 0.0101,
      "step": 689
    },
    {
      "epoch": 1.104,
      "grad_norm": 0.25829554558104056,
      "learning_rate": 1.4621317696275563e-05,
      "loss": 0.0116,
      "step": 690
    },
    {
      "epoch": 1.1056,
      "grad_norm": 0.26790109240876864,
      "learning_rate": 1.4605986272741748e-05,
      "loss": 0.0094,
      "step": 691
    },
    {
      "epoch": 1.1072,
      "grad_norm": 0.42271209243570657,
      "learning_rate": 1.4590641095033786e-05,
      "loss": 0.0092,
      "step": 692
    },
    {
      "epoch": 1.1088,
      "grad_norm": 0.4486221873645635,
      "learning_rate": 1.4575282208974704e-05,
      "loss": 0.0143,
      "step": 693
    },
    {
      "epoch": 1.1104,
      "grad_norm": 0.214394965355476,
      "learning_rate": 1.4559909660428469e-05,
      "loss": 0.0089,
      "step": 694
    },
    {
      "epoch": 1.112,
      "grad_norm": 0.38287545940258694,
      "learning_rate": 1.4544523495299843e-05,
      "loss": 0.0122,
      "step": 695
    },
    {
      "epoch": 1.1136,
      "grad_norm": 0.4091777970517161,
      "learning_rate": 1.4529123759534253e-05,
      "loss": 0.0096,
      "step": 696
    },
    {
      "epoch": 1.1152,
      "grad_norm": 0.3551233469036031,
      "learning_rate": 1.4513710499117648e-05,
      "loss": 0.0106,
      "step": 697
    },
    {
      "epoch": 1.1168,
      "grad_norm": 0.35675441570055094,
      "learning_rate": 1.4498283760076362e-05,
      "loss": 0.008,
      "step": 698
    },
    {
      "epoch": 1.1184,
      "grad_norm": 0.4761855955476139,
      "learning_rate": 1.4482843588476976e-05,
      "loss": 0.0113,
      "step": 699
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.3129622710980147,
      "learning_rate": 1.4467390030426187e-05,
      "loss": 0.0096,
      "step": 700
    },
    {
      "epoch": 1.1216,
      "grad_norm": 0.36202464550526253,
      "learning_rate": 1.445192313207067e-05,
      "loss": 0.0084,
      "step": 701
    },
    {
      "epoch": 1.1232,
      "grad_norm": 0.2751302476380293,
      "learning_rate": 1.443644293959693e-05,
      "loss": 0.0089,
      "step": 702
    },
    {
      "epoch": 1.1248,
      "grad_norm": 0.44668265606662855,
      "learning_rate": 1.4420949499231172e-05,
      "loss": 0.0119,
      "step": 703
    },
    {
      "epoch": 1.1264,
      "grad_norm": 0.6115934261309857,
      "learning_rate": 1.4405442857239151e-05,
      "loss": 0.0126,
      "step": 704
    },
    {
      "epoch": 1.1280000000000001,
      "grad_norm": 0.3178399997249513,
      "learning_rate": 1.4389923059926064e-05,
      "loss": 0.012,
      "step": 705
    },
    {
      "epoch": 1.1296,
      "grad_norm": 0.22483155185595363,
      "learning_rate": 1.437439015363638e-05,
      "loss": 0.0108,
      "step": 706
    },
    {
      "epoch": 1.1312,
      "grad_norm": 0.27827436859396276,
      "learning_rate": 1.4358844184753713e-05,
      "loss": 0.0103,
      "step": 707
    },
    {
      "epoch": 1.1328,
      "grad_norm": 0.40643842085489884,
      "learning_rate": 1.4343285199700685e-05,
      "loss": 0.008,
      "step": 708
    },
    {
      "epoch": 1.1344,
      "grad_norm": 0.17923007058434895,
      "learning_rate": 1.432771324493879e-05,
      "loss": 0.0084,
      "step": 709
    },
    {
      "epoch": 1.1360000000000001,
      "grad_norm": 0.36990937482255815,
      "learning_rate": 1.4312128366968244e-05,
      "loss": 0.0112,
      "step": 710
    },
    {
      "epoch": 1.1376,
      "grad_norm": 0.30074565334559583,
      "learning_rate": 1.4296530612327864e-05,
      "loss": 0.0126,
      "step": 711
    },
    {
      "epoch": 1.1392,
      "grad_norm": 0.32382377558034564,
      "learning_rate": 1.428092002759491e-05,
      "loss": 0.01,
      "step": 712
    },
    {
      "epoch": 1.1408,
      "grad_norm": 0.3153553781926328,
      "learning_rate": 1.4265296659384956e-05,
      "loss": 0.0123,
      "step": 713
    },
    {
      "epoch": 1.1424,
      "grad_norm": 0.3566687628901846,
      "learning_rate": 1.4249660554351752e-05,
      "loss": 0.011,
      "step": 714
    },
    {
      "epoch": 1.144,
      "grad_norm": 0.29709278445318327,
      "learning_rate": 1.4234011759187084e-05,
      "loss": 0.0105,
      "step": 715
    },
    {
      "epoch": 1.1456,
      "grad_norm": 0.43818116357185727,
      "learning_rate": 1.4218350320620625e-05,
      "loss": 0.0105,
      "step": 716
    },
    {
      "epoch": 1.1472,
      "grad_norm": 0.34110667072697853,
      "learning_rate": 1.4202676285419811e-05,
      "loss": 0.0083,
      "step": 717
    },
    {
      "epoch": 1.1488,
      "grad_norm": 0.5110187407254071,
      "learning_rate": 1.4186989700389689e-05,
      "loss": 0.0099,
      "step": 718
    },
    {
      "epoch": 1.1504,
      "grad_norm": 0.6325184130355656,
      "learning_rate": 1.4171290612372781e-05,
      "loss": 0.0133,
      "step": 719
    },
    {
      "epoch": 1.152,
      "grad_norm": 0.225304624696556,
      "learning_rate": 1.4155579068248951e-05,
      "loss": 0.0106,
      "step": 720
    },
    {
      "epoch": 1.1536,
      "grad_norm": 0.2884387125345757,
      "learning_rate": 1.4139855114935253e-05,
      "loss": 0.0086,
      "step": 721
    },
    {
      "epoch": 1.1552,
      "grad_norm": 0.15131765398328514,
      "learning_rate": 1.4124118799385797e-05,
      "loss": 0.0095,
      "step": 722
    },
    {
      "epoch": 1.1568,
      "grad_norm": 0.2563227551597657,
      "learning_rate": 1.410837016859161e-05,
      "loss": 0.0082,
      "step": 723
    },
    {
      "epoch": 1.1584,
      "grad_norm": 0.43243462497917795,
      "learning_rate": 1.4092609269580498e-05,
      "loss": 0.0108,
      "step": 724
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.30652501002357746,
      "learning_rate": 1.4076836149416889e-05,
      "loss": 0.0129,
      "step": 725
    },
    {
      "epoch": 1.1616,
      "grad_norm": 0.2600559640710353,
      "learning_rate": 1.4061050855201723e-05,
      "loss": 0.0122,
      "step": 726
    },
    {
      "epoch": 1.1632,
      "grad_norm": 0.2751909063838169,
      "learning_rate": 1.4045253434072278e-05,
      "loss": 0.0095,
      "step": 727
    },
    {
      "epoch": 1.1648,
      "grad_norm": 0.2677495381514669,
      "learning_rate": 1.4029443933202059e-05,
      "loss": 0.0087,
      "step": 728
    },
    {
      "epoch": 1.1663999999999999,
      "grad_norm": 0.329153787441638,
      "learning_rate": 1.4013622399800628e-05,
      "loss": 0.0118,
      "step": 729
    },
    {
      "epoch": 1.168,
      "grad_norm": 0.3954741610104781,
      "learning_rate": 1.399778888111349e-05,
      "loss": 0.0096,
      "step": 730
    },
    {
      "epoch": 1.1696,
      "grad_norm": 0.39323399280494176,
      "learning_rate": 1.3981943424421932e-05,
      "loss": 0.0112,
      "step": 731
    },
    {
      "epoch": 1.1712,
      "grad_norm": 0.2878151028081222,
      "learning_rate": 1.3966086077042891e-05,
      "loss": 0.0083,
      "step": 732
    },
    {
      "epoch": 1.1728,
      "grad_norm": 0.505797224428351,
      "learning_rate": 1.3950216886328818e-05,
      "loss": 0.014,
      "step": 733
    },
    {
      "epoch": 1.1743999999999999,
      "grad_norm": 0.2790678628939786,
      "learning_rate": 1.3934335899667526e-05,
      "loss": 0.0107,
      "step": 734
    },
    {
      "epoch": 1.176,
      "grad_norm": 0.3277167621850828,
      "learning_rate": 1.3918443164482048e-05,
      "loss": 0.0159,
      "step": 735
    },
    {
      "epoch": 1.1776,
      "grad_norm": 0.600242811130678,
      "learning_rate": 1.3902538728230502e-05,
      "loss": 0.0106,
      "step": 736
    },
    {
      "epoch": 1.1792,
      "grad_norm": 0.391801850174501,
      "learning_rate": 1.3886622638405953e-05,
      "loss": 0.011,
      "step": 737
    },
    {
      "epoch": 1.1808,
      "grad_norm": 0.2984427536737696,
      "learning_rate": 1.387069494253626e-05,
      "loss": 0.0133,
      "step": 738
    },
    {
      "epoch": 1.1824,
      "grad_norm": 0.19798876151872777,
      "learning_rate": 1.3854755688183941e-05,
      "loss": 0.0108,
      "step": 739
    },
    {
      "epoch": 1.184,
      "grad_norm": 0.3120296494732397,
      "learning_rate": 1.3838804922946027e-05,
      "loss": 0.0099,
      "step": 740
    },
    {
      "epoch": 1.1856,
      "grad_norm": 0.5741221288191662,
      "learning_rate": 1.3822842694453923e-05,
      "loss": 0.0143,
      "step": 741
    },
    {
      "epoch": 1.1872,
      "grad_norm": 0.5326283032765481,
      "learning_rate": 1.380686905037327e-05,
      "loss": 0.0116,
      "step": 742
    },
    {
      "epoch": 1.1888,
      "grad_norm": 0.311440966690227,
      "learning_rate": 1.3790884038403796e-05,
      "loss": 0.0116,
      "step": 743
    },
    {
      "epoch": 1.1904,
      "grad_norm": 0.425289385624851,
      "learning_rate": 1.3774887706279165e-05,
      "loss": 0.0094,
      "step": 744
    },
    {
      "epoch": 1.192,
      "grad_norm": 0.26822941847179554,
      "learning_rate": 1.375888010176686e-05,
      "loss": 0.0133,
      "step": 745
    },
    {
      "epoch": 1.1936,
      "grad_norm": 0.23449670744636084,
      "learning_rate": 1.374286127266801e-05,
      "loss": 0.0081,
      "step": 746
    },
    {
      "epoch": 1.1952,
      "grad_norm": 0.24419781837141116,
      "learning_rate": 1.3726831266817278e-05,
      "loss": 0.0111,
      "step": 747
    },
    {
      "epoch": 1.1968,
      "grad_norm": 0.2503170694406675,
      "learning_rate": 1.3710790132082693e-05,
      "loss": 0.0098,
      "step": 748
    },
    {
      "epoch": 1.1984,
      "grad_norm": 0.2924145590049586,
      "learning_rate": 1.3694737916365517e-05,
      "loss": 0.0107,
      "step": 749
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.413991031494989,
      "learning_rate": 1.3678674667600102e-05,
      "loss": 0.0124,
      "step": 750
    },
    {
      "epoch": 1.2016,
      "grad_norm": 0.2641875819723108,
      "learning_rate": 1.3662600433753746e-05,
      "loss": 0.0095,
      "step": 751
    },
    {
      "epoch": 1.2032,
      "grad_norm": 0.2530973588426655,
      "learning_rate": 1.3646515262826551e-05,
      "loss": 0.0095,
      "step": 752
    },
    {
      "epoch": 1.2048,
      "grad_norm": 0.13083092986175077,
      "learning_rate": 1.3630419202851287e-05,
      "loss": 0.0068,
      "step": 753
    },
    {
      "epoch": 1.2064,
      "grad_norm": 0.2826431152828746,
      "learning_rate": 1.3614312301893222e-05,
      "loss": 0.0091,
      "step": 754
    },
    {
      "epoch": 1.208,
      "grad_norm": 0.4859207050471821,
      "learning_rate": 1.3598194608050011e-05,
      "loss": 0.0109,
      "step": 755
    },
    {
      "epoch": 1.2096,
      "grad_norm": 0.3528244903100441,
      "learning_rate": 1.3582066169451535e-05,
      "loss": 0.0081,
      "step": 756
    },
    {
      "epoch": 1.2112,
      "grad_norm": 0.23735550136193914,
      "learning_rate": 1.3565927034259757e-05,
      "loss": 0.0072,
      "step": 757
    },
    {
      "epoch": 1.2128,
      "grad_norm": 0.26982430126630624,
      "learning_rate": 1.354977725066859e-05,
      "loss": 0.013,
      "step": 758
    },
    {
      "epoch": 1.2144,
      "grad_norm": 0.5291448926408648,
      "learning_rate": 1.3533616866903736e-05,
      "loss": 0.0127,
      "step": 759
    },
    {
      "epoch": 1.216,
      "grad_norm": 0.28414696636536013,
      "learning_rate": 1.351744593122255e-05,
      "loss": 0.012,
      "step": 760
    },
    {
      "epoch": 1.2176,
      "grad_norm": 0.4694574939461318,
      "learning_rate": 1.3501264491913909e-05,
      "loss": 0.0121,
      "step": 761
    },
    {
      "epoch": 1.2192,
      "grad_norm": 0.35695596199339735,
      "learning_rate": 1.3485072597298038e-05,
      "loss": 0.0085,
      "step": 762
    },
    {
      "epoch": 1.2208,
      "grad_norm": 0.19716908170889477,
      "learning_rate": 1.3468870295726399e-05,
      "loss": 0.0075,
      "step": 763
    },
    {
      "epoch": 1.2224,
      "grad_norm": 0.4402510192967036,
      "learning_rate": 1.3452657635581521e-05,
      "loss": 0.0097,
      "step": 764
    },
    {
      "epoch": 1.224,
      "grad_norm": 0.4574247653241456,
      "learning_rate": 1.3436434665276865e-05,
      "loss": 0.0097,
      "step": 765
    },
    {
      "epoch": 1.2256,
      "grad_norm": 0.31040237291608896,
      "learning_rate": 1.342020143325669e-05,
      "loss": 0.011,
      "step": 766
    },
    {
      "epoch": 1.2272,
      "grad_norm": 0.310907702558819,
      "learning_rate": 1.3403957987995884e-05,
      "loss": 0.0095,
      "step": 767
    },
    {
      "epoch": 1.2288000000000001,
      "grad_norm": 0.3833100006565884,
      "learning_rate": 1.3387704377999842e-05,
      "loss": 0.0113,
      "step": 768
    },
    {
      "epoch": 1.2304,
      "grad_norm": 0.4376843592039469,
      "learning_rate": 1.3371440651804313e-05,
      "loss": 0.0105,
      "step": 769
    },
    {
      "epoch": 1.232,
      "grad_norm": 0.3461601821191171,
      "learning_rate": 1.335516685797525e-05,
      "loss": 0.0105,
      "step": 770
    },
    {
      "epoch": 1.2336,
      "grad_norm": 0.4206013244953963,
      "learning_rate": 1.3338883045108674e-05,
      "loss": 0.01,
      "step": 771
    },
    {
      "epoch": 1.2352,
      "grad_norm": 0.3969044404932267,
      "learning_rate": 1.3322589261830517e-05,
      "loss": 0.0078,
      "step": 772
    },
    {
      "epoch": 1.2368000000000001,
      "grad_norm": 0.29849669127198464,
      "learning_rate": 1.3306285556796494e-05,
      "loss": 0.0091,
      "step": 773
    },
    {
      "epoch": 1.2384,
      "grad_norm": 0.36916239678537005,
      "learning_rate": 1.328997197869194e-05,
      "loss": 0.0081,
      "step": 774
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.42480431464993185,
      "learning_rate": 1.327364857623168e-05,
      "loss": 0.0109,
      "step": 775
    },
    {
      "epoch": 1.2416,
      "grad_norm": 0.3006130760508486,
      "learning_rate": 1.3257315398159865e-05,
      "loss": 0.0107,
      "step": 776
    },
    {
      "epoch": 1.2432,
      "grad_norm": 0.3883539027864325,
      "learning_rate": 1.3240972493249846e-05,
      "loss": 0.0096,
      "step": 777
    },
    {
      "epoch": 1.2448,
      "grad_norm": 0.1962408454725349,
      "learning_rate": 1.3224619910304019e-05,
      "loss": 0.0072,
      "step": 778
    },
    {
      "epoch": 1.2464,
      "grad_norm": 0.45417639089173384,
      "learning_rate": 1.3208257698153677e-05,
      "loss": 0.014,
      "step": 779
    },
    {
      "epoch": 1.248,
      "grad_norm": 0.3029083926751555,
      "learning_rate": 1.3191885905658873e-05,
      "loss": 0.0085,
      "step": 780
    },
    {
      "epoch": 1.2496,
      "grad_norm": 0.36776562477996894,
      "learning_rate": 1.3175504581708261e-05,
      "loss": 0.0106,
      "step": 781
    },
    {
      "epoch": 1.2511999999999999,
      "grad_norm": 0.3492474614748319,
      "learning_rate": 1.3159113775218963e-05,
      "loss": 0.0128,
      "step": 782
    },
    {
      "epoch": 1.2528000000000001,
      "grad_norm": 0.404422451829511,
      "learning_rate": 1.3142713535136413e-05,
      "loss": 0.0132,
      "step": 783
    },
    {
      "epoch": 1.2544,
      "grad_norm": 0.4650420812195629,
      "learning_rate": 1.3126303910434215e-05,
      "loss": 0.0086,
      "step": 784
    },
    {
      "epoch": 1.256,
      "grad_norm": 0.29560248543685946,
      "learning_rate": 1.3109884950114007e-05,
      "loss": 0.0095,
      "step": 785
    },
    {
      "epoch": 1.2576,
      "grad_norm": 0.17946868198520566,
      "learning_rate": 1.309345670320529e-05,
      "loss": 0.0071,
      "step": 786
    },
    {
      "epoch": 1.2591999999999999,
      "grad_norm": 0.4780712178705736,
      "learning_rate": 1.3077019218765306e-05,
      "loss": 0.0084,
      "step": 787
    },
    {
      "epoch": 1.2608,
      "grad_norm": 0.4451640294949449,
      "learning_rate": 1.3060572545878875e-05,
      "loss": 0.0137,
      "step": 788
    },
    {
      "epoch": 1.2624,
      "grad_norm": 0.42930639864800635,
      "learning_rate": 1.3044116733658261e-05,
      "loss": 0.0109,
      "step": 789
    },
    {
      "epoch": 1.264,
      "grad_norm": 0.23600539351190586,
      "learning_rate": 1.302765183124302e-05,
      "loss": 0.0068,
      "step": 790
    },
    {
      "epoch": 1.2656,
      "grad_norm": 0.37736631357521966,
      "learning_rate": 1.3011177887799846e-05,
      "loss": 0.0115,
      "step": 791
    },
    {
      "epoch": 1.2671999999999999,
      "grad_norm": 0.2326939370182424,
      "learning_rate": 1.2994694952522435e-05,
      "loss": 0.0089,
      "step": 792
    },
    {
      "epoch": 1.2688,
      "grad_norm": 0.2437980610117518,
      "learning_rate": 1.2978203074631335e-05,
      "loss": 0.008,
      "step": 793
    },
    {
      "epoch": 1.2704,
      "grad_norm": 0.42895841526592327,
      "learning_rate": 1.2961702303373795e-05,
      "loss": 0.0084,
      "step": 794
    },
    {
      "epoch": 1.272,
      "grad_norm": 0.4639404626719677,
      "learning_rate": 1.2945192688023625e-05,
      "loss": 0.0112,
      "step": 795
    },
    {
      "epoch": 1.2736,
      "grad_norm": 0.33281863236309694,
      "learning_rate": 1.2928674277881041e-05,
      "loss": 0.0071,
      "step": 796
    },
    {
      "epoch": 1.2752,
      "grad_norm": 0.5574126754068867,
      "learning_rate": 1.2912147122272523e-05,
      "loss": 0.0109,
      "step": 797
    },
    {
      "epoch": 1.2768,
      "grad_norm": 0.2797917785944588,
      "learning_rate": 1.2895611270550666e-05,
      "loss": 0.0067,
      "step": 798
    },
    {
      "epoch": 1.2784,
      "grad_norm": 0.3907726779123967,
      "learning_rate": 1.287906677209403e-05,
      "loss": 0.0084,
      "step": 799
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.3629313864103953,
      "learning_rate": 1.2862513676307009e-05,
      "loss": 0.0099,
      "step": 800
    },
    {
      "epoch": 1.2816,
      "grad_norm": 0.2278749102527171,
      "learning_rate": 1.2845952032619651e-05,
      "loss": 0.0082,
      "step": 801
    },
    {
      "epoch": 1.2832,
      "grad_norm": 0.32473684811473386,
      "learning_rate": 1.2829381890487536e-05,
      "loss": 0.0072,
      "step": 802
    },
    {
      "epoch": 1.2848,
      "grad_norm": 0.3133256938184833,
      "learning_rate": 1.2812803299391629e-05,
      "loss": 0.0082,
      "step": 803
    },
    {
      "epoch": 1.2864,
      "grad_norm": 0.24712866401984404,
      "learning_rate": 1.2796216308838116e-05,
      "loss": 0.009,
      "step": 804
    },
    {
      "epoch": 1.288,
      "grad_norm": 0.329561597917042,
      "learning_rate": 1.2779620968358276e-05,
      "loss": 0.0099,
      "step": 805
    },
    {
      "epoch": 1.2896,
      "grad_norm": 0.3674812322915868,
      "learning_rate": 1.2763017327508304e-05,
      "loss": 0.0086,
      "step": 806
    },
    {
      "epoch": 1.2912,
      "grad_norm": 0.5298508650102428,
      "learning_rate": 1.2746405435869198e-05,
      "loss": 0.0096,
      "step": 807
    },
    {
      "epoch": 1.2928,
      "grad_norm": 0.36287308162787724,
      "learning_rate": 1.2729785343046587e-05,
      "loss": 0.0104,
      "step": 808
    },
    {
      "epoch": 1.2944,
      "grad_norm": 0.4298281989328197,
      "learning_rate": 1.271315709867059e-05,
      "loss": 0.012,
      "step": 809
    },
    {
      "epoch": 1.296,
      "grad_norm": 0.35227509950172525,
      "learning_rate": 1.2696520752395671e-05,
      "loss": 0.0149,
      "step": 810
    },
    {
      "epoch": 1.2976,
      "grad_norm": 0.29161432482983723,
      "learning_rate": 1.2679876353900482e-05,
      "loss": 0.0069,
      "step": 811
    },
    {
      "epoch": 1.2992,
      "grad_norm": 0.39437309067730203,
      "learning_rate": 1.2663223952887724e-05,
      "loss": 0.0092,
      "step": 812
    },
    {
      "epoch": 1.3008,
      "grad_norm": 0.44244490142285775,
      "learning_rate": 1.2646563599083997e-05,
      "loss": 0.0129,
      "step": 813
    },
    {
      "epoch": 1.3024,
      "grad_norm": 0.25192147580518404,
      "learning_rate": 1.2629895342239643e-05,
      "loss": 0.0083,
      "step": 814
    },
    {
      "epoch": 1.304,
      "grad_norm": 0.3250667811497033,
      "learning_rate": 1.2613219232128608e-05,
      "loss": 0.0068,
      "step": 815
    },
    {
      "epoch": 1.3056,
      "grad_norm": 0.23326374829818386,
      "learning_rate": 1.2596535318548288e-05,
      "loss": 0.0088,
      "step": 816
    },
    {
      "epoch": 1.3072,
      "grad_norm": 0.32973641631747075,
      "learning_rate": 1.2579843651319382e-05,
      "loss": 0.0112,
      "step": 817
    },
    {
      "epoch": 1.3088,
      "grad_norm": 0.5578728814053906,
      "learning_rate": 1.2563144280285742e-05,
      "loss": 0.0102,
      "step": 818
    },
    {
      "epoch": 1.3104,
      "grad_norm": 0.30049060516878323,
      "learning_rate": 1.2546437255314223e-05,
      "loss": 0.0081,
      "step": 819
    },
    {
      "epoch": 1.312,
      "grad_norm": 0.5041765014830494,
      "learning_rate": 1.252972262629454e-05,
      "loss": 0.0125,
      "step": 820
    },
    {
      "epoch": 1.3136,
      "grad_norm": 0.2746317455129088,
      "learning_rate": 1.2513000443139112e-05,
      "loss": 0.0079,
      "step": 821
    },
    {
      "epoch": 1.3152,
      "grad_norm": 0.22655587379623343,
      "learning_rate": 1.2496270755782913e-05,
      "loss": 0.0061,
      "step": 822
    },
    {
      "epoch": 1.3168,
      "grad_norm": 0.43052405453673703,
      "learning_rate": 1.2479533614183334e-05,
      "loss": 0.012,
      "step": 823
    },
    {
      "epoch": 1.3184,
      "grad_norm": 0.5377323132322557,
      "learning_rate": 1.2462789068320016e-05,
      "loss": 0.0142,
      "step": 824
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.1647183059052667,
      "learning_rate": 1.2446037168194716e-05,
      "loss": 0.0062,
      "step": 825
    },
    {
      "epoch": 1.3216,
      "grad_norm": 0.32227842625423486,
      "learning_rate": 1.2429277963831147e-05,
      "loss": 0.0075,
      "step": 826
    },
    {
      "epoch": 1.3232,
      "grad_norm": 0.37578055706478664,
      "learning_rate": 1.2412511505274845e-05,
      "loss": 0.0097,
      "step": 827
    },
    {
      "epoch": 1.3248,
      "grad_norm": 0.34278180839303224,
      "learning_rate": 1.2395737842592997e-05,
      "loss": 0.0092,
      "step": 828
    },
    {
      "epoch": 1.3264,
      "grad_norm": 0.44306859364790885,
      "learning_rate": 1.23789570258743e-05,
      "loss": 0.0079,
      "step": 829
    },
    {
      "epoch": 1.328,
      "grad_norm": 0.24536713994929546,
      "learning_rate": 1.2362169105228828e-05,
      "loss": 0.0066,
      "step": 830
    },
    {
      "epoch": 1.3296000000000001,
      "grad_norm": 0.37085850581722773,
      "learning_rate": 1.2345374130787855e-05,
      "loss": 0.0104,
      "step": 831
    },
    {
      "epoch": 1.3312,
      "grad_norm": 0.37085850581722773,
      "learning_rate": 1.2345374130787855e-05,
      "loss": 0.0066,
      "step": 832
    },
    {
      "epoch": 1.3328,
      "grad_norm": 0.30848218442545855,
      "learning_rate": 1.2328572152703726e-05,
      "loss": 0.0103,
      "step": 833
    },
    {
      "epoch": 1.3344,
      "grad_norm": 0.3627469324704478,
      "learning_rate": 1.23117632211497e-05,
      "loss": 0.0099,
      "step": 834
    },
    {
      "epoch": 1.336,
      "grad_norm": 0.3160131608223671,
      "learning_rate": 1.2294947386319793e-05,
      "loss": 0.0122,
      "step": 835
    },
    {
      "epoch": 1.3376000000000001,
      "grad_norm": 0.2648370818046249,
      "learning_rate": 1.2278124698428643e-05,
      "loss": 0.0085,
      "step": 836
    },
    {
      "epoch": 1.3392,
      "grad_norm": 0.22951803230834383,
      "learning_rate": 1.2261295207711347e-05,
      "loss": 0.0082,
      "step": 837
    },
    {
      "epoch": 1.3408,
      "grad_norm": 0.2807748471689655,
      "learning_rate": 1.2244458964423328e-05,
      "loss": 0.0064,
      "step": 838
    },
    {
      "epoch": 1.3424,
      "grad_norm": 0.424777188861044,
      "learning_rate": 1.2227616018840154e-05,
      "loss": 0.0105,
      "step": 839
    },
    {
      "epoch": 1.3439999999999999,
      "grad_norm": 0.4438776391511739,
      "learning_rate": 1.221076642125742e-05,
      "loss": 0.0106,
      "step": 840
    },
    {
      "epoch": 1.3456000000000001,
      "grad_norm": 0.4737141697531989,
      "learning_rate": 1.2193910221990582e-05,
      "loss": 0.0091,
      "step": 841
    },
    {
      "epoch": 1.3472,
      "grad_norm": 0.22483995906080545,
      "learning_rate": 1.2177047471374808e-05,
      "loss": 0.0066,
      "step": 842
    },
    {
      "epoch": 1.3488,
      "grad_norm": 0.2966432665008213,
      "learning_rate": 1.2160178219764838e-05,
      "loss": 0.006,
      "step": 843
    },
    {
      "epoch": 1.3504,
      "grad_norm": 0.2922693904592179,
      "learning_rate": 1.214330251753481e-05,
      "loss": 0.0096,
      "step": 844
    },
    {
      "epoch": 1.3519999999999999,
      "grad_norm": 0.3977189742221552,
      "learning_rate": 1.2126420415078133e-05,
      "loss": 0.0118,
      "step": 845
    },
    {
      "epoch": 1.3536000000000001,
      "grad_norm": 0.3505353141532119,
      "learning_rate": 1.2109531962807333e-05,
      "loss": 0.0076,
      "step": 846
    },
    {
      "epoch": 1.3552,
      "grad_norm": 0.23791890889957615,
      "learning_rate": 1.2092637211153885e-05,
      "loss": 0.0061,
      "step": 847
    },
    {
      "epoch": 1.3568,
      "grad_norm": 0.3527591437661224,
      "learning_rate": 1.207573621056809e-05,
      "loss": 0.0079,
      "step": 848
    },
    {
      "epoch": 1.3584,
      "grad_norm": 0.40075379763225033,
      "learning_rate": 1.2058829011518896e-05,
      "loss": 0.0083,
      "step": 849
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 0.29395282358595143,
      "learning_rate": 1.2041915664493763e-05,
      "loss": 0.0093,
      "step": 850
    },
    {
      "epoch": 1.3616,
      "grad_norm": 0.38626848636739086,
      "learning_rate": 1.2024996219998517e-05,
      "loss": 0.0089,
      "step": 851
    },
    {
      "epoch": 1.3632,
      "grad_norm": 0.362595249446703,
      "learning_rate": 1.2008070728557186e-05,
      "loss": 0.0112,
      "step": 852
    },
    {
      "epoch": 1.3648,
      "grad_norm": 0.34489646540213964,
      "learning_rate": 1.1991139240711857e-05,
      "loss": 0.0067,
      "step": 853
    },
    {
      "epoch": 1.3664,
      "grad_norm": 0.20804277398188623,
      "learning_rate": 1.1974201807022525e-05,
      "loss": 0.0053,
      "step": 854
    },
    {
      "epoch": 1.3679999999999999,
      "grad_norm": 0.629465566724322,
      "learning_rate": 1.195725847806693e-05,
      "loss": 0.0122,
      "step": 855
    },
    {
      "epoch": 1.3696,
      "grad_norm": 0.29739632175228264,
      "learning_rate": 1.1940309304440434e-05,
      "loss": 0.0057,
      "step": 856
    },
    {
      "epoch": 1.3712,
      "grad_norm": 0.36866457031721195,
      "learning_rate": 1.1923354336755835e-05,
      "loss": 0.0099,
      "step": 857
    },
    {
      "epoch": 1.3728,
      "grad_norm": 0.2528063106828151,
      "learning_rate": 1.1906393625643244e-05,
      "loss": 0.0068,
      "step": 858
    },
    {
      "epoch": 1.3744,
      "grad_norm": 0.5200816773004503,
      "learning_rate": 1.1889427221749916e-05,
      "loss": 0.0085,
      "step": 859
    },
    {
      "epoch": 1.376,
      "grad_norm": 0.4688297926594098,
      "learning_rate": 1.1872455175740111e-05,
      "loss": 0.0086,
      "step": 860
    },
    {
      "epoch": 1.3776,
      "grad_norm": 0.42639147595917204,
      "learning_rate": 1.1855477538294934e-05,
      "loss": 0.0115,
      "step": 861
    },
    {
      "epoch": 1.3792,
      "grad_norm": 0.36162926831552683,
      "learning_rate": 1.1838494360112185e-05,
      "loss": 0.0075,
      "step": 862
    },
    {
      "epoch": 1.3808,
      "grad_norm": 0.377285170693332,
      "learning_rate": 1.1821505691906216e-05,
      "loss": 0.008,
      "step": 863
    },
    {
      "epoch": 1.3824,
      "grad_norm": 0.26461382490495844,
      "learning_rate": 1.1804511584407763e-05,
      "loss": 0.0057,
      "step": 864
    },
    {
      "epoch": 1.384,
      "grad_norm": 0.5373846258454587,
      "learning_rate": 1.1787512088363817e-05,
      "loss": 0.01,
      "step": 865
    },
    {
      "epoch": 1.3856,
      "grad_norm": 0.2500826668207825,
      "learning_rate": 1.1770507254537454e-05,
      "loss": 0.006,
      "step": 866
    },
    {
      "epoch": 1.3872,
      "grad_norm": 0.3530484206619875,
      "learning_rate": 1.1753497133707678e-05,
      "loss": 0.0076,
      "step": 867
    },
    {
      "epoch": 1.3888,
      "grad_norm": 0.580708563505262,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 0.0077,
      "step": 868
    },
    {
      "epoch": 1.3904,
      "grad_norm": 0.39933089144151734,
      "learning_rate": 1.1719461234232765e-05,
      "loss": 0.0096,
      "step": 869
    },
    {
      "epoch": 1.392,
      "grad_norm": 0.418183007425909,
      "learning_rate": 1.1702435557223988e-05,
      "loss": 0.0083,
      "step": 870
    },
    {
      "epoch": 1.3936,
      "grad_norm": 0.36181371076167196,
      "learning_rate": 1.1685404796484226e-05,
      "loss": 0.0077,
      "step": 871
    },
    {
      "epoch": 1.3952,
      "grad_norm": 0.5490040008065915,
      "learning_rate": 1.1668369002869912e-05,
      "loss": 0.01,
      "step": 872
    },
    {
      "epoch": 1.3968,
      "grad_norm": 0.5014599138442379,
      "learning_rate": 1.1651328227252516e-05,
      "loss": 0.0123,
      "step": 873
    },
    {
      "epoch": 1.3984,
      "grad_norm": 0.35423008795715893,
      "learning_rate": 1.1634282520518382e-05,
      "loss": 0.0089,
      "step": 874
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.2803589900876237,
      "learning_rate": 1.1617231933568579e-05,
      "loss": 0.0067,
      "step": 875
    },
    {
      "epoch": 1.4016,
      "grad_norm": 0.4036687277196133,
      "learning_rate": 1.1600176517318742e-05,
      "loss": 0.0091,
      "step": 876
    },
    {
      "epoch": 1.4032,
      "grad_norm": 0.5007461160934455,
      "learning_rate": 1.1583116322698936e-05,
      "loss": 0.0095,
      "step": 877
    },
    {
      "epoch": 1.4048,
      "grad_norm": 0.38342918619428096,
      "learning_rate": 1.1566051400653486e-05,
      "loss": 0.0129,
      "step": 878
    },
    {
      "epoch": 1.4064,
      "grad_norm": 0.305523477144802,
      "learning_rate": 1.1548981802140849e-05,
      "loss": 0.0082,
      "step": 879
    },
    {
      "epoch": 1.408,
      "grad_norm": 0.34758149839035163,
      "learning_rate": 1.153190757813343e-05,
      "loss": 0.01,
      "step": 880
    },
    {
      "epoch": 1.4096,
      "grad_norm": 0.29249731470991963,
      "learning_rate": 1.151482877961746e-05,
      "loss": 0.0074,
      "step": 881
    },
    {
      "epoch": 1.4112,
      "grad_norm": 0.26005852708222554,
      "learning_rate": 1.1497745457592817e-05,
      "loss": 0.0068,
      "step": 882
    },
    {
      "epoch": 1.4128,
      "grad_norm": 0.37864495517615604,
      "learning_rate": 1.1480657663072896e-05,
      "loss": 0.0095,
      "step": 883
    },
    {
      "epoch": 1.4144,
      "grad_norm": 0.26262297208151797,
      "learning_rate": 1.1463565447084446e-05,
      "loss": 0.0053,
      "step": 884
    },
    {
      "epoch": 1.416,
      "grad_norm": 0.457891954315086,
      "learning_rate": 1.1446468860667422e-05,
      "loss": 0.0106,
      "step": 885
    },
    {
      "epoch": 1.4176,
      "grad_norm": 0.3154968911620306,
      "learning_rate": 1.142936795487482e-05,
      "loss": 0.0091,
      "step": 886
    },
    {
      "epoch": 1.4192,
      "grad_norm": 0.34682416254437193,
      "learning_rate": 1.141226278077254e-05,
      "loss": 0.0076,
      "step": 887
    },
    {
      "epoch": 1.4208,
      "grad_norm": 0.2687387433904091,
      "learning_rate": 1.1395153389439232e-05,
      "loss": 0.0095,
      "step": 888
    },
    {
      "epoch": 1.4224,
      "grad_norm": 0.29596337803682293,
      "learning_rate": 1.1378039831966134e-05,
      "loss": 0.007,
      "step": 889
    },
    {
      "epoch": 1.424,
      "grad_norm": 0.36130970492428094,
      "learning_rate": 1.1360922159456929e-05,
      "loss": 0.0079,
      "step": 890
    },
    {
      "epoch": 1.4256,
      "grad_norm": 0.31624384251425164,
      "learning_rate": 1.1343800423027583e-05,
      "loss": 0.0118,
      "step": 891
    },
    {
      "epoch": 1.4272,
      "grad_norm": 0.36534736351875546,
      "learning_rate": 1.1326674673806195e-05,
      "loss": 0.0128,
      "step": 892
    },
    {
      "epoch": 1.4288,
      "grad_norm": 0.2851068872408835,
      "learning_rate": 1.1309544962932861e-05,
      "loss": 0.0073,
      "step": 893
    },
    {
      "epoch": 1.4304000000000001,
      "grad_norm": 0.3145959788278407,
      "learning_rate": 1.129241134155949e-05,
      "loss": 0.0077,
      "step": 894
    },
    {
      "epoch": 1.432,
      "grad_norm": 0.29982859274561346,
      "learning_rate": 1.1275273860849684e-05,
      "loss": 0.0079,
      "step": 895
    },
    {
      "epoch": 1.4336,
      "grad_norm": 0.36762779188609057,
      "learning_rate": 1.1258132571978555e-05,
      "loss": 0.0099,
      "step": 896
    },
    {
      "epoch": 1.4352,
      "grad_norm": 0.3316833150262433,
      "learning_rate": 1.1240987526132595e-05,
      "loss": 0.0092,
      "step": 897
    },
    {
      "epoch": 1.4368,
      "grad_norm": 0.3136355214354166,
      "learning_rate": 1.1223838774509515e-05,
      "loss": 0.0078,
      "step": 898
    },
    {
      "epoch": 1.4384000000000001,
      "grad_norm": 0.17982547977128643,
      "learning_rate": 1.1206686368318087e-05,
      "loss": 0.006,
      "step": 899
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.33169527081978134,
      "learning_rate": 1.1189530358778005e-05,
      "loss": 0.0066,
      "step": 900
    },
    {
      "epoch": 1.4416,
      "grad_norm": 0.3896441580222428,
      "learning_rate": 1.1172370797119711e-05,
      "loss": 0.0078,
      "step": 901
    },
    {
      "epoch": 1.4432,
      "grad_norm": 0.5813066018410238,
      "learning_rate": 1.1155207734584264e-05,
      "loss": 0.0102,
      "step": 902
    },
    {
      "epoch": 1.4447999999999999,
      "grad_norm": 0.4291550297920064,
      "learning_rate": 1.1138041222423177e-05,
      "loss": 0.0093,
      "step": 903
    },
    {
      "epoch": 1.4464000000000001,
      "grad_norm": 0.25388174158389226,
      "learning_rate": 1.1120871311898254e-05,
      "loss": 0.0084,
      "step": 904
    },
    {
      "epoch": 1.448,
      "grad_norm": 0.3056076978259771,
      "learning_rate": 1.110369805428146e-05,
      "loss": 0.0072,
      "step": 905
    },
    {
      "epoch": 1.4496,
      "grad_norm": 0.3181572825323978,
      "learning_rate": 1.1086521500854746e-05,
      "loss": 0.008,
      "step": 906
    },
    {
      "epoch": 1.4512,
      "grad_norm": 0.3642763841578639,
      "learning_rate": 1.106934170290991e-05,
      "loss": 0.0085,
      "step": 907
    },
    {
      "epoch": 1.4527999999999999,
      "grad_norm": 0.2851410982407962,
      "learning_rate": 1.1052158711748435e-05,
      "loss": 0.0082,
      "step": 908
    },
    {
      "epoch": 1.4544000000000001,
      "grad_norm": 0.3791614811930034,
      "learning_rate": 1.1034972578681338e-05,
      "loss": 0.0119,
      "step": 909
    },
    {
      "epoch": 1.456,
      "grad_norm": 0.3177905578295756,
      "learning_rate": 1.1017783355029027e-05,
      "loss": 0.0066,
      "step": 910
    },
    {
      "epoch": 1.4576,
      "grad_norm": 0.47774781874768774,
      "learning_rate": 1.1000591092121126e-05,
      "loss": 0.0113,
      "step": 911
    },
    {
      "epoch": 1.4592,
      "grad_norm": 0.3274200811499365,
      "learning_rate": 1.0983395841296349e-05,
      "loss": 0.0088,
      "step": 912
    },
    {
      "epoch": 1.4607999999999999,
      "grad_norm": 0.356859262010724,
      "learning_rate": 1.0966197653902319e-05,
      "loss": 0.0077,
      "step": 913
    },
    {
      "epoch": 1.4624,
      "grad_norm": 0.3265891416269014,
      "learning_rate": 1.0948996581295437e-05,
      "loss": 0.0096,
      "step": 914
    },
    {
      "epoch": 1.464,
      "grad_norm": 0.43926335025409535,
      "learning_rate": 1.0931792674840718e-05,
      "loss": 0.0083,
      "step": 915
    },
    {
      "epoch": 1.4656,
      "grad_norm": 0.46204903984552365,
      "learning_rate": 1.0914585985911632e-05,
      "loss": 0.0073,
      "step": 916
    },
    {
      "epoch": 1.4672,
      "grad_norm": 0.3547883499422264,
      "learning_rate": 1.0897376565889972e-05,
      "loss": 0.0089,
      "step": 917
    },
    {
      "epoch": 1.4687999999999999,
      "grad_norm": 0.31694588750563024,
      "learning_rate": 1.0880164466165675e-05,
      "loss": 0.006,
      "step": 918
    },
    {
      "epoch": 1.4704,
      "grad_norm": 0.3493040167389327,
      "learning_rate": 1.0862949738136682e-05,
      "loss": 0.01,
      "step": 919
    },
    {
      "epoch": 1.472,
      "grad_norm": 0.2912440674322754,
      "learning_rate": 1.084573243320878e-05,
      "loss": 0.0078,
      "step": 920
    },
    {
      "epoch": 1.4736,
      "grad_norm": 0.26374265472204816,
      "learning_rate": 1.0828512602795462e-05,
      "loss": 0.0072,
      "step": 921
    },
    {
      "epoch": 1.4752,
      "grad_norm": 0.5666887421121791,
      "learning_rate": 1.0811290298317755e-05,
      "loss": 0.0092,
      "step": 922
    },
    {
      "epoch": 1.4768,
      "grad_norm": 0.6282685860166906,
      "learning_rate": 1.0794065571204073e-05,
      "loss": 0.0118,
      "step": 923
    },
    {
      "epoch": 1.4784,
      "grad_norm": 0.3907357745744409,
      "learning_rate": 1.0776838472890065e-05,
      "loss": 0.0071,
      "step": 924
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.24739569338370557,
      "learning_rate": 1.0759609054818459e-05,
      "loss": 0.0075,
      "step": 925
    },
    {
      "epoch": 1.4816,
      "grad_norm": 0.5111206065703442,
      "learning_rate": 1.0742377368438915e-05,
      "loss": 0.0085,
      "step": 926
    },
    {
      "epoch": 1.4832,
      "grad_norm": 0.37201673466090657,
      "learning_rate": 1.0725143465207868e-05,
      "loss": 0.0069,
      "step": 927
    },
    {
      "epoch": 1.4848,
      "grad_norm": 0.3401601865747001,
      "learning_rate": 1.0707907396588362e-05,
      "loss": 0.0075,
      "step": 928
    },
    {
      "epoch": 1.4864,
      "grad_norm": 0.28814263010972685,
      "learning_rate": 1.069066921404992e-05,
      "loss": 0.0076,
      "step": 929
    },
    {
      "epoch": 1.488,
      "grad_norm": 0.44674818094500446,
      "learning_rate": 1.0673428969068365e-05,
      "loss": 0.0081,
      "step": 930
    },
    {
      "epoch": 1.4896,
      "grad_norm": 0.39018171835308646,
      "learning_rate": 1.065618671312569e-05,
      "loss": 0.0079,
      "step": 931
    },
    {
      "epoch": 1.4912,
      "grad_norm": 0.425351217131543,
      "learning_rate": 1.063894249770989e-05,
      "loss": 0.0063,
      "step": 932
    },
    {
      "epoch": 1.4928,
      "grad_norm": 0.24044988708296725,
      "learning_rate": 1.0621696374314807e-05,
      "loss": 0.005,
      "step": 933
    },
    {
      "epoch": 1.4944,
      "grad_norm": 0.39605196566520534,
      "learning_rate": 1.0604448394439983e-05,
      "loss": 0.0082,
      "step": 934
    },
    {
      "epoch": 1.496,
      "grad_norm": 0.2975298766800573,
      "learning_rate": 1.0587198609590505e-05,
      "loss": 0.0053,
      "step": 935
    },
    {
      "epoch": 1.4976,
      "grad_norm": 0.1918028503772937,
      "learning_rate": 1.0569947071276847e-05,
      "loss": 0.0056,
      "step": 936
    },
    {
      "epoch": 1.4992,
      "grad_norm": 0.5225196763397442,
      "learning_rate": 1.0552693831014726e-05,
      "loss": 0.0073,
      "step": 937
    },
    {
      "epoch": 1.5008,
      "grad_norm": 0.4740052207484797,
      "learning_rate": 1.053543894032493e-05,
      "loss": 0.0086,
      "step": 938
    },
    {
      "epoch": 1.5024,
      "grad_norm": 0.38194720723505593,
      "learning_rate": 1.0518182450733185e-05,
      "loss": 0.0065,
      "step": 939
    },
    {
      "epoch": 1.504,
      "grad_norm": 0.5756849630159504,
      "learning_rate": 1.0500924413769988e-05,
      "loss": 0.0128,
      "step": 940
    },
    {
      "epoch": 1.5056,
      "grad_norm": 0.510189764918798,
      "learning_rate": 1.0483664880970456e-05,
      "loss": 0.008,
      "step": 941
    },
    {
      "epoch": 1.5072,
      "grad_norm": 0.3040717534673568,
      "learning_rate": 1.0466403903874176e-05,
      "loss": 0.0063,
      "step": 942
    },
    {
      "epoch": 1.5088,
      "grad_norm": 0.3804735334346367,
      "learning_rate": 1.0449141534025044e-05,
      "loss": 0.0085,
      "step": 943
    },
    {
      "epoch": 1.5104,
      "grad_norm": 0.3791484229367134,
      "learning_rate": 1.0431877822971118e-05,
      "loss": 0.0056,
      "step": 944
    },
    {
      "epoch": 1.512,
      "grad_norm": 0.5429868292228728,
      "learning_rate": 1.0414612822264457e-05,
      "loss": 0.0111,
      "step": 945
    },
    {
      "epoch": 1.5135999999999998,
      "grad_norm": 0.48848710775177834,
      "learning_rate": 1.0397346583460972e-05,
      "loss": 0.0141,
      "step": 946
    },
    {
      "epoch": 1.5152,
      "grad_norm": 0.41318570074900424,
      "learning_rate": 1.038007915812028e-05,
      "loss": 0.0078,
      "step": 947
    },
    {
      "epoch": 1.5168,
      "grad_norm": 0.2347124585806924,
      "learning_rate": 1.0362810597805526e-05,
      "loss": 0.0073,
      "step": 948
    },
    {
      "epoch": 1.5184,
      "grad_norm": 0.2584944658144249,
      "learning_rate": 1.034554095408326e-05,
      "loss": 0.0061,
      "step": 949
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.3665948347221613,
      "learning_rate": 1.0328270278523256e-05,
      "loss": 0.0083,
      "step": 950
    },
    {
      "epoch": 1.5215999999999998,
      "grad_norm": 0.27856771161565724,
      "learning_rate": 1.031099862269837e-05,
      "loss": 0.0066,
      "step": 951
    },
    {
      "epoch": 1.5232,
      "grad_norm": 0.5144509420873353,
      "learning_rate": 1.0293726038184393e-05,
      "loss": 0.0097,
      "step": 952
    },
    {
      "epoch": 1.5248,
      "grad_norm": 0.44385552507763787,
      "learning_rate": 1.0276452576559878e-05,
      "loss": 0.0073,
      "step": 953
    },
    {
      "epoch": 1.5264,
      "grad_norm": 0.22407647544607326,
      "learning_rate": 1.0259178289406011e-05,
      "loss": 0.0061,
      "step": 954
    },
    {
      "epoch": 1.528,
      "grad_norm": 0.2805711214308104,
      "learning_rate": 1.024190322830643e-05,
      "loss": 0.0058,
      "step": 955
    },
    {
      "epoch": 1.5295999999999998,
      "grad_norm": 0.404703550361594,
      "learning_rate": 1.022462744484709e-05,
      "loss": 0.0076,
      "step": 956
    },
    {
      "epoch": 1.5312000000000001,
      "grad_norm": 0.2694424159798581,
      "learning_rate": 1.0207350990616107e-05,
      "loss": 0.0083,
      "step": 957
    },
    {
      "epoch": 1.5328,
      "grad_norm": 0.21143467606914215,
      "learning_rate": 1.019007391720359e-05,
      "loss": 0.0044,
      "step": 958
    },
    {
      "epoch": 1.5344,
      "grad_norm": 0.5338154828197106,
      "learning_rate": 1.0172796276201504e-05,
      "loss": 0.0083,
      "step": 959
    },
    {
      "epoch": 1.536,
      "grad_norm": 0.2364234305089969,
      "learning_rate": 1.0155518119203511e-05,
      "loss": 0.0062,
      "step": 960
    },
    {
      "epoch": 1.5375999999999999,
      "grad_norm": 0.4413487129601474,
      "learning_rate": 1.0138239497804804e-05,
      "loss": 0.0068,
      "step": 961
    },
    {
      "epoch": 1.5392000000000001,
      "grad_norm": 0.5534092282345944,
      "learning_rate": 1.0120960463601977e-05,
      "loss": 0.0124,
      "step": 962
    },
    {
      "epoch": 1.5408,
      "grad_norm": 0.17683301558606265,
      "learning_rate": 1.0103681068192845e-05,
      "loss": 0.0042,
      "step": 963
    },
    {
      "epoch": 1.5424,
      "grad_norm": 0.36698442517605695,
      "learning_rate": 1.0086401363176306e-05,
      "loss": 0.0065,
      "step": 964
    },
    {
      "epoch": 1.544,
      "grad_norm": 0.49359052658995684,
      "learning_rate": 1.0069121400152182e-05,
      "loss": 0.0105,
      "step": 965
    },
    {
      "epoch": 1.5455999999999999,
      "grad_norm": 0.3855375904953211,
      "learning_rate": 1.0051841230721065e-05,
      "loss": 0.0084,
      "step": 966
    },
    {
      "epoch": 1.5472000000000001,
      "grad_norm": 0.2950845647855933,
      "learning_rate": 1.0034560906484161e-05,
      "loss": 0.0048,
      "step": 967
    },
    {
      "epoch": 1.5488,
      "grad_norm": 0.8075864977538052,
      "learning_rate": 1.0017280479043148e-05,
      "loss": 0.0096,
      "step": 968
    },
    {
      "epoch": 1.5504,
      "grad_norm": 0.6355076338574505,
      "learning_rate": 1e-05,
      "loss": 0.0144,
      "step": 969
    },
    {
      "epoch": 1.552,
      "grad_norm": 0.3166150019684947,
      "learning_rate": 9.982719520956856e-06,
      "loss": 0.0095,
      "step": 970
    },
    {
      "epoch": 1.5535999999999999,
      "grad_norm": 0.5387976539923057,
      "learning_rate": 9.965439093515842e-06,
      "loss": 0.012,
      "step": 971
    },
    {
      "epoch": 1.5552000000000001,
      "grad_norm": 0.3613434605845047,
      "learning_rate": 9.948158769278939e-06,
      "loss": 0.0077,
      "step": 972
    },
    {
      "epoch": 1.5568,
      "grad_norm": 0.31729947652365154,
      "learning_rate": 9.930878599847822e-06,
      "loss": 0.0082,
      "step": 973
    },
    {
      "epoch": 1.5584,
      "grad_norm": 0.404159685281789,
      "learning_rate": 9.913598636823694e-06,
      "loss": 0.007,
      "step": 974
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.38660573811621596,
      "learning_rate": 9.896318931807155e-06,
      "loss": 0.0079,
      "step": 975
    },
    {
      "epoch": 1.5615999999999999,
      "grad_norm": 0.3594232944435872,
      "learning_rate": 9.879039536398023e-06,
      "loss": 0.0069,
      "step": 976
    },
    {
      "epoch": 1.5632000000000001,
      "grad_norm": 0.40316798951088684,
      "learning_rate": 9.861760502195197e-06,
      "loss": 0.0075,
      "step": 977
    },
    {
      "epoch": 1.5648,
      "grad_norm": 0.27917997843188214,
      "learning_rate": 9.844481880796492e-06,
      "loss": 0.0066,
      "step": 978
    },
    {
      "epoch": 1.5664,
      "grad_norm": 0.2498848300811543,
      "learning_rate": 9.827203723798498e-06,
      "loss": 0.0071,
      "step": 979
    },
    {
      "epoch": 1.568,
      "grad_norm": 0.31125100047983123,
      "learning_rate": 9.809926082796415e-06,
      "loss": 0.0065,
      "step": 980
    },
    {
      "epoch": 1.5695999999999999,
      "grad_norm": 0.3937536406425979,
      "learning_rate": 9.7926490093839e-06,
      "loss": 0.0063,
      "step": 981
    },
    {
      "epoch": 1.5712000000000002,
      "grad_norm": 0.3458932138437726,
      "learning_rate": 9.775372555152912e-06,
      "loss": 0.0078,
      "step": 982
    },
    {
      "epoch": 1.5728,
      "grad_norm": 0.34898050713560863,
      "learning_rate": 9.758096771693574e-06,
      "loss": 0.0077,
      "step": 983
    },
    {
      "epoch": 1.5744,
      "grad_norm": 0.2970444107821782,
      "learning_rate": 9.740821710593989e-06,
      "loss": 0.0055,
      "step": 984
    },
    {
      "epoch": 1.576,
      "grad_norm": 0.15653928019538896,
      "learning_rate": 9.723547423440122e-06,
      "loss": 0.0051,
      "step": 985
    },
    {
      "epoch": 1.5776,
      "grad_norm": 0.28161861859072923,
      "learning_rate": 9.70627396181561e-06,
      "loss": 0.0064,
      "step": 986
    },
    {
      "epoch": 1.5792000000000002,
      "grad_norm": 0.4613769443728261,
      "learning_rate": 9.689001377301634e-06,
      "loss": 0.0091,
      "step": 987
    },
    {
      "epoch": 1.5808,
      "grad_norm": 0.4574576544358834,
      "learning_rate": 9.671729721476747e-06,
      "loss": 0.0099,
      "step": 988
    },
    {
      "epoch": 1.5824,
      "grad_norm": 0.33530952880607473,
      "learning_rate": 9.654459045916743e-06,
      "loss": 0.0078,
      "step": 989
    },
    {
      "epoch": 1.584,
      "grad_norm": 0.28242677597079807,
      "learning_rate": 9.637189402194477e-06,
      "loss": 0.009,
      "step": 990
    },
    {
      "epoch": 1.5856,
      "grad_norm": 0.19422839526194863,
      "learning_rate": 9.619920841879726e-06,
      "loss": 0.0056,
      "step": 991
    },
    {
      "epoch": 1.5872000000000002,
      "grad_norm": 0.3148931713956359,
      "learning_rate": 9.602653416539031e-06,
      "loss": 0.0049,
      "step": 992
    },
    {
      "epoch": 1.5888,
      "grad_norm": 0.399269780526929,
      "learning_rate": 9.585387177735548e-06,
      "loss": 0.0077,
      "step": 993
    },
    {
      "epoch": 1.5904,
      "grad_norm": 0.19831642838835764,
      "learning_rate": 9.568122177028884e-06,
      "loss": 0.0054,
      "step": 994
    },
    {
      "epoch": 1.592,
      "grad_norm": 0.31601962588213334,
      "learning_rate": 9.550858465974958e-06,
      "loss": 0.0052,
      "step": 995
    },
    {
      "epoch": 1.5936,
      "grad_norm": 0.2582794553323895,
      "learning_rate": 9.533596096125826e-06,
      "loss": 0.0048,
      "step": 996
    },
    {
      "epoch": 1.5952,
      "grad_norm": 0.3926592333535472,
      "learning_rate": 9.516335119029547e-06,
      "loss": 0.0094,
      "step": 997
    },
    {
      "epoch": 1.5968,
      "grad_norm": 0.4871992079292709,
      "learning_rate": 9.499075586230014e-06,
      "loss": 0.0091,
      "step": 998
    },
    {
      "epoch": 1.5984,
      "grad_norm": 0.3472080830513595,
      "learning_rate": 9.481817549266817e-06,
      "loss": 0.0089,
      "step": 999
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.32112872839677564,
      "learning_rate": 9.464561059675073e-06,
      "loss": 0.0088,
      "step": 1000
    },
    {
      "epoch": 1.6016,
      "grad_norm": 0.3858214720607361,
      "learning_rate": 9.44730616898528e-06,
      "loss": 0.0091,
      "step": 1001
    },
    {
      "epoch": 1.6032,
      "grad_norm": 0.24614269707964617,
      "learning_rate": 9.430052928723153e-06,
      "loss": 0.0043,
      "step": 1002
    },
    {
      "epoch": 1.6048,
      "grad_norm": 0.4562912736776646,
      "learning_rate": 9.412801390409496e-06,
      "loss": 0.0056,
      "step": 1003
    },
    {
      "epoch": 1.6064,
      "grad_norm": 0.32457799252073727,
      "learning_rate": 9.395551605560018e-06,
      "loss": 0.0044,
      "step": 1004
    },
    {
      "epoch": 1.608,
      "grad_norm": 0.548586587859223,
      "learning_rate": 9.378303625685196e-06,
      "loss": 0.0078,
      "step": 1005
    },
    {
      "epoch": 1.6096,
      "grad_norm": 0.18743960053979214,
      "learning_rate": 9.361057502290112e-06,
      "loss": 0.0044,
      "step": 1006
    },
    {
      "epoch": 1.6112,
      "grad_norm": 0.2672443254835976,
      "learning_rate": 9.343813286874312e-06,
      "loss": 0.0047,
      "step": 1007
    },
    {
      "epoch": 1.6128,
      "grad_norm": 0.31131607843452186,
      "learning_rate": 9.326571030931636e-06,
      "loss": 0.0079,
      "step": 1008
    },
    {
      "epoch": 1.6143999999999998,
      "grad_norm": 0.344647486890118,
      "learning_rate": 9.309330785950086e-06,
      "loss": 0.0075,
      "step": 1009
    },
    {
      "epoch": 1.616,
      "grad_norm": 0.5556824055660701,
      "learning_rate": 9.292092603411642e-06,
      "loss": 0.0099,
      "step": 1010
    },
    {
      "epoch": 1.6176,
      "grad_norm": 0.4352851050232399,
      "learning_rate": 9.274856534792138e-06,
      "loss": 0.0063,
      "step": 1011
    },
    {
      "epoch": 1.6192,
      "grad_norm": 0.23656950577275335,
      "learning_rate": 9.257622631561085e-06,
      "loss": 0.0041,
      "step": 1012
    },
    {
      "epoch": 1.6208,
      "grad_norm": 0.29931781512890643,
      "learning_rate": 9.240390945181543e-06,
      "loss": 0.007,
      "step": 1013
    },
    {
      "epoch": 1.6223999999999998,
      "grad_norm": 0.2650588085569112,
      "learning_rate": 9.223161527109938e-06,
      "loss": 0.0051,
      "step": 1014
    },
    {
      "epoch": 1.624,
      "grad_norm": 0.2714204527615275,
      "learning_rate": 9.205934428795929e-06,
      "loss": 0.0073,
      "step": 1015
    },
    {
      "epoch": 1.6256,
      "grad_norm": 0.3800341050458012,
      "learning_rate": 9.188709701682246e-06,
      "loss": 0.0096,
      "step": 1016
    },
    {
      "epoch": 1.6272,
      "grad_norm": 0.3126363797153251,
      "learning_rate": 9.17148739720454e-06,
      "loss": 0.0051,
      "step": 1017
    },
    {
      "epoch": 1.6288,
      "grad_norm": 0.3386373119408399,
      "learning_rate": 9.154267566791224e-06,
      "loss": 0.0056,
      "step": 1018
    },
    {
      "epoch": 1.6303999999999998,
      "grad_norm": 0.45698087996637005,
      "learning_rate": 9.137050261863323e-06,
      "loss": 0.0089,
      "step": 1019
    },
    {
      "epoch": 1.6320000000000001,
      "grad_norm": 0.342596772259811,
      "learning_rate": 9.119835533834332e-06,
      "loss": 0.0059,
      "step": 1020
    },
    {
      "epoch": 1.6336,
      "grad_norm": 0.4333113205185105,
      "learning_rate": 9.102623434110028e-06,
      "loss": 0.0081,
      "step": 1021
    },
    {
      "epoch": 1.6352,
      "grad_norm": 0.330647644777723,
      "learning_rate": 9.085414014088368e-06,
      "loss": 0.0081,
      "step": 1022
    },
    {
      "epoch": 1.6368,
      "grad_norm": 0.3686712495111263,
      "learning_rate": 9.068207325159285e-06,
      "loss": 0.0069,
      "step": 1023
    },
    {
      "epoch": 1.6383999999999999,
      "grad_norm": 0.42138138717299284,
      "learning_rate": 9.051003418704566e-06,
      "loss": 0.0062,
      "step": 1024
    },
    {
      "epoch": 1.6400000000000001,
      "grad_norm": 0.3040381923635361,
      "learning_rate": 9.033802346097683e-06,
      "loss": 0.0062,
      "step": 1025
    },
    {
      "epoch": 1.6416,
      "grad_norm": 0.24752072470183697,
      "learning_rate": 9.016604158703654e-06,
      "loss": 0.0073,
      "step": 1026
    },
    {
      "epoch": 1.6432,
      "grad_norm": 0.40757468787887463,
      "learning_rate": 8.999408907878877e-06,
      "loss": 0.0061,
      "step": 1027
    },
    {
      "epoch": 1.6448,
      "grad_norm": 0.4507691465257879,
      "learning_rate": 8.982216644970978e-06,
      "loss": 0.0073,
      "step": 1028
    },
    {
      "epoch": 1.6463999999999999,
      "grad_norm": 0.2753376263149583,
      "learning_rate": 8.965027421318666e-06,
      "loss": 0.0053,
      "step": 1029
    },
    {
      "epoch": 1.6480000000000001,
      "grad_norm": 0.2693047821995353,
      "learning_rate": 8.947841288251568e-06,
      "loss": 0.0056,
      "step": 1030
    },
    {
      "epoch": 1.6496,
      "grad_norm": 0.2931505194078567,
      "learning_rate": 8.930658297090092e-06,
      "loss": 0.0077,
      "step": 1031
    },
    {
      "epoch": 1.6512,
      "grad_norm": 0.39489564425611573,
      "learning_rate": 8.913478499145255e-06,
      "loss": 0.0071,
      "step": 1032
    },
    {
      "epoch": 1.6528,
      "grad_norm": 0.31392814320075263,
      "learning_rate": 8.896301945718541e-06,
      "loss": 0.0054,
      "step": 1033
    },
    {
      "epoch": 1.6543999999999999,
      "grad_norm": 0.36157939324989835,
      "learning_rate": 8.879128688101749e-06,
      "loss": 0.0071,
      "step": 1034
    },
    {
      "epoch": 1.6560000000000001,
      "grad_norm": 0.5234260678164566,
      "learning_rate": 8.861958777576826e-06,
      "loss": 0.0056,
      "step": 1035
    },
    {
      "epoch": 1.6576,
      "grad_norm": 0.5320931361265643,
      "learning_rate": 8.844792265415738e-06,
      "loss": 0.0129,
      "step": 1036
    },
    {
      "epoch": 1.6592,
      "grad_norm": 0.3858984330716074,
      "learning_rate": 8.827629202880294e-06,
      "loss": 0.0074,
      "step": 1037
    },
    {
      "epoch": 1.6608,
      "grad_norm": 0.29373407740042495,
      "learning_rate": 8.810469641222001e-06,
      "loss": 0.0073,
      "step": 1038
    },
    {
      "epoch": 1.6623999999999999,
      "grad_norm": 0.37216888978843427,
      "learning_rate": 8.793313631681915e-06,
      "loss": 0.006,
      "step": 1039
    },
    {
      "epoch": 1.6640000000000001,
      "grad_norm": 0.3994723130190849,
      "learning_rate": 8.776161225490488e-06,
      "loss": 0.0071,
      "step": 1040
    },
    {
      "epoch": 1.6656,
      "grad_norm": 0.3248017020873982,
      "learning_rate": 8.759012473867407e-06,
      "loss": 0.0095,
      "step": 1041
    },
    {
      "epoch": 1.6672,
      "grad_norm": 0.4572658899164843,
      "learning_rate": 8.741867428021447e-06,
      "loss": 0.0089,
      "step": 1042
    },
    {
      "epoch": 1.6688,
      "grad_norm": 0.29580859378645913,
      "learning_rate": 8.72472613915032e-06,
      "loss": 0.0048,
      "step": 1043
    },
    {
      "epoch": 1.6703999999999999,
      "grad_norm": 0.20178460492639025,
      "learning_rate": 8.707588658440511e-06,
      "loss": 0.0047,
      "step": 1044
    },
    {
      "epoch": 1.6720000000000002,
      "grad_norm": 0.2449233525366233,
      "learning_rate": 8.690455037067142e-06,
      "loss": 0.0066,
      "step": 1045
    },
    {
      "epoch": 1.6736,
      "grad_norm": 0.2636928045909801,
      "learning_rate": 8.673325326193806e-06,
      "loss": 0.0057,
      "step": 1046
    },
    {
      "epoch": 1.6752,
      "grad_norm": 0.3635025762512276,
      "learning_rate": 8.656199576972424e-06,
      "loss": 0.0077,
      "step": 1047
    },
    {
      "epoch": 1.6768,
      "grad_norm": 0.4118276246432996,
      "learning_rate": 8.639077840543078e-06,
      "loss": 0.0085,
      "step": 1048
    },
    {
      "epoch": 1.6784,
      "grad_norm": 0.30512670055196606,
      "learning_rate": 8.621960168033868e-06,
      "loss": 0.0068,
      "step": 1049
    },
    {
      "epoch": 1.6800000000000002,
      "grad_norm": 0.38760281582430434,
      "learning_rate": 8.604846610560771e-06,
      "loss": 0.0069,
      "step": 1050
    },
    {
      "epoch": 1.6816,
      "grad_norm": 0.7300646784773884,
      "learning_rate": 8.587737219227462e-06,
      "loss": 0.0167,
      "step": 1051
    },
    {
      "epoch": 1.6832,
      "grad_norm": 0.3130205403934896,
      "learning_rate": 8.570632045125185e-06,
      "loss": 0.005,
      "step": 1052
    },
    {
      "epoch": 1.6848,
      "grad_norm": 0.35157751233453793,
      "learning_rate": 8.553531139332583e-06,
      "loss": 0.0052,
      "step": 1053
    },
    {
      "epoch": 1.6864,
      "grad_norm": 0.3671764642336765,
      "learning_rate": 8.536434552915555e-06,
      "loss": 0.0102,
      "step": 1054
    },
    {
      "epoch": 1.688,
      "grad_norm": 0.40735028939010787,
      "learning_rate": 8.519342336927106e-06,
      "loss": 0.0079,
      "step": 1055
    },
    {
      "epoch": 1.6896,
      "grad_norm": 0.20121176475177047,
      "learning_rate": 8.502254542407186e-06,
      "loss": 0.0041,
      "step": 1056
    },
    {
      "epoch": 1.6912,
      "grad_norm": 0.15124424016459737,
      "learning_rate": 8.485171220382545e-06,
      "loss": 0.0041,
      "step": 1057
    },
    {
      "epoch": 1.6928,
      "grad_norm": 1.221110816048998,
      "learning_rate": 8.468092421866575e-06,
      "loss": 0.0083,
      "step": 1058
    },
    {
      "epoch": 1.6944,
      "grad_norm": 0.32146297311032346,
      "learning_rate": 8.451018197859153e-06,
      "loss": 0.0065,
      "step": 1059
    },
    {
      "epoch": 1.696,
      "grad_norm": 0.19352312694489443,
      "learning_rate": 8.433948599346516e-06,
      "loss": 0.0047,
      "step": 1060
    },
    {
      "epoch": 1.6976,
      "grad_norm": 0.2887348899685335,
      "learning_rate": 8.41688367730107e-06,
      "loss": 0.0056,
      "step": 1061
    },
    {
      "epoch": 1.6992,
      "grad_norm": 0.25308885584787627,
      "learning_rate": 8.399823482681263e-06,
      "loss": 0.0055,
      "step": 1062
    },
    {
      "epoch": 1.7008,
      "grad_norm": 0.4536632873651025,
      "learning_rate": 8.382768066431427e-06,
      "loss": 0.0114,
      "step": 1063
    },
    {
      "epoch": 1.7024,
      "grad_norm": 0.3334783544636052,
      "learning_rate": 8.36571747948162e-06,
      "loss": 0.0077,
      "step": 1064
    },
    {
      "epoch": 1.704,
      "grad_norm": 0.5080509457816318,
      "learning_rate": 8.348671772747488e-06,
      "loss": 0.0064,
      "step": 1065
    },
    {
      "epoch": 1.7056,
      "grad_norm": 0.21655922542846942,
      "learning_rate": 8.331630997130091e-06,
      "loss": 0.0051,
      "step": 1066
    },
    {
      "epoch": 1.7072,
      "grad_norm": 0.32059563902166266,
      "learning_rate": 8.314595203515781e-06,
      "loss": 0.0065,
      "step": 1067
    },
    {
      "epoch": 1.7088,
      "grad_norm": 0.19953685775494243,
      "learning_rate": 8.297564442776014e-06,
      "loss": 0.005,
      "step": 1068
    },
    {
      "epoch": 1.7104,
      "grad_norm": 0.3299807282834272,
      "learning_rate": 8.280538765767236e-06,
      "loss": 0.0068,
      "step": 1069
    },
    {
      "epoch": 1.712,
      "grad_norm": 0.400352453437766,
      "learning_rate": 8.263518223330698e-06,
      "loss": 0.0062,
      "step": 1070
    },
    {
      "epoch": 1.7136,
      "grad_norm": 0.2856335747990602,
      "learning_rate": 8.246502866292324e-06,
      "loss": 0.0056,
      "step": 1071
    },
    {
      "epoch": 1.7151999999999998,
      "grad_norm": 0.2527366785060869,
      "learning_rate": 8.229492745462551e-06,
      "loss": 0.0072,
      "step": 1072
    },
    {
      "epoch": 1.7168,
      "grad_norm": 0.24010903888241203,
      "learning_rate": 8.212487911636185e-06,
      "loss": 0.0056,
      "step": 1073
    },
    {
      "epoch": 1.7184,
      "grad_norm": 0.23896142647226756,
      "learning_rate": 8.195488415592238e-06,
      "loss": 0.0044,
      "step": 1074
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.44078832518865635,
      "learning_rate": 8.17849430809379e-06,
      "loss": 0.011,
      "step": 1075
    },
    {
      "epoch": 1.7216,
      "grad_norm": 0.2928645606679149,
      "learning_rate": 8.161505639887818e-06,
      "loss": 0.0051,
      "step": 1076
    },
    {
      "epoch": 1.7231999999999998,
      "grad_norm": 0.22448380268567095,
      "learning_rate": 8.144522461705067e-06,
      "loss": 0.0053,
      "step": 1077
    },
    {
      "epoch": 1.7248,
      "grad_norm": 0.3144154252736782,
      "learning_rate": 8.12754482425989e-06,
      "loss": 0.0045,
      "step": 1078
    },
    {
      "epoch": 1.7264,
      "grad_norm": 0.39472540262238637,
      "learning_rate": 8.110572778250086e-06,
      "loss": 0.0064,
      "step": 1079
    },
    {
      "epoch": 1.728,
      "grad_norm": 0.22917733716831562,
      "learning_rate": 8.09360637435676e-06,
      "loss": 0.0031,
      "step": 1080
    },
    {
      "epoch": 1.7296,
      "grad_norm": 0.3201761452865257,
      "learning_rate": 8.076645663244168e-06,
      "loss": 0.0071,
      "step": 1081
    },
    {
      "epoch": 1.7311999999999999,
      "grad_norm": 0.3584742117015553,
      "learning_rate": 8.05969069555957e-06,
      "loss": 0.0056,
      "step": 1082
    },
    {
      "epoch": 1.7328000000000001,
      "grad_norm": 0.7920760407669944,
      "learning_rate": 8.042741521933071e-06,
      "loss": 0.0083,
      "step": 1083
    },
    {
      "epoch": 1.7344,
      "grad_norm": 0.4012984458746515,
      "learning_rate": 8.025798192977482e-06,
      "loss": 0.0035,
      "step": 1084
    },
    {
      "epoch": 1.736,
      "grad_norm": 0.3498836585904627,
      "learning_rate": 8.008860759288148e-06,
      "loss": 0.0059,
      "step": 1085
    },
    {
      "epoch": 1.7376,
      "grad_norm": 0.23419600252095124,
      "learning_rate": 7.991929271442817e-06,
      "loss": 0.003,
      "step": 1086
    },
    {
      "epoch": 1.7391999999999999,
      "grad_norm": 0.15684536876271404,
      "learning_rate": 7.975003780001486e-06,
      "loss": 0.0031,
      "step": 1087
    },
    {
      "epoch": 1.7408000000000001,
      "grad_norm": 0.27088599771134847,
      "learning_rate": 7.958084335506239e-06,
      "loss": 0.0051,
      "step": 1088
    },
    {
      "epoch": 1.7424,
      "grad_norm": 0.2713339506776891,
      "learning_rate": 7.941170988481108e-06,
      "loss": 0.0033,
      "step": 1089
    },
    {
      "epoch": 1.744,
      "grad_norm": 0.444222794720414,
      "learning_rate": 7.924263789431913e-06,
      "loss": 0.0073,
      "step": 1090
    },
    {
      "epoch": 1.7456,
      "grad_norm": 0.3699148307486384,
      "learning_rate": 7.907362788846116e-06,
      "loss": 0.0057,
      "step": 1091
    },
    {
      "epoch": 1.7471999999999999,
      "grad_norm": 0.5512431022822822,
      "learning_rate": 7.89046803719267e-06,
      "loss": 0.0068,
      "step": 1092
    },
    {
      "epoch": 1.7488000000000001,
      "grad_norm": 0.2578013990883215,
      "learning_rate": 7.873579584921869e-06,
      "loss": 0.0041,
      "step": 1093
    },
    {
      "epoch": 1.7504,
      "grad_norm": 0.3321770414037301,
      "learning_rate": 7.856697482465195e-06,
      "loss": 0.0058,
      "step": 1094
    },
    {
      "epoch": 1.752,
      "grad_norm": 0.32220117094247036,
      "learning_rate": 7.839821780235168e-06,
      "loss": 0.0034,
      "step": 1095
    },
    {
      "epoch": 1.7536,
      "grad_norm": 0.6863037206004958,
      "learning_rate": 7.822952528625192e-06,
      "loss": 0.0088,
      "step": 1096
    },
    {
      "epoch": 1.7551999999999999,
      "grad_norm": 0.24408042299840918,
      "learning_rate": 7.806089778009421e-06,
      "loss": 0.0049,
      "step": 1097
    },
    {
      "epoch": 1.7568000000000001,
      "grad_norm": 0.5797822426576273,
      "learning_rate": 7.789233578742583e-06,
      "loss": 0.0089,
      "step": 1098
    },
    {
      "epoch": 1.7584,
      "grad_norm": 0.5290109447804685,
      "learning_rate": 7.77238398115985e-06,
      "loss": 0.0109,
      "step": 1099
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.41951629898132015,
      "learning_rate": 7.755541035576677e-06,
      "loss": 0.0058,
      "step": 1100
    },
    {
      "epoch": 1.7616,
      "grad_norm": 0.4877789394560355,
      "learning_rate": 7.738704792288654e-06,
      "loss": 0.0059,
      "step": 1101
    },
    {
      "epoch": 1.7631999999999999,
      "grad_norm": 0.5335087112253809,
      "learning_rate": 7.721875301571359e-06,
      "loss": 0.0097,
      "step": 1102
    },
    {
      "epoch": 1.7648000000000001,
      "grad_norm": 0.47224661389922745,
      "learning_rate": 7.705052613680212e-06,
      "loss": 0.0118,
      "step": 1103
    },
    {
      "epoch": 1.7664,
      "grad_norm": 0.41273094189268117,
      "learning_rate": 7.688236778850307e-06,
      "loss": 0.0058,
      "step": 1104
    },
    {
      "epoch": 1.768,
      "grad_norm": 0.6097866252752729,
      "learning_rate": 7.671427847296274e-06,
      "loss": 0.0084,
      "step": 1105
    },
    {
      "epoch": 1.7696,
      "grad_norm": 0.39561876270305457,
      "learning_rate": 7.654625869212147e-06,
      "loss": 0.0069,
      "step": 1106
    },
    {
      "epoch": 1.7711999999999999,
      "grad_norm": 0.5717886178279439,
      "learning_rate": 7.637830894771176e-06,
      "loss": 0.0089,
      "step": 1107
    },
    {
      "epoch": 1.7728000000000002,
      "grad_norm": 0.5155788333858049,
      "learning_rate": 7.621042974125701e-06,
      "loss": 0.0058,
      "step": 1108
    },
    {
      "epoch": 1.7744,
      "grad_norm": 0.3477078857840802,
      "learning_rate": 7.604262157407008e-06,
      "loss": 0.0039,
      "step": 1109
    },
    {
      "epoch": 1.776,
      "grad_norm": 0.23435713399109273,
      "learning_rate": 7.587488494725157e-06,
      "loss": 0.0052,
      "step": 1110
    },
    {
      "epoch": 1.7776,
      "grad_norm": 0.5578341526611292,
      "learning_rate": 7.570722036168855e-06,
      "loss": 0.0098,
      "step": 1111
    },
    {
      "epoch": 1.7792,
      "grad_norm": 0.655412385837667,
      "learning_rate": 7.553962831805291e-06,
      "loss": 0.0085,
      "step": 1112
    },
    {
      "epoch": 1.7808000000000002,
      "grad_norm": 0.5802910339710066,
      "learning_rate": 7.537210931679988e-06,
      "loss": 0.0076,
      "step": 1113
    },
    {
      "epoch": 1.7824,
      "grad_norm": 0.6665626528983685,
      "learning_rate": 7.520466385816672e-06,
      "loss": 0.0087,
      "step": 1114
    },
    {
      "epoch": 1.784,
      "grad_norm": 0.33197254843442225,
      "learning_rate": 7.5037292442170865e-06,
      "loss": 0.0039,
      "step": 1115
    },
    {
      "epoch": 1.7856,
      "grad_norm": 0.5695030088373062,
      "learning_rate": 7.48699955686089e-06,
      "loss": 0.0098,
      "step": 1116
    },
    {
      "epoch": 1.7872,
      "grad_norm": 0.33307819834062863,
      "learning_rate": 7.470277373705461e-06,
      "loss": 0.0067,
      "step": 1117
    },
    {
      "epoch": 1.7888,
      "grad_norm": 0.47288221727470514,
      "learning_rate": 7.453562744685779e-06,
      "loss": 0.005,
      "step": 1118
    },
    {
      "epoch": 1.7904,
      "grad_norm": 0.3960654098825163,
      "learning_rate": 7.4368557197142596e-06,
      "loss": 0.0065,
      "step": 1119
    },
    {
      "epoch": 1.792,
      "grad_norm": 0.3449626116345571,
      "learning_rate": 7.420156348680621e-06,
      "loss": 0.0067,
      "step": 1120
    },
    {
      "epoch": 1.7936,
      "grad_norm": 0.4244578336020816,
      "learning_rate": 7.4034646814517155e-06,
      "loss": 0.0065,
      "step": 1121
    },
    {
      "epoch": 1.7952,
      "grad_norm": 0.4169819548917528,
      "learning_rate": 7.3867807678713965e-06,
      "loss": 0.0069,
      "step": 1122
    },
    {
      "epoch": 1.7968,
      "grad_norm": 0.32289014794153065,
      "learning_rate": 7.3701046577603605e-06,
      "loss": 0.0085,
      "step": 1123
    },
    {
      "epoch": 1.7984,
      "grad_norm": 0.472551606096138,
      "learning_rate": 7.353436400916006e-06,
      "loss": 0.0063,
      "step": 1124
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.33829886594761394,
      "learning_rate": 7.336776047112277e-06,
      "loss": 0.0045,
      "step": 1125
    },
    {
      "epoch": 1.8016,
      "grad_norm": 0.479782119071177,
      "learning_rate": 7.32012364609952e-06,
      "loss": 0.0084,
      "step": 1126
    },
    {
      "epoch": 1.8032,
      "grad_norm": 0.2724659089240648,
      "learning_rate": 7.303479247604333e-06,
      "loss": 0.0036,
      "step": 1127
    },
    {
      "epoch": 1.8048,
      "grad_norm": 0.3327499456295506,
      "learning_rate": 7.286842901329413e-06,
      "loss": 0.0054,
      "step": 1128
    },
    {
      "epoch": 1.8064,
      "grad_norm": 0.33234081270148413,
      "learning_rate": 7.270214656953415e-06,
      "loss": 0.0047,
      "step": 1129
    },
    {
      "epoch": 1.808,
      "grad_norm": 0.5063193642113647,
      "learning_rate": 7.253594564130804e-06,
      "loss": 0.0067,
      "step": 1130
    },
    {
      "epoch": 1.8096,
      "grad_norm": 0.24443282410498932,
      "learning_rate": 7.236982672491699e-06,
      "loss": 0.0056,
      "step": 1131
    },
    {
      "epoch": 1.8112,
      "grad_norm": 0.47491223334505844,
      "learning_rate": 7.22037903164173e-06,
      "loss": 0.0056,
      "step": 1132
    },
    {
      "epoch": 1.8128,
      "grad_norm": 0.190968013316911,
      "learning_rate": 7.203783691161883e-06,
      "loss": 0.0043,
      "step": 1133
    },
    {
      "epoch": 1.8144,
      "grad_norm": 0.5665726247730363,
      "learning_rate": 7.187196700608373e-06,
      "loss": 0.0079,
      "step": 1134
    },
    {
      "epoch": 1.8159999999999998,
      "grad_norm": 0.3578735453323942,
      "learning_rate": 7.170618109512465e-06,
      "loss": 0.0074,
      "step": 1135
    },
    {
      "epoch": 1.8176,
      "grad_norm": 0.26292486166299694,
      "learning_rate": 7.154047967380353e-06,
      "loss": 0.0055,
      "step": 1136
    },
    {
      "epoch": 1.8192,
      "grad_norm": 0.3287946223540181,
      "learning_rate": 7.137486323692994e-06,
      "loss": 0.0045,
      "step": 1137
    },
    {
      "epoch": 1.8208,
      "grad_norm": 0.4821439032277905,
      "learning_rate": 7.120933227905971e-06,
      "loss": 0.0151,
      "step": 1138
    },
    {
      "epoch": 1.8224,
      "grad_norm": 0.34486469359537525,
      "learning_rate": 7.104388729449338e-06,
      "loss": 0.006,
      "step": 1139
    },
    {
      "epoch": 1.8239999999999998,
      "grad_norm": 0.2618247231236417,
      "learning_rate": 7.0878528777274814e-06,
      "loss": 0.0051,
      "step": 1140
    },
    {
      "epoch": 1.8256000000000001,
      "grad_norm": 0.4052418483681234,
      "learning_rate": 7.0713257221189635e-06,
      "loss": 0.0057,
      "step": 1141
    },
    {
      "epoch": 1.8272,
      "grad_norm": 0.45680366105962966,
      "learning_rate": 7.05480731197638e-06,
      "loss": 0.0053,
      "step": 1142
    },
    {
      "epoch": 1.8288,
      "grad_norm": 0.3486466160337022,
      "learning_rate": 7.0382976966262065e-06,
      "loss": 0.0061,
      "step": 1143
    },
    {
      "epoch": 1.8304,
      "grad_norm": 0.18148153052257054,
      "learning_rate": 7.021796925368667e-06,
      "loss": 0.0035,
      "step": 1144
    },
    {
      "epoch": 1.8319999999999999,
      "grad_norm": 0.30417352075414195,
      "learning_rate": 7.005305047477566e-06,
      "loss": 0.0073,
      "step": 1145
    },
    {
      "epoch": 1.8336000000000001,
      "grad_norm": 0.3316981633490078,
      "learning_rate": 6.988822112200157e-06,
      "loss": 0.0073,
      "step": 1146
    },
    {
      "epoch": 1.8352,
      "grad_norm": 0.3646192862545398,
      "learning_rate": 6.9723481687569836e-06,
      "loss": 0.0042,
      "step": 1147
    },
    {
      "epoch": 1.8368,
      "grad_norm": 0.6013002629062065,
      "learning_rate": 6.955883266341741e-06,
      "loss": 0.0042,
      "step": 1148
    },
    {
      "epoch": 1.8384,
      "grad_norm": 0.29652681955969035,
      "learning_rate": 6.939427454121128e-06,
      "loss": 0.0061,
      "step": 1149
    },
    {
      "epoch": 1.8399999999999999,
      "grad_norm": 0.34240071160797086,
      "learning_rate": 6.9229807812346985e-06,
      "loss": 0.0096,
      "step": 1150
    },
    {
      "epoch": 1.8416000000000001,
      "grad_norm": 0.17463564552842203,
      "learning_rate": 6.9065432967947145e-06,
      "loss": 0.004,
      "step": 1151
    },
    {
      "epoch": 1.8432,
      "grad_norm": 0.38200139192617427,
      "learning_rate": 6.890115049885995e-06,
      "loss": 0.0077,
      "step": 1152
    },
    {
      "epoch": 1.8448,
      "grad_norm": 0.4949151787689372,
      "learning_rate": 6.8736960895657854e-06,
      "loss": 0.0108,
      "step": 1153
    },
    {
      "epoch": 1.8464,
      "grad_norm": 0.27918929940467996,
      "learning_rate": 6.85728646486359e-06,
      "loss": 0.0042,
      "step": 1154
    },
    {
      "epoch": 1.8479999999999999,
      "grad_norm": 0.2761223100165974,
      "learning_rate": 6.840886224781039e-06,
      "loss": 0.0067,
      "step": 1155
    },
    {
      "epoch": 1.8496000000000001,
      "grad_norm": 0.3091314244102595,
      "learning_rate": 6.824495418291741e-06,
      "loss": 0.008,
      "step": 1156
    },
    {
      "epoch": 1.8512,
      "grad_norm": 0.4653825842474054,
      "learning_rate": 6.8081140943411296e-06,
      "loss": 0.0072,
      "step": 1157
    },
    {
      "epoch": 1.8528,
      "grad_norm": 0.31221517262585574,
      "learning_rate": 6.791742301846325e-06,
      "loss": 0.0059,
      "step": 1158
    },
    {
      "epoch": 1.8544,
      "grad_norm": 0.36726590327954306,
      "learning_rate": 6.775380089695986e-06,
      "loss": 0.0046,
      "step": 1159
    },
    {
      "epoch": 1.8559999999999999,
      "grad_norm": 0.3242158265475158,
      "learning_rate": 6.759027506750159e-06,
      "loss": 0.0064,
      "step": 1160
    },
    {
      "epoch": 1.8576000000000001,
      "grad_norm": 0.31247436508664383,
      "learning_rate": 6.742684601840142e-06,
      "loss": 0.0049,
      "step": 1161
    },
    {
      "epoch": 1.8592,
      "grad_norm": 0.20875497548440738,
      "learning_rate": 6.726351423768323e-06,
      "loss": 0.0027,
      "step": 1162
    },
    {
      "epoch": 1.8608,
      "grad_norm": 0.33207426145196084,
      "learning_rate": 6.710028021308061e-06,
      "loss": 0.0049,
      "step": 1163
    },
    {
      "epoch": 1.8624,
      "grad_norm": 0.40260970705133703,
      "learning_rate": 6.693714443203507e-06,
      "loss": 0.0047,
      "step": 1164
    },
    {
      "epoch": 1.8639999999999999,
      "grad_norm": 0.30987382333167335,
      "learning_rate": 6.677410738169485e-06,
      "loss": 0.0056,
      "step": 1165
    },
    {
      "epoch": 1.8656000000000001,
      "grad_norm": 0.4130657667581579,
      "learning_rate": 6.661116954891329e-06,
      "loss": 0.0055,
      "step": 1166
    },
    {
      "epoch": 1.8672,
      "grad_norm": 0.32794991888696495,
      "learning_rate": 6.644833142024752e-06,
      "loss": 0.0042,
      "step": 1167
    },
    {
      "epoch": 1.8688,
      "grad_norm": 0.522406636948986,
      "learning_rate": 6.62855934819569e-06,
      "loss": 0.0065,
      "step": 1168
    },
    {
      "epoch": 1.8704,
      "grad_norm": 0.3902449243177916,
      "learning_rate": 6.612295622000162e-06,
      "loss": 0.0088,
      "step": 1169
    },
    {
      "epoch": 1.8719999999999999,
      "grad_norm": 0.213505902046021,
      "learning_rate": 6.59604201200412e-06,
      "loss": 0.005,
      "step": 1170
    },
    {
      "epoch": 1.8736000000000002,
      "grad_norm": 0.2757910385921279,
      "learning_rate": 6.579798566743314e-06,
      "loss": 0.006,
      "step": 1171
    },
    {
      "epoch": 1.8752,
      "grad_norm": 0.2563207989287682,
      "learning_rate": 6.563565334723134e-06,
      "loss": 0.004,
      "step": 1172
    },
    {
      "epoch": 1.8768,
      "grad_norm": 0.3429964571294477,
      "learning_rate": 6.547342364418482e-06,
      "loss": 0.0047,
      "step": 1173
    },
    {
      "epoch": 1.8784,
      "grad_norm": 0.17895805114654723,
      "learning_rate": 6.5311297042736046e-06,
      "loss": 0.0031,
      "step": 1174
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.3836987346654857,
      "learning_rate": 6.514927402701965e-06,
      "loss": 0.0063,
      "step": 1175
    },
    {
      "epoch": 1.8816000000000002,
      "grad_norm": 0.45059071335600803,
      "learning_rate": 6.498735508086094e-06,
      "loss": 0.008,
      "step": 1176
    },
    {
      "epoch": 1.8832,
      "grad_norm": 0.33214015016144555,
      "learning_rate": 6.482554068777451e-06,
      "loss": 0.0042,
      "step": 1177
    },
    {
      "epoch": 1.8848,
      "grad_norm": 0.4595121731217701,
      "learning_rate": 6.466383133096268e-06,
      "loss": 0.0063,
      "step": 1178
    },
    {
      "epoch": 1.8864,
      "grad_norm": 0.42499357639905216,
      "learning_rate": 6.450222749331414e-06,
      "loss": 0.0063,
      "step": 1179
    },
    {
      "epoch": 1.888,
      "grad_norm": 0.3630709788454369,
      "learning_rate": 6.4340729657402424e-06,
      "loss": 0.0056,
      "step": 1180
    },
    {
      "epoch": 1.8896,
      "grad_norm": 0.32849173455121033,
      "learning_rate": 6.4179338305484675e-06,
      "loss": 0.0042,
      "step": 1181
    },
    {
      "epoch": 1.8912,
      "grad_norm": 0.47697285205391693,
      "learning_rate": 6.40180539194999e-06,
      "loss": 0.0053,
      "step": 1182
    },
    {
      "epoch": 1.8928,
      "grad_norm": 0.214089479958329,
      "learning_rate": 6.385687698106781e-06,
      "loss": 0.004,
      "step": 1183
    },
    {
      "epoch": 1.8944,
      "grad_norm": 0.18666853565973923,
      "learning_rate": 6.3695807971487175e-06,
      "loss": 0.0044,
      "step": 1184
    },
    {
      "epoch": 1.896,
      "grad_norm": 0.37642768242623026,
      "learning_rate": 6.35348473717345e-06,
      "loss": 0.0063,
      "step": 1185
    },
    {
      "epoch": 1.8976,
      "grad_norm": 0.27832937638872407,
      "learning_rate": 6.337399566246257e-06,
      "loss": 0.0073,
      "step": 1186
    },
    {
      "epoch": 1.8992,
      "grad_norm": 0.3143922382058477,
      "learning_rate": 6.321325332399904e-06,
      "loss": 0.0034,
      "step": 1187
    },
    {
      "epoch": 1.9008,
      "grad_norm": 0.33682375227488676,
      "learning_rate": 6.305262083634488e-06,
      "loss": 0.0039,
      "step": 1188
    },
    {
      "epoch": 1.9024,
      "grad_norm": 0.4044456913665351,
      "learning_rate": 6.289209867917312e-06,
      "loss": 0.0059,
      "step": 1189
    },
    {
      "epoch": 1.904,
      "grad_norm": 0.6500081303210257,
      "learning_rate": 6.2731687331827214e-06,
      "loss": 0.0068,
      "step": 1190
    },
    {
      "epoch": 1.9056,
      "grad_norm": 0.5839569583167884,
      "learning_rate": 6.2571387273319905e-06,
      "loss": 0.0072,
      "step": 1191
    },
    {
      "epoch": 1.9072,
      "grad_norm": 0.29813248946273585,
      "learning_rate": 6.2411198982331435e-06,
      "loss": 0.0059,
      "step": 1192
    },
    {
      "epoch": 1.9088,
      "grad_norm": 0.395552190000539,
      "learning_rate": 6.225112293720836e-06,
      "loss": 0.0056,
      "step": 1193
    },
    {
      "epoch": 1.9104,
      "grad_norm": 0.28649578761298766,
      "learning_rate": 6.209115961596208e-06,
      "loss": 0.0032,
      "step": 1194
    },
    {
      "epoch": 1.912,
      "grad_norm": 0.33399463181181227,
      "learning_rate": 6.193130949626731e-06,
      "loss": 0.0076,
      "step": 1195
    },
    {
      "epoch": 1.9136,
      "grad_norm": 0.31191182808084145,
      "learning_rate": 6.177157305546077e-06,
      "loss": 0.006,
      "step": 1196
    },
    {
      "epoch": 1.9152,
      "grad_norm": 0.28527792964216175,
      "learning_rate": 6.1611950770539766e-06,
      "loss": 0.0046,
      "step": 1197
    },
    {
      "epoch": 1.9167999999999998,
      "grad_norm": 0.4323818678849387,
      "learning_rate": 6.145244311816063e-06,
      "loss": 0.0155,
      "step": 1198
    },
    {
      "epoch": 1.9184,
      "grad_norm": 0.26984275027804744,
      "learning_rate": 6.129305057463741e-06,
      "loss": 0.0039,
      "step": 1199
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.3258817250599572,
      "learning_rate": 6.113377361594048e-06,
      "loss": 0.0066,
      "step": 1200
    },
    {
      "epoch": 1.9216,
      "grad_norm": 0.40964896634709624,
      "learning_rate": 6.0974612717695e-06,
      "loss": 0.0054,
      "step": 1201
    },
    {
      "epoch": 1.9232,
      "grad_norm": 0.36917883995698975,
      "learning_rate": 6.081556835517955e-06,
      "loss": 0.0052,
      "step": 1202
    },
    {
      "epoch": 1.9247999999999998,
      "grad_norm": 0.2358165396682365,
      "learning_rate": 6.065664100332478e-06,
      "loss": 0.0037,
      "step": 1203
    },
    {
      "epoch": 1.9264000000000001,
      "grad_norm": 0.36764085459176077,
      "learning_rate": 6.049783113671184e-06,
      "loss": 0.0043,
      "step": 1204
    },
    {
      "epoch": 1.928,
      "grad_norm": 0.41252331455261154,
      "learning_rate": 6.033913922957112e-06,
      "loss": 0.0043,
      "step": 1205
    },
    {
      "epoch": 1.9296,
      "grad_norm": 0.18441031104306468,
      "learning_rate": 6.018056575578075e-06,
      "loss": 0.0033,
      "step": 1206
    },
    {
      "epoch": 1.9312,
      "grad_norm": 0.22195801965442705,
      "learning_rate": 6.002211118886514e-06,
      "loss": 0.0027,
      "step": 1207
    },
    {
      "epoch": 1.9327999999999999,
      "grad_norm": 0.4037529762288358,
      "learning_rate": 5.986377600199371e-06,
      "loss": 0.0064,
      "step": 1208
    },
    {
      "epoch": 1.9344000000000001,
      "grad_norm": 0.3270423790951842,
      "learning_rate": 5.970556066797941e-06,
      "loss": 0.0055,
      "step": 1209
    },
    {
      "epoch": 1.936,
      "grad_norm": 0.3849544612859768,
      "learning_rate": 5.9547465659277215e-06,
      "loss": 0.005,
      "step": 1210
    },
    {
      "epoch": 1.9376,
      "grad_norm": 0.19885859640069262,
      "learning_rate": 5.93894914479828e-06,
      "loss": 0.004,
      "step": 1211
    },
    {
      "epoch": 1.9392,
      "grad_norm": 0.2511154473625223,
      "learning_rate": 5.923163850583114e-06,
      "loss": 0.0032,
      "step": 1212
    },
    {
      "epoch": 1.9407999999999999,
      "grad_norm": 0.3459117674365608,
      "learning_rate": 5.907390730419506e-06,
      "loss": 0.0065,
      "step": 1213
    },
    {
      "epoch": 1.9424000000000001,
      "grad_norm": 0.3154974281342526,
      "learning_rate": 5.891629831408392e-06,
      "loss": 0.0037,
      "step": 1214
    },
    {
      "epoch": 1.944,
      "grad_norm": 0.35432322451329235,
      "learning_rate": 5.875881200614208e-06,
      "loss": 0.0057,
      "step": 1215
    },
    {
      "epoch": 1.9456,
      "grad_norm": 0.27041268208410507,
      "learning_rate": 5.8601448850647515e-06,
      "loss": 0.0047,
      "step": 1216
    },
    {
      "epoch": 1.9472,
      "grad_norm": 0.571194663835566,
      "learning_rate": 5.8444209317510515e-06,
      "loss": 0.0061,
      "step": 1217
    },
    {
      "epoch": 1.9487999999999999,
      "grad_norm": 0.6282359016961699,
      "learning_rate": 5.828709387627219e-06,
      "loss": 0.0127,
      "step": 1218
    },
    {
      "epoch": 1.9504000000000001,
      "grad_norm": 0.34402025436940703,
      "learning_rate": 5.813010299610313e-06,
      "loss": 0.0084,
      "step": 1219
    },
    {
      "epoch": 1.952,
      "grad_norm": 0.3480649745220345,
      "learning_rate": 5.797323714580192e-06,
      "loss": 0.0039,
      "step": 1220
    },
    {
      "epoch": 1.9536,
      "grad_norm": 0.4691213849698909,
      "learning_rate": 5.781649679379379e-06,
      "loss": 0.0059,
      "step": 1221
    },
    {
      "epoch": 1.9552,
      "grad_norm": 0.21565957592354734,
      "learning_rate": 5.7659882408129204e-06,
      "loss": 0.0038,
      "step": 1222
    },
    {
      "epoch": 1.9567999999999999,
      "grad_norm": 0.44118541711284726,
      "learning_rate": 5.750339445648252e-06,
      "loss": 0.0094,
      "step": 1223
    },
    {
      "epoch": 1.9584000000000001,
      "grad_norm": 0.4957908051911234,
      "learning_rate": 5.7347033406150494e-06,
      "loss": 0.0072,
      "step": 1224
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.2937561406067032,
      "learning_rate": 5.7190799724050924e-06,
      "loss": 0.0038,
      "step": 1225
    },
    {
      "epoch": 1.9616,
      "grad_norm": 0.7194650498636356,
      "learning_rate": 5.703469387672138e-06,
      "loss": 0.0071,
      "step": 1226
    },
    {
      "epoch": 1.9632,
      "grad_norm": 0.4604277229426392,
      "learning_rate": 5.687871633031754e-06,
      "loss": 0.0062,
      "step": 1227
    },
    {
      "epoch": 1.9647999999999999,
      "grad_norm": 0.5455836140271865,
      "learning_rate": 5.672286755061212e-06,
      "loss": 0.0087,
      "step": 1228
    },
    {
      "epoch": 1.9664000000000001,
      "grad_norm": 0.4672016066833358,
      "learning_rate": 5.656714800299317e-06,
      "loss": 0.0072,
      "step": 1229
    },
    {
      "epoch": 1.968,
      "grad_norm": 0.21241233709878102,
      "learning_rate": 5.64115581524629e-06,
      "loss": 0.0053,
      "step": 1230
    },
    {
      "epoch": 1.9696,
      "grad_norm": 0.34005297435351134,
      "learning_rate": 5.625609846363622e-06,
      "loss": 0.003,
      "step": 1231
    },
    {
      "epoch": 1.9712,
      "grad_norm": 0.42052160854254317,
      "learning_rate": 5.610076940073939e-06,
      "loss": 0.0069,
      "step": 1232
    },
    {
      "epoch": 1.9727999999999999,
      "grad_norm": 0.3944203490898746,
      "learning_rate": 5.594557142760853e-06,
      "loss": 0.0048,
      "step": 1233
    },
    {
      "epoch": 1.9744000000000002,
      "grad_norm": 0.21058513543549798,
      "learning_rate": 5.579050500768837e-06,
      "loss": 0.0047,
      "step": 1234
    },
    {
      "epoch": 1.976,
      "grad_norm": 0.26725426932475405,
      "learning_rate": 5.563557060403071e-06,
      "loss": 0.0037,
      "step": 1235
    },
    {
      "epoch": 1.9776,
      "grad_norm": 0.33849781411866225,
      "learning_rate": 5.548076867929331e-06,
      "loss": 0.0067,
      "step": 1236
    },
    {
      "epoch": 1.9792,
      "grad_norm": 0.29629666107985536,
      "learning_rate": 5.53260996957381e-06,
      "loss": 0.0062,
      "step": 1237
    },
    {
      "epoch": 1.9808,
      "grad_norm": 0.42531169332494806,
      "learning_rate": 5.517156411523026e-06,
      "loss": 0.0065,
      "step": 1238
    },
    {
      "epoch": 1.9824000000000002,
      "grad_norm": 0.4259366912829729,
      "learning_rate": 5.501716239923642e-06,
      "loss": 0.0072,
      "step": 1239
    },
    {
      "epoch": 1.984,
      "grad_norm": 0.6024658910530383,
      "learning_rate": 5.486289500882355e-06,
      "loss": 0.0132,
      "step": 1240
    },
    {
      "epoch": 1.9856,
      "grad_norm": 0.30824009184805684,
      "learning_rate": 5.47087624046575e-06,
      "loss": 0.004,
      "step": 1241
    },
    {
      "epoch": 1.9872,
      "grad_norm": 0.22725954434890408,
      "learning_rate": 5.455476504700161e-06,
      "loss": 0.0042,
      "step": 1242
    },
    {
      "epoch": 1.9888,
      "grad_norm": 0.4197519865595487,
      "learning_rate": 5.440090339571537e-06,
      "loss": 0.0064,
      "step": 1243
    },
    {
      "epoch": 1.9904,
      "grad_norm": 0.44957031880341963,
      "learning_rate": 5.424717791025302e-06,
      "loss": 0.0078,
      "step": 1244
    },
    {
      "epoch": 1.992,
      "grad_norm": 0.35123516194158083,
      "learning_rate": 5.4093589049662175e-06,
      "loss": 0.0073,
      "step": 1245
    },
    {
      "epoch": 1.9936,
      "grad_norm": 0.282971493471759,
      "learning_rate": 5.3940137272582534e-06,
      "loss": 0.0052,
      "step": 1246
    },
    {
      "epoch": 1.9952,
      "grad_norm": 0.38016149664339494,
      "learning_rate": 5.378682303724435e-06,
      "loss": 0.0038,
      "step": 1247
    },
    {
      "epoch": 1.9968,
      "grad_norm": 0.5098739325476022,
      "learning_rate": 5.3633646801467255e-06,
      "loss": 0.007,
      "step": 1248
    },
    {
      "epoch": 1.9984,
      "grad_norm": 0.6140692262656893,
      "learning_rate": 5.348060902265871e-06,
      "loss": 0.0075,
      "step": 1249
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.3501279716540246,
      "learning_rate": 5.332771015781275e-06,
      "loss": 0.0064,
      "step": 1250
    },
    {
      "epoch": 2.0016,
      "grad_norm": 0.27946530367808153,
      "learning_rate": 5.31749506635086e-06,
      "loss": 0.0064,
      "step": 1251
    },
    {
      "epoch": 2.0032,
      "grad_norm": 0.30440282450736217,
      "learning_rate": 5.302233099590928e-06,
      "loss": 0.0046,
      "step": 1252
    },
    {
      "epoch": 2.0048,
      "grad_norm": 0.44397766833706465,
      "learning_rate": 5.286985161076029e-06,
      "loss": 0.0062,
      "step": 1253
    },
    {
      "epoch": 2.0064,
      "grad_norm": 0.478560493359205,
      "learning_rate": 5.271751296338823e-06,
      "loss": 0.0061,
      "step": 1254
    },
    {
      "epoch": 2.008,
      "grad_norm": 0.6615866501216844,
      "learning_rate": 5.2565315508699374e-06,
      "loss": 0.0063,
      "step": 1255
    },
    {
      "epoch": 2.0096,
      "grad_norm": 0.2802755002339857,
      "learning_rate": 5.241325970117851e-06,
      "loss": 0.0042,
      "step": 1256
    },
    {
      "epoch": 2.0112,
      "grad_norm": 0.4109368046645932,
      "learning_rate": 5.226134599488728e-06,
      "loss": 0.0053,
      "step": 1257
    },
    {
      "epoch": 2.0128,
      "grad_norm": 0.3481720552299007,
      "learning_rate": 5.210957484346314e-06,
      "loss": 0.0082,
      "step": 1258
    },
    {
      "epoch": 2.0144,
      "grad_norm": 0.3029754374591977,
      "learning_rate": 5.195794670011775e-06,
      "loss": 0.0062,
      "step": 1259
    },
    {
      "epoch": 2.016,
      "grad_norm": 0.5502723232435336,
      "learning_rate": 5.1806462017635775e-06,
      "loss": 0.0052,
      "step": 1260
    },
    {
      "epoch": 2.0176,
      "grad_norm": 0.330718899882051,
      "learning_rate": 5.165512124837344e-06,
      "loss": 0.007,
      "step": 1261
    },
    {
      "epoch": 2.0192,
      "grad_norm": 0.48764245936206446,
      "learning_rate": 5.150392484425728e-06,
      "loss": 0.0049,
      "step": 1262
    },
    {
      "epoch": 2.0208,
      "grad_norm": 0.3342124162357661,
      "learning_rate": 5.135287325678271e-06,
      "loss": 0.0066,
      "step": 1263
    },
    {
      "epoch": 2.0224,
      "grad_norm": 0.4202488105628078,
      "learning_rate": 5.120196693701267e-06,
      "loss": 0.0042,
      "step": 1264
    },
    {
      "epoch": 2.024,
      "grad_norm": 0.40577032886305797,
      "learning_rate": 5.105120633557634e-06,
      "loss": 0.0072,
      "step": 1265
    },
    {
      "epoch": 2.0256,
      "grad_norm": 0.26383658827264916,
      "learning_rate": 5.090059190266779e-06,
      "loss": 0.0036,
      "step": 1266
    },
    {
      "epoch": 2.0272,
      "grad_norm": 0.37091203157656494,
      "learning_rate": 5.075012408804458e-06,
      "loss": 0.0034,
      "step": 1267
    },
    {
      "epoch": 2.0288,
      "grad_norm": 0.353219243476177,
      "learning_rate": 5.059980334102637e-06,
      "loss": 0.0047,
      "step": 1268
    },
    {
      "epoch": 2.0304,
      "grad_norm": 0.23227630423010162,
      "learning_rate": 5.044963011049384e-06,
      "loss": 0.0025,
      "step": 1269
    },
    {
      "epoch": 2.032,
      "grad_norm": 0.2737992961702637,
      "learning_rate": 5.0299604844886985e-06,
      "loss": 0.0058,
      "step": 1270
    },
    {
      "epoch": 2.0336,
      "grad_norm": 0.49954303192035526,
      "learning_rate": 5.0149727992204034e-06,
      "loss": 0.0058,
      "step": 1271
    },
    {
      "epoch": 2.0352,
      "grad_norm": 0.34112852678238753,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.0068,
      "step": 1272
    },
    {
      "epoch": 2.0368,
      "grad_norm": 0.5429394939769813,
      "learning_rate": 4.985042131538545e-06,
      "loss": 0.008,
      "step": 1273
    },
    {
      "epoch": 2.0384,
      "grad_norm": 0.35273691884209735,
      "learning_rate": 4.970099238502494e-06,
      "loss": 0.0045,
      "step": 1274
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.41262386203301665,
      "learning_rate": 4.955171365513603e-06,
      "loss": 0.0057,
      "step": 1275
    },
    {
      "epoch": 2.0416,
      "grad_norm": 0.261370128060724,
      "learning_rate": 4.940258557148765e-06,
      "loss": 0.0054,
      "step": 1276
    },
    {
      "epoch": 2.0432,
      "grad_norm": 0.627636888025991,
      "learning_rate": 4.925360857939886e-06,
      "loss": 0.0073,
      "step": 1277
    },
    {
      "epoch": 2.0448,
      "grad_norm": 0.2226146536129725,
      "learning_rate": 4.910478312373757e-06,
      "loss": 0.0035,
      "step": 1278
    },
    {
      "epoch": 2.0464,
      "grad_norm": 0.2127764817558698,
      "learning_rate": 4.895610964891923e-06,
      "loss": 0.0036,
      "step": 1279
    },
    {
      "epoch": 2.048,
      "grad_norm": 0.25214117233057504,
      "learning_rate": 4.8807588598905364e-06,
      "loss": 0.0037,
      "step": 1280
    },
    {
      "epoch": 2.0496,
      "grad_norm": 0.31500180906439734,
      "learning_rate": 4.865922041720239e-06,
      "loss": 0.0067,
      "step": 1281
    },
    {
      "epoch": 2.0512,
      "grad_norm": 0.2577048344610542,
      "learning_rate": 4.8511005546860214e-06,
      "loss": 0.005,
      "step": 1282
    },
    {
      "epoch": 2.0528,
      "grad_norm": 0.25911780144857743,
      "learning_rate": 4.836294443047088e-06,
      "loss": 0.0043,
      "step": 1283
    },
    {
      "epoch": 2.0544,
      "grad_norm": 0.24808506280824796,
      "learning_rate": 4.821503751016746e-06,
      "loss": 0.004,
      "step": 1284
    },
    {
      "epoch": 2.056,
      "grad_norm": 0.19062806653577596,
      "learning_rate": 4.806728522762241e-06,
      "loss": 0.0042,
      "step": 1285
    },
    {
      "epoch": 2.0576,
      "grad_norm": 0.1565566885639028,
      "learning_rate": 4.791968802404648e-06,
      "loss": 0.0035,
      "step": 1286
    },
    {
      "epoch": 2.0592,
      "grad_norm": 0.3309828031131677,
      "learning_rate": 4.777224634018732e-06,
      "loss": 0.006,
      "step": 1287
    },
    {
      "epoch": 2.0608,
      "grad_norm": 0.23958773308236364,
      "learning_rate": 4.762496061632814e-06,
      "loss": 0.0034,
      "step": 1288
    },
    {
      "epoch": 2.0624,
      "grad_norm": 0.23239980452297915,
      "learning_rate": 4.7477831292286555e-06,
      "loss": 0.0041,
      "step": 1289
    },
    {
      "epoch": 2.064,
      "grad_norm": 0.3107452959938814,
      "learning_rate": 4.733085880741301e-06,
      "loss": 0.0049,
      "step": 1290
    },
    {
      "epoch": 2.0656,
      "grad_norm": 0.21814879569393347,
      "learning_rate": 4.7184043600589655e-06,
      "loss": 0.0038,
      "step": 1291
    },
    {
      "epoch": 2.0672,
      "grad_norm": 0.3002218278383945,
      "learning_rate": 4.703738611022899e-06,
      "loss": 0.0038,
      "step": 1292
    },
    {
      "epoch": 2.0688,
      "grad_norm": 0.17988878791249202,
      "learning_rate": 4.689088677427249e-06,
      "loss": 0.0028,
      "step": 1293
    },
    {
      "epoch": 2.0704,
      "grad_norm": 0.1918665097047004,
      "learning_rate": 4.674454603018949e-06,
      "loss": 0.0029,
      "step": 1294
    },
    {
      "epoch": 2.072,
      "grad_norm": 0.2138417184341068,
      "learning_rate": 4.659836431497563e-06,
      "loss": 0.0039,
      "step": 1295
    },
    {
      "epoch": 2.0736,
      "grad_norm": 0.17185071863166368,
      "learning_rate": 4.645234206515171e-06,
      "loss": 0.0022,
      "step": 1296
    },
    {
      "epoch": 2.0752,
      "grad_norm": 0.3512312119574598,
      "learning_rate": 4.630647971676232e-06,
      "loss": 0.0051,
      "step": 1297
    },
    {
      "epoch": 2.0768,
      "grad_norm": 0.1967479613686531,
      "learning_rate": 4.616077770537453e-06,
      "loss": 0.0031,
      "step": 1298
    },
    {
      "epoch": 2.0784,
      "grad_norm": 0.30824071063703085,
      "learning_rate": 4.601523646607675e-06,
      "loss": 0.0034,
      "step": 1299
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.2695497026040452,
      "learning_rate": 4.586985643347716e-06,
      "loss": 0.0039,
      "step": 1300
    },
    {
      "epoch": 2.0816,
      "grad_norm": 0.4598031790299854,
      "learning_rate": 4.572463804170263e-06,
      "loss": 0.0042,
      "step": 1301
    },
    {
      "epoch": 2.0832,
      "grad_norm": 0.22011607876674072,
      "learning_rate": 4.557958172439726e-06,
      "loss": 0.0019,
      "step": 1302
    },
    {
      "epoch": 2.0848,
      "grad_norm": 0.3459979438376907,
      "learning_rate": 4.543468791472131e-06,
      "loss": 0.0034,
      "step": 1303
    },
    {
      "epoch": 2.0864,
      "grad_norm": 0.29612633919353937,
      "learning_rate": 4.5289957045349655e-06,
      "loss": 0.0032,
      "step": 1304
    },
    {
      "epoch": 2.088,
      "grad_norm": 0.33327682757677773,
      "learning_rate": 4.5145389548470645e-06,
      "loss": 0.0054,
      "step": 1305
    },
    {
      "epoch": 2.0896,
      "grad_norm": 0.2522281062899099,
      "learning_rate": 4.500098585578475e-06,
      "loss": 0.0035,
      "step": 1306
    },
    {
      "epoch": 2.0912,
      "grad_norm": 0.3154381874423506,
      "learning_rate": 4.485674639850334e-06,
      "loss": 0.0033,
      "step": 1307
    },
    {
      "epoch": 2.0928,
      "grad_norm": 0.7806681135321053,
      "learning_rate": 4.471267160734731e-06,
      "loss": 0.0082,
      "step": 1308
    },
    {
      "epoch": 2.0944,
      "grad_norm": 0.20924011580648477,
      "learning_rate": 4.456876191254582e-06,
      "loss": 0.002,
      "step": 1309
    },
    {
      "epoch": 2.096,
      "grad_norm": 0.27343581761958907,
      "learning_rate": 4.4425017743835155e-06,
      "loss": 0.0042,
      "step": 1310
    },
    {
      "epoch": 2.0976,
      "grad_norm": 0.43634599723703343,
      "learning_rate": 4.4281439530457174e-06,
      "loss": 0.0053,
      "step": 1311
    },
    {
      "epoch": 2.0992,
      "grad_norm": 0.27198385642671596,
      "learning_rate": 4.413802770115816e-06,
      "loss": 0.0033,
      "step": 1312
    },
    {
      "epoch": 2.1008,
      "grad_norm": 0.23484207052201267,
      "learning_rate": 4.399478268418771e-06,
      "loss": 0.0018,
      "step": 1313
    },
    {
      "epoch": 2.1024,
      "grad_norm": 0.17246406743748,
      "learning_rate": 4.385170490729712e-06,
      "loss": 0.0019,
      "step": 1314
    },
    {
      "epoch": 2.104,
      "grad_norm": 0.2028269985930605,
      "learning_rate": 4.370879479773837e-06,
      "loss": 0.0027,
      "step": 1315
    },
    {
      "epoch": 2.1056,
      "grad_norm": 0.26292786274161534,
      "learning_rate": 4.356605278226274e-06,
      "loss": 0.0029,
      "step": 1316
    },
    {
      "epoch": 2.1072,
      "grad_norm": 0.1867765717999834,
      "learning_rate": 4.342347928711953e-06,
      "loss": 0.0023,
      "step": 1317
    },
    {
      "epoch": 2.1088,
      "grad_norm": 0.5299678610445645,
      "learning_rate": 4.328107473805487e-06,
      "loss": 0.0067,
      "step": 1318
    },
    {
      "epoch": 2.1104,
      "grad_norm": 0.3684095285885515,
      "learning_rate": 4.313883956031031e-06,
      "loss": 0.006,
      "step": 1319
    },
    {
      "epoch": 2.112,
      "grad_norm": 0.21733768699759115,
      "learning_rate": 4.299677417862174e-06,
      "loss": 0.0024,
      "step": 1320
    },
    {
      "epoch": 2.1136,
      "grad_norm": 0.339185518787034,
      "learning_rate": 4.28548790172179e-06,
      "loss": 0.0038,
      "step": 1321
    },
    {
      "epoch": 2.1152,
      "grad_norm": 0.20615281771551228,
      "learning_rate": 4.2713154499819345e-06,
      "loss": 0.0027,
      "step": 1322
    },
    {
      "epoch": 2.1168,
      "grad_norm": 0.35306789971032443,
      "learning_rate": 4.257160104963695e-06,
      "loss": 0.0043,
      "step": 1323
    },
    {
      "epoch": 2.1184,
      "grad_norm": 0.42183599454548737,
      "learning_rate": 4.243021908937083e-06,
      "loss": 0.0051,
      "step": 1324
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.3939500329035106,
      "learning_rate": 4.228900904120895e-06,
      "loss": 0.0044,
      "step": 1325
    },
    {
      "epoch": 2.1216,
      "grad_norm": 0.6441545692827979,
      "learning_rate": 4.214797132682597e-06,
      "loss": 0.0057,
      "step": 1326
    },
    {
      "epoch": 2.1232,
      "grad_norm": 0.46152235798316077,
      "learning_rate": 4.200710636738189e-06,
      "loss": 0.0054,
      "step": 1327
    },
    {
      "epoch": 2.1248,
      "grad_norm": 0.40725285654205506,
      "learning_rate": 4.186641458352088e-06,
      "loss": 0.0059,
      "step": 1328
    },
    {
      "epoch": 2.1264,
      "grad_norm": 0.24332009358464296,
      "learning_rate": 4.172589639536992e-06,
      "loss": 0.0043,
      "step": 1329
    },
    {
      "epoch": 2.128,
      "grad_norm": 0.41815713704336355,
      "learning_rate": 4.158555222253772e-06,
      "loss": 0.0058,
      "step": 1330
    },
    {
      "epoch": 2.1296,
      "grad_norm": 0.4771775277273845,
      "learning_rate": 4.144538248411321e-06,
      "loss": 0.004,
      "step": 1331
    },
    {
      "epoch": 2.1312,
      "grad_norm": 0.33752406994186707,
      "learning_rate": 4.130538759866457e-06,
      "loss": 0.0059,
      "step": 1332
    },
    {
      "epoch": 2.1328,
      "grad_norm": 0.4575903127414404,
      "learning_rate": 4.116556798423776e-06,
      "loss": 0.0052,
      "step": 1333
    },
    {
      "epoch": 2.1344,
      "grad_norm": 0.23976033424310222,
      "learning_rate": 4.102592405835536e-06,
      "loss": 0.0045,
      "step": 1334
    },
    {
      "epoch": 2.136,
      "grad_norm": 0.4474444698419231,
      "learning_rate": 4.088645623801534e-06,
      "loss": 0.0046,
      "step": 1335
    },
    {
      "epoch": 2.1376,
      "grad_norm": 0.7501073711538025,
      "learning_rate": 4.074716493968976e-06,
      "loss": 0.01,
      "step": 1336
    },
    {
      "epoch": 2.1391999999999998,
      "grad_norm": 0.4176418804505741,
      "learning_rate": 4.060805057932359e-06,
      "loss": 0.008,
      "step": 1337
    },
    {
      "epoch": 2.1408,
      "grad_norm": 0.41124851299783816,
      "learning_rate": 4.046911357233343e-06,
      "loss": 0.005,
      "step": 1338
    },
    {
      "epoch": 2.1424,
      "grad_norm": 0.10633178217740394,
      "learning_rate": 4.033035433360624e-06,
      "loss": 0.0012,
      "step": 1339
    },
    {
      "epoch": 2.144,
      "grad_norm": 0.5906271452129809,
      "learning_rate": 4.019177327749822e-06,
      "loss": 0.0064,
      "step": 1340
    },
    {
      "epoch": 2.1456,
      "grad_norm": 0.42624283355835463,
      "learning_rate": 4.00533708178334e-06,
      "loss": 0.0062,
      "step": 1341
    },
    {
      "epoch": 2.1471999999999998,
      "grad_norm": 0.3947213443096026,
      "learning_rate": 3.991514736790259e-06,
      "loss": 0.0067,
      "step": 1342
    },
    {
      "epoch": 2.1488,
      "grad_norm": 0.2380767675553356,
      "learning_rate": 3.977710334046193e-06,
      "loss": 0.0028,
      "step": 1343
    },
    {
      "epoch": 2.1504,
      "grad_norm": 0.3204449051112425,
      "learning_rate": 3.9639239147731865e-06,
      "loss": 0.003,
      "step": 1344
    },
    {
      "epoch": 2.152,
      "grad_norm": 0.2398994875704844,
      "learning_rate": 3.950155520139581e-06,
      "loss": 0.005,
      "step": 1345
    },
    {
      "epoch": 2.1536,
      "grad_norm": 0.17312681016523704,
      "learning_rate": 3.936405191259891e-06,
      "loss": 0.0027,
      "step": 1346
    },
    {
      "epoch": 2.1552,
      "grad_norm": 0.5201426071877979,
      "learning_rate": 3.9226729691946865e-06,
      "loss": 0.0043,
      "step": 1347
    },
    {
      "epoch": 2.1568,
      "grad_norm": 0.26525210748144445,
      "learning_rate": 3.908958894950465e-06,
      "loss": 0.0038,
      "step": 1348
    },
    {
      "epoch": 2.1584,
      "grad_norm": 0.3930269687027248,
      "learning_rate": 3.895263009479534e-06,
      "loss": 0.0033,
      "step": 1349
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.41311565597600647,
      "learning_rate": 3.881585353679891e-06,
      "loss": 0.0055,
      "step": 1350
    },
    {
      "epoch": 2.1616,
      "grad_norm": 0.1639356228958999,
      "learning_rate": 3.867925968395085e-06,
      "loss": 0.0022,
      "step": 1351
    },
    {
      "epoch": 2.1632,
      "grad_norm": 0.2127497575095144,
      "learning_rate": 3.854284894414122e-06,
      "loss": 0.0028,
      "step": 1352
    },
    {
      "epoch": 2.1648,
      "grad_norm": 0.35107268142876874,
      "learning_rate": 3.840662172471315e-06,
      "loss": 0.0061,
      "step": 1353
    },
    {
      "epoch": 2.1664,
      "grad_norm": 0.4737783292996846,
      "learning_rate": 3.827057843246181e-06,
      "loss": 0.0057,
      "step": 1354
    },
    {
      "epoch": 2.168,
      "grad_norm": 0.30491471805866227,
      "learning_rate": 3.8134719473633098e-06,
      "loss": 0.0039,
      "step": 1355
    },
    {
      "epoch": 2.1696,
      "grad_norm": 0.30209230149364336,
      "learning_rate": 3.799904525392251e-06,
      "loss": 0.0042,
      "step": 1356
    },
    {
      "epoch": 2.1712,
      "grad_norm": 0.5563014157232996,
      "learning_rate": 3.786355617847385e-06,
      "loss": 0.0064,
      "step": 1357
    },
    {
      "epoch": 2.1728,
      "grad_norm": 0.5629153450463303,
      "learning_rate": 3.7728252651878018e-06,
      "loss": 0.0042,
      "step": 1358
    },
    {
      "epoch": 2.1744,
      "grad_norm": 0.206585291044635,
      "learning_rate": 3.759313507817196e-06,
      "loss": 0.0017,
      "step": 1359
    },
    {
      "epoch": 2.176,
      "grad_norm": 0.3337582064935557,
      "learning_rate": 3.745820386083724e-06,
      "loss": 0.004,
      "step": 1360
    },
    {
      "epoch": 2.1776,
      "grad_norm": 0.29096226358428184,
      "learning_rate": 3.7323459402798936e-06,
      "loss": 0.0046,
      "step": 1361
    },
    {
      "epoch": 2.1792,
      "grad_norm": 0.39433867019101887,
      "learning_rate": 3.718890210642442e-06,
      "loss": 0.0043,
      "step": 1362
    },
    {
      "epoch": 2.1808,
      "grad_norm": 0.2635655806134531,
      "learning_rate": 3.705453237352227e-06,
      "loss": 0.0033,
      "step": 1363
    },
    {
      "epoch": 2.1824,
      "grad_norm": 0.3111796237224152,
      "learning_rate": 3.6920350605340883e-06,
      "loss": 0.0036,
      "step": 1364
    },
    {
      "epoch": 2.184,
      "grad_norm": 0.3144976699122512,
      "learning_rate": 3.6786357202567367e-06,
      "loss": 0.0031,
      "step": 1365
    },
    {
      "epoch": 2.1856,
      "grad_norm": 0.4266913195390291,
      "learning_rate": 3.6652552565326382e-06,
      "loss": 0.0068,
      "step": 1366
    },
    {
      "epoch": 2.1872,
      "grad_norm": 0.2717717217634583,
      "learning_rate": 3.6518937093178873e-06,
      "loss": 0.004,
      "step": 1367
    },
    {
      "epoch": 2.1888,
      "grad_norm": 0.508523412760655,
      "learning_rate": 3.638551118512089e-06,
      "loss": 0.0061,
      "step": 1368
    },
    {
      "epoch": 2.1904,
      "grad_norm": 0.273214646716238,
      "learning_rate": 3.6252275239582522e-06,
      "loss": 0.0047,
      "step": 1369
    },
    {
      "epoch": 2.192,
      "grad_norm": 0.3764530890741371,
      "learning_rate": 3.611922965442648e-06,
      "loss": 0.004,
      "step": 1370
    },
    {
      "epoch": 2.1936,
      "grad_norm": 0.24772481640114252,
      "learning_rate": 3.5986374826947067e-06,
      "loss": 0.0022,
      "step": 1371
    },
    {
      "epoch": 2.1952,
      "grad_norm": 0.30233854902217266,
      "learning_rate": 3.5853711153868962e-06,
      "loss": 0.0035,
      "step": 1372
    },
    {
      "epoch": 2.1968,
      "grad_norm": 0.4774706382490888,
      "learning_rate": 3.5721239031346067e-06,
      "loss": 0.0088,
      "step": 1373
    },
    {
      "epoch": 2.1984,
      "grad_norm": 0.2645838272243305,
      "learning_rate": 3.558895885496023e-06,
      "loss": 0.0029,
      "step": 1374
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.310741726897126,
      "learning_rate": 3.545687101972013e-06,
      "loss": 0.0031,
      "step": 1375
    },
    {
      "epoch": 2.2016,
      "grad_norm": 0.2528778399021895,
      "learning_rate": 3.53249759200601e-06,
      "loss": 0.0031,
      "step": 1376
    },
    {
      "epoch": 2.2032,
      "grad_norm": 0.3168647834878494,
      "learning_rate": 3.519327394983888e-06,
      "loss": 0.0038,
      "step": 1377
    },
    {
      "epoch": 2.2048,
      "grad_norm": 0.3048180293845282,
      "learning_rate": 3.506176550233863e-06,
      "loss": 0.0044,
      "step": 1378
    },
    {
      "epoch": 2.2064,
      "grad_norm": 0.5099909796721364,
      "learning_rate": 3.4930450970263485e-06,
      "loss": 0.0062,
      "step": 1379
    },
    {
      "epoch": 2.208,
      "grad_norm": 0.37016840677232016,
      "learning_rate": 3.479933074573858e-06,
      "loss": 0.004,
      "step": 1380
    },
    {
      "epoch": 2.2096,
      "grad_norm": 0.3865414949322783,
      "learning_rate": 3.4668405220308797e-06,
      "loss": 0.0065,
      "step": 1381
    },
    {
      "epoch": 2.2112,
      "grad_norm": 0.3237177904121443,
      "learning_rate": 3.453767478493761e-06,
      "loss": 0.0052,
      "step": 1382
    },
    {
      "epoch": 2.2128,
      "grad_norm": 0.17672305270254257,
      "learning_rate": 3.440713983000601e-06,
      "loss": 0.0018,
      "step": 1383
    },
    {
      "epoch": 2.2144,
      "grad_norm": 0.37446742117760956,
      "learning_rate": 3.4276800745311135e-06,
      "loss": 0.0039,
      "step": 1384
    },
    {
      "epoch": 2.216,
      "grad_norm": 0.37271036327618084,
      "learning_rate": 3.4146657920065286e-06,
      "loss": 0.0043,
      "step": 1385
    },
    {
      "epoch": 2.2176,
      "grad_norm": 0.24469565507957217,
      "learning_rate": 3.401671174289469e-06,
      "loss": 0.0034,
      "step": 1386
    },
    {
      "epoch": 2.2192,
      "grad_norm": 0.2746037035440998,
      "learning_rate": 3.3886962601838327e-06,
      "loss": 0.0028,
      "step": 1387
    },
    {
      "epoch": 2.2208,
      "grad_norm": 0.3230434848869537,
      "learning_rate": 3.37574108843469e-06,
      "loss": 0.0044,
      "step": 1388
    },
    {
      "epoch": 2.2224,
      "grad_norm": 0.3526293628919445,
      "learning_rate": 3.3628056977281456e-06,
      "loss": 0.0059,
      "step": 1389
    },
    {
      "epoch": 2.224,
      "grad_norm": 0.2789922220782755,
      "learning_rate": 3.3498901266912397e-06,
      "loss": 0.0029,
      "step": 1390
    },
    {
      "epoch": 2.2256,
      "grad_norm": 0.3246182344755697,
      "learning_rate": 3.3369944138918286e-06,
      "loss": 0.0055,
      "step": 1391
    },
    {
      "epoch": 2.2272,
      "grad_norm": 0.26651265660656137,
      "learning_rate": 3.3241185978384636e-06,
      "loss": 0.0026,
      "step": 1392
    },
    {
      "epoch": 2.2288,
      "grad_norm": 0.43057393091251184,
      "learning_rate": 3.3112627169802948e-06,
      "loss": 0.0058,
      "step": 1393
    },
    {
      "epoch": 2.2304,
      "grad_norm": 0.31773344106889506,
      "learning_rate": 3.2984268097069284e-06,
      "loss": 0.0054,
      "step": 1394
    },
    {
      "epoch": 2.232,
      "grad_norm": 0.39789740348144614,
      "learning_rate": 3.2856109143483316e-06,
      "loss": 0.0048,
      "step": 1395
    },
    {
      "epoch": 2.2336,
      "grad_norm": 0.3131342354025925,
      "learning_rate": 3.2728150691747117e-06,
      "loss": 0.0033,
      "step": 1396
    },
    {
      "epoch": 2.2352,
      "grad_norm": 0.20721870152903424,
      "learning_rate": 3.2600393123964114e-06,
      "loss": 0.0022,
      "step": 1397
    },
    {
      "epoch": 2.2368,
      "grad_norm": 0.311970033505521,
      "learning_rate": 3.2472836821637744e-06,
      "loss": 0.0033,
      "step": 1398
    },
    {
      "epoch": 2.2384,
      "grad_norm": 0.3479014445667459,
      "learning_rate": 3.2345482165670493e-06,
      "loss": 0.0047,
      "step": 1399
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.3117659770003126,
      "learning_rate": 3.22183295363627e-06,
      "loss": 0.0044,
      "step": 1400
    },
    {
      "epoch": 2.2416,
      "grad_norm": 0.36393325077159877,
      "learning_rate": 3.209137931341143e-06,
      "loss": 0.0037,
      "step": 1401
    },
    {
      "epoch": 2.2432,
      "grad_norm": 0.47793755851724995,
      "learning_rate": 3.196463187590929e-06,
      "loss": 0.0047,
      "step": 1402
    },
    {
      "epoch": 2.2448,
      "grad_norm": 0.42910510869632124,
      "learning_rate": 3.183808760234335e-06,
      "loss": 0.0051,
      "step": 1403
    },
    {
      "epoch": 2.2464,
      "grad_norm": 0.2169222373711283,
      "learning_rate": 3.1711746870594083e-06,
      "loss": 0.0036,
      "step": 1404
    },
    {
      "epoch": 2.248,
      "grad_norm": 0.3028100781268857,
      "learning_rate": 3.1585610057934022e-06,
      "loss": 0.0032,
      "step": 1405
    },
    {
      "epoch": 2.2496,
      "grad_norm": 0.5514473213221643,
      "learning_rate": 3.145967754102691e-06,
      "loss": 0.0055,
      "step": 1406
    },
    {
      "epoch": 2.2512,
      "grad_norm": 0.4049785329070876,
      "learning_rate": 3.1333949695926323e-06,
      "loss": 0.0046,
      "step": 1407
    },
    {
      "epoch": 2.2528,
      "grad_norm": 0.40058983163715545,
      "learning_rate": 3.1208426898074685e-06,
      "loss": 0.0067,
      "step": 1408
    },
    {
      "epoch": 2.2544,
      "grad_norm": 0.30103543975515684,
      "learning_rate": 3.1083109522302124e-06,
      "loss": 0.0026,
      "step": 1409
    },
    {
      "epoch": 2.2560000000000002,
      "grad_norm": 0.35093667823401603,
      "learning_rate": 3.0957997942825337e-06,
      "loss": 0.0054,
      "step": 1410
    },
    {
      "epoch": 2.2576,
      "grad_norm": 0.46305339166405457,
      "learning_rate": 3.083309253324651e-06,
      "loss": 0.0032,
      "step": 1411
    },
    {
      "epoch": 2.2592,
      "grad_norm": 0.3253390462209622,
      "learning_rate": 3.070839366655215e-06,
      "loss": 0.0061,
      "step": 1412
    },
    {
      "epoch": 2.2608,
      "grad_norm": 0.4677775103626948,
      "learning_rate": 3.0583901715111965e-06,
      "loss": 0.0037,
      "step": 1413
    },
    {
      "epoch": 2.2624,
      "grad_norm": 0.4040927546843886,
      "learning_rate": 3.045961705067787e-06,
      "loss": 0.0047,
      "step": 1414
    },
    {
      "epoch": 2.2640000000000002,
      "grad_norm": 0.37462510699491497,
      "learning_rate": 3.0335540044382693e-06,
      "loss": 0.0059,
      "step": 1415
    },
    {
      "epoch": 2.2656,
      "grad_norm": 0.36590491327213365,
      "learning_rate": 3.021167106673928e-06,
      "loss": 0.0039,
      "step": 1416
    },
    {
      "epoch": 2.2672,
      "grad_norm": 0.4126162047296427,
      "learning_rate": 3.008801048763914e-06,
      "loss": 0.0086,
      "step": 1417
    },
    {
      "epoch": 2.2688,
      "grad_norm": 0.19225835760207746,
      "learning_rate": 2.996455867635155e-06,
      "loss": 0.0019,
      "step": 1418
    },
    {
      "epoch": 2.2704,
      "grad_norm": 0.2745973327167802,
      "learning_rate": 2.9841316001522345e-06,
      "loss": 0.0026,
      "step": 1419
    },
    {
      "epoch": 2.2720000000000002,
      "grad_norm": 0.4391947025341379,
      "learning_rate": 2.9718282831172885e-06,
      "loss": 0.0043,
      "step": 1420
    },
    {
      "epoch": 2.2736,
      "grad_norm": 0.37677518935738363,
      "learning_rate": 2.9595459532698854e-06,
      "loss": 0.0055,
      "step": 1421
    },
    {
      "epoch": 2.2752,
      "grad_norm": 0.20114083698483004,
      "learning_rate": 2.94728464728693e-06,
      "loss": 0.0025,
      "step": 1422
    },
    {
      "epoch": 2.2768,
      "grad_norm": 0.5013196354873555,
      "learning_rate": 2.9350444017825385e-06,
      "loss": 0.0049,
      "step": 1423
    },
    {
      "epoch": 2.2784,
      "grad_norm": 0.21358665618820144,
      "learning_rate": 2.922825253307947e-06,
      "loss": 0.0032,
      "step": 1424
    },
    {
      "epoch": 2.2800000000000002,
      "grad_norm": 0.3422376824346592,
      "learning_rate": 2.910627238351383e-06,
      "loss": 0.0057,
      "step": 1425
    },
    {
      "epoch": 2.2816,
      "grad_norm": 0.22579646148322313,
      "learning_rate": 2.898450393337977e-06,
      "loss": 0.0031,
      "step": 1426
    },
    {
      "epoch": 2.2832,
      "grad_norm": 0.48227216495460973,
      "learning_rate": 2.886294754629632e-06,
      "loss": 0.0108,
      "step": 1427
    },
    {
      "epoch": 2.2848,
      "grad_norm": 0.44696733275716655,
      "learning_rate": 2.8741603585249312e-06,
      "loss": 0.0045,
      "step": 1428
    },
    {
      "epoch": 2.2864,
      "grad_norm": 0.3841451203939599,
      "learning_rate": 2.8620472412590227e-06,
      "loss": 0.0057,
      "step": 1429
    },
    {
      "epoch": 2.288,
      "grad_norm": 0.29580487326759286,
      "learning_rate": 2.8499554390035144e-06,
      "loss": 0.0024,
      "step": 1430
    },
    {
      "epoch": 2.2896,
      "grad_norm": 0.27885549878850224,
      "learning_rate": 2.837884987866363e-06,
      "loss": 0.0036,
      "step": 1431
    },
    {
      "epoch": 2.2912,
      "grad_norm": 0.25723153018368067,
      "learning_rate": 2.8258359238917665e-06,
      "loss": 0.0047,
      "step": 1432
    },
    {
      "epoch": 2.2928,
      "grad_norm": 0.3956945495465071,
      "learning_rate": 2.8138082830600556e-06,
      "loss": 0.0037,
      "step": 1433
    },
    {
      "epoch": 2.2944,
      "grad_norm": 0.3694973738962662,
      "learning_rate": 2.8018021012875994e-06,
      "loss": 0.0062,
      "step": 1434
    },
    {
      "epoch": 2.296,
      "grad_norm": 0.25758641245570874,
      "learning_rate": 2.789817414426673e-06,
      "loss": 0.0038,
      "step": 1435
    },
    {
      "epoch": 2.2976,
      "grad_norm": 0.4575486542931288,
      "learning_rate": 2.7778542582653746e-06,
      "loss": 0.0098,
      "step": 1436
    },
    {
      "epoch": 2.2992,
      "grad_norm": 0.1979079546178225,
      "learning_rate": 2.7659126685275028e-06,
      "loss": 0.0031,
      "step": 1437
    },
    {
      "epoch": 2.3008,
      "grad_norm": 0.2243893806489553,
      "learning_rate": 2.753992680872457e-06,
      "loss": 0.0029,
      "step": 1438
    },
    {
      "epoch": 2.3024,
      "grad_norm": 0.3084199013963842,
      "learning_rate": 2.7420943308951287e-06,
      "loss": 0.0065,
      "step": 1439
    },
    {
      "epoch": 2.304,
      "grad_norm": 0.4512154129946682,
      "learning_rate": 2.7302176541257984e-06,
      "loss": 0.0089,
      "step": 1440
    },
    {
      "epoch": 2.3056,
      "grad_norm": 0.33548598217031145,
      "learning_rate": 2.718362686030025e-06,
      "loss": 0.0086,
      "step": 1441
    },
    {
      "epoch": 2.3072,
      "grad_norm": 0.19048531501627833,
      "learning_rate": 2.7065294620085425e-06,
      "loss": 0.002,
      "step": 1442
    },
    {
      "epoch": 2.3088,
      "grad_norm": 0.12113369849410853,
      "learning_rate": 2.694718017397151e-06,
      "loss": 0.0017,
      "step": 1443
    },
    {
      "epoch": 2.3104,
      "grad_norm": 0.2889296533710736,
      "learning_rate": 2.6829283874666236e-06,
      "loss": 0.0042,
      "step": 1444
    },
    {
      "epoch": 2.312,
      "grad_norm": 0.19654119828654645,
      "learning_rate": 2.6711606074225783e-06,
      "loss": 0.0027,
      "step": 1445
    },
    {
      "epoch": 2.3136,
      "grad_norm": 0.21111769539052078,
      "learning_rate": 2.6594147124053983e-06,
      "loss": 0.0027,
      "step": 1446
    },
    {
      "epoch": 2.3152,
      "grad_norm": 0.2984545484710624,
      "learning_rate": 2.6476907374901062e-06,
      "loss": 0.0028,
      "step": 1447
    },
    {
      "epoch": 2.3168,
      "grad_norm": 0.27245977028805857,
      "learning_rate": 2.635988717686272e-06,
      "loss": 0.0033,
      "step": 1448
    },
    {
      "epoch": 2.3184,
      "grad_norm": 0.16091645943495417,
      "learning_rate": 2.6243086879379e-06,
      "loss": 0.0021,
      "step": 1449
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.31031957579565217,
      "learning_rate": 2.6126506831233343e-06,
      "loss": 0.0036,
      "step": 1450
    },
    {
      "epoch": 2.3216,
      "grad_norm": 0.30000907332964405,
      "learning_rate": 2.6010147380551474e-06,
      "loss": 0.0047,
      "step": 1451
    },
    {
      "epoch": 2.3232,
      "grad_norm": 0.4743264477032393,
      "learning_rate": 2.5894008874800323e-06,
      "loss": 0.0067,
      "step": 1452
    },
    {
      "epoch": 2.3247999999999998,
      "grad_norm": 0.31529185585749747,
      "learning_rate": 2.577809166078716e-06,
      "loss": 0.0031,
      "step": 1453
    },
    {
      "epoch": 2.3264,
      "grad_norm": 0.2339476450070033,
      "learning_rate": 2.5662396084658383e-06,
      "loss": 0.0021,
      "step": 1454
    },
    {
      "epoch": 2.328,
      "grad_norm": 0.3713041003942751,
      "learning_rate": 2.5546922491898497e-06,
      "loss": 0.0074,
      "step": 1455
    },
    {
      "epoch": 2.3296,
      "grad_norm": 0.2339589802070238,
      "learning_rate": 2.543167122732918e-06,
      "loss": 0.0031,
      "step": 1456
    },
    {
      "epoch": 2.3312,
      "grad_norm": 0.5717784940488866,
      "learning_rate": 2.5316642635108247e-06,
      "loss": 0.0081,
      "step": 1457
    },
    {
      "epoch": 2.3327999999999998,
      "grad_norm": 0.22448609320259003,
      "learning_rate": 2.5201837058728506e-06,
      "loss": 0.0032,
      "step": 1458
    },
    {
      "epoch": 2.3344,
      "grad_norm": 0.25023791492443204,
      "learning_rate": 2.508725484101684e-06,
      "loss": 0.0029,
      "step": 1459
    },
    {
      "epoch": 2.336,
      "grad_norm": 0.20091058624312233,
      "learning_rate": 2.4972896324133143e-06,
      "loss": 0.0031,
      "step": 1460
    },
    {
      "epoch": 2.3376,
      "grad_norm": 0.24096170695328817,
      "learning_rate": 2.485876184956928e-06,
      "loss": 0.002,
      "step": 1461
    },
    {
      "epoch": 2.3392,
      "grad_norm": 0.16494454119077795,
      "learning_rate": 2.474485175814816e-06,
      "loss": 0.0028,
      "step": 1462
    },
    {
      "epoch": 2.3407999999999998,
      "grad_norm": 0.3569701654101745,
      "learning_rate": 2.4631166390022574e-06,
      "loss": 0.0076,
      "step": 1463
    },
    {
      "epoch": 2.3424,
      "grad_norm": 0.3297034067855959,
      "learning_rate": 2.451770608467432e-06,
      "loss": 0.0044,
      "step": 1464
    },
    {
      "epoch": 2.344,
      "grad_norm": 0.23806871388529877,
      "learning_rate": 2.440447118091306e-06,
      "loss": 0.0024,
      "step": 1465
    },
    {
      "epoch": 2.3456,
      "grad_norm": 0.2863711800753721,
      "learning_rate": 2.429146201687538e-06,
      "loss": 0.005,
      "step": 1466
    },
    {
      "epoch": 2.3472,
      "grad_norm": 0.2957686438641965,
      "learning_rate": 2.417867893002387e-06,
      "loss": 0.0022,
      "step": 1467
    },
    {
      "epoch": 2.3487999999999998,
      "grad_norm": 0.2722652520341511,
      "learning_rate": 2.4066122257145898e-06,
      "loss": 0.003,
      "step": 1468
    },
    {
      "epoch": 2.3504,
      "grad_norm": 0.2559967992120655,
      "learning_rate": 2.3953792334352787e-06,
      "loss": 0.0027,
      "step": 1469
    },
    {
      "epoch": 2.352,
      "grad_norm": 0.19374332836986866,
      "learning_rate": 2.3841689497078746e-06,
      "loss": 0.0015,
      "step": 1470
    },
    {
      "epoch": 2.3536,
      "grad_norm": 0.35084170140341775,
      "learning_rate": 2.3729814080079815e-06,
      "loss": 0.0038,
      "step": 1471
    },
    {
      "epoch": 2.3552,
      "grad_norm": 0.28875005932803777,
      "learning_rate": 2.361816641743303e-06,
      "loss": 0.0031,
      "step": 1472
    },
    {
      "epoch": 2.3568,
      "grad_norm": 0.2639017927454099,
      "learning_rate": 2.3506746842535244e-06,
      "loss": 0.0032,
      "step": 1473
    },
    {
      "epoch": 2.3584,
      "grad_norm": 0.3922182807822737,
      "learning_rate": 2.339555568810221e-06,
      "loss": 0.0053,
      "step": 1474
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.475777040855533,
      "learning_rate": 2.328459328616759e-06,
      "loss": 0.0037,
      "step": 1475
    },
    {
      "epoch": 2.3616,
      "grad_norm": 0.26708534180003957,
      "learning_rate": 2.317385996808195e-06,
      "loss": 0.0033,
      "step": 1476
    },
    {
      "epoch": 2.3632,
      "grad_norm": 0.4313635677032356,
      "learning_rate": 2.306335606451181e-06,
      "loss": 0.0059,
      "step": 1477
    },
    {
      "epoch": 2.3648,
      "grad_norm": 0.3618264943721742,
      "learning_rate": 2.295308190543859e-06,
      "loss": 0.0043,
      "step": 1478
    },
    {
      "epoch": 2.3664,
      "grad_norm": 0.12036963531136798,
      "learning_rate": 2.2843037820157678e-06,
      "loss": 0.0014,
      "step": 1479
    },
    {
      "epoch": 2.368,
      "grad_norm": 0.36595545738178586,
      "learning_rate": 2.2733224137277366e-06,
      "loss": 0.0054,
      "step": 1480
    },
    {
      "epoch": 2.3696,
      "grad_norm": 0.26787082388335076,
      "learning_rate": 2.2623641184718048e-06,
      "loss": 0.0035,
      "step": 1481
    },
    {
      "epoch": 2.3712,
      "grad_norm": 0.28522072693267997,
      "learning_rate": 2.251428928971102e-06,
      "loss": 0.0043,
      "step": 1482
    },
    {
      "epoch": 2.3728,
      "grad_norm": 0.5210748077944257,
      "learning_rate": 2.240516877879765e-06,
      "loss": 0.0055,
      "step": 1483
    },
    {
      "epoch": 2.3744,
      "grad_norm": 0.5219477323930597,
      "learning_rate": 2.229627997782834e-06,
      "loss": 0.006,
      "step": 1484
    },
    {
      "epoch": 2.376,
      "grad_norm": 0.33335896703198864,
      "learning_rate": 2.218762321196156e-06,
      "loss": 0.0026,
      "step": 1485
    },
    {
      "epoch": 2.3776,
      "grad_norm": 0.37122145250464333,
      "learning_rate": 2.2079198805662917e-06,
      "loss": 0.0031,
      "step": 1486
    },
    {
      "epoch": 2.3792,
      "grad_norm": 0.311804841509584,
      "learning_rate": 2.1971007082704167e-06,
      "loss": 0.003,
      "step": 1487
    },
    {
      "epoch": 2.3808,
      "grad_norm": 0.33656546755716665,
      "learning_rate": 2.186304836616221e-06,
      "loss": 0.0034,
      "step": 1488
    },
    {
      "epoch": 2.3824,
      "grad_norm": 0.20226762823370856,
      "learning_rate": 2.1755322978418134e-06,
      "loss": 0.0022,
      "step": 1489
    },
    {
      "epoch": 2.384,
      "grad_norm": 0.2478589945954889,
      "learning_rate": 2.1647831241156304e-06,
      "loss": 0.0027,
      "step": 1490
    },
    {
      "epoch": 2.3856,
      "grad_norm": 0.41468702245089384,
      "learning_rate": 2.1540573475363402e-06,
      "loss": 0.0035,
      "step": 1491
    },
    {
      "epoch": 2.3872,
      "grad_norm": 0.3191295821954367,
      "learning_rate": 2.1433550001327376e-06,
      "loss": 0.0036,
      "step": 1492
    },
    {
      "epoch": 2.3888,
      "grad_norm": 0.36544274076642663,
      "learning_rate": 2.1326761138636555e-06,
      "loss": 0.0082,
      "step": 1493
    },
    {
      "epoch": 2.3904,
      "grad_norm": 0.27671618926189717,
      "learning_rate": 2.122020720617869e-06,
      "loss": 0.002,
      "step": 1494
    },
    {
      "epoch": 2.392,
      "grad_norm": 0.30829653757205117,
      "learning_rate": 2.111388852214001e-06,
      "loss": 0.0032,
      "step": 1495
    },
    {
      "epoch": 2.3936,
      "grad_norm": 0.24967904868047844,
      "learning_rate": 2.1007805404004247e-06,
      "loss": 0.0032,
      "step": 1496
    },
    {
      "epoch": 2.3952,
      "grad_norm": 0.34782125592838925,
      "learning_rate": 2.090195816855164e-06,
      "loss": 0.0044,
      "step": 1497
    },
    {
      "epoch": 2.3968,
      "grad_norm": 0.41020445003946276,
      "learning_rate": 2.0796347131858187e-06,
      "loss": 0.0045,
      "step": 1498
    },
    {
      "epoch": 2.3984,
      "grad_norm": 0.34054741896721913,
      "learning_rate": 2.069097260929439e-06,
      "loss": 0.007,
      "step": 1499
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.33057868952454816,
      "learning_rate": 2.058583491552465e-06,
      "loss": 0.0049,
      "step": 1500
    },
    {
      "epoch": 2.4016,
      "grad_norm": 0.20827812995867526,
      "learning_rate": 2.048093436450603e-06,
      "loss": 0.0019,
      "step": 1501
    },
    {
      "epoch": 2.4032,
      "grad_norm": 0.5795964386944271,
      "learning_rate": 2.037627126948751e-06,
      "loss": 0.0109,
      "step": 1502
    },
    {
      "epoch": 2.4048,
      "grad_norm": 0.2061432574165304,
      "learning_rate": 2.0271845943008984e-06,
      "loss": 0.0024,
      "step": 1503
    },
    {
      "epoch": 2.4064,
      "grad_norm": 0.30377799341148626,
      "learning_rate": 2.0167658696900317e-06,
      "loss": 0.0036,
      "step": 1504
    },
    {
      "epoch": 2.408,
      "grad_norm": 0.43473012373512426,
      "learning_rate": 2.006370984228043e-06,
      "loss": 0.0039,
      "step": 1505
    },
    {
      "epoch": 2.4096,
      "grad_norm": 0.3505689933040014,
      "learning_rate": 1.9959999689556407e-06,
      "loss": 0.0043,
      "step": 1506
    },
    {
      "epoch": 2.4112,
      "grad_norm": 0.5849372527808584,
      "learning_rate": 1.985652854842247e-06,
      "loss": 0.0052,
      "step": 1507
    },
    {
      "epoch": 2.4128,
      "grad_norm": 0.4538261118657633,
      "learning_rate": 1.9753296727859195e-06,
      "loss": 0.0038,
      "step": 1508
    },
    {
      "epoch": 2.4144,
      "grad_norm": 0.5685819220855983,
      "learning_rate": 1.9650304536132426e-06,
      "loss": 0.0106,
      "step": 1509
    },
    {
      "epoch": 2.416,
      "grad_norm": 0.3575602209191756,
      "learning_rate": 1.9547552280792528e-06,
      "loss": 0.0027,
      "step": 1510
    },
    {
      "epoch": 2.4176,
      "grad_norm": 0.30509551541333807,
      "learning_rate": 1.9445040268673297e-06,
      "loss": 0.0049,
      "step": 1511
    },
    {
      "epoch": 2.4192,
      "grad_norm": 0.3524991773850427,
      "learning_rate": 1.9342768805891176e-06,
      "loss": 0.0035,
      "step": 1512
    },
    {
      "epoch": 2.4208,
      "grad_norm": 0.17541571972789952,
      "learning_rate": 1.924073819784428e-06,
      "loss": 0.0016,
      "step": 1513
    },
    {
      "epoch": 2.4224,
      "grad_norm": 0.2290188668830978,
      "learning_rate": 1.9138948749211473e-06,
      "loss": 0.0023,
      "step": 1514
    },
    {
      "epoch": 2.424,
      "grad_norm": 0.3169526663891749,
      "learning_rate": 1.9037400763951508e-06,
      "loss": 0.0031,
      "step": 1515
    },
    {
      "epoch": 2.4256,
      "grad_norm": 0.26479670963805474,
      "learning_rate": 1.8936094545302098e-06,
      "loss": 0.0055,
      "step": 1516
    },
    {
      "epoch": 2.4272,
      "grad_norm": 0.18324366456774377,
      "learning_rate": 1.8835030395778941e-06,
      "loss": 0.002,
      "step": 1517
    },
    {
      "epoch": 2.4288,
      "grad_norm": 0.4203047399682966,
      "learning_rate": 1.8734208617174986e-06,
      "loss": 0.0038,
      "step": 1518
    },
    {
      "epoch": 2.4304,
      "grad_norm": 0.20604756226606,
      "learning_rate": 1.8633629510559315e-06,
      "loss": 0.0025,
      "step": 1519
    },
    {
      "epoch": 2.432,
      "grad_norm": 0.2766323590013184,
      "learning_rate": 1.8533293376276473e-06,
      "loss": 0.0042,
      "step": 1520
    },
    {
      "epoch": 2.4336,
      "grad_norm": 0.44343747501893677,
      "learning_rate": 1.8433200513945338e-06,
      "loss": 0.0055,
      "step": 1521
    },
    {
      "epoch": 2.4352,
      "grad_norm": 0.24799125963285085,
      "learning_rate": 1.8333351222458407e-06,
      "loss": 0.0047,
      "step": 1522
    },
    {
      "epoch": 2.4368,
      "grad_norm": 0.2526019202655489,
      "learning_rate": 1.8233745799980818e-06,
      "loss": 0.0038,
      "step": 1523
    },
    {
      "epoch": 2.4384,
      "grad_norm": 0.2039656728483351,
      "learning_rate": 1.813438454394948e-06,
      "loss": 0.0029,
      "step": 1524
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.4609040856616715,
      "learning_rate": 1.8035267751072172e-06,
      "loss": 0.004,
      "step": 1525
    },
    {
      "epoch": 2.4416,
      "grad_norm": 0.3248920744267325,
      "learning_rate": 1.7936395717326705e-06,
      "loss": 0.0057,
      "step": 1526
    },
    {
      "epoch": 2.4432,
      "grad_norm": 0.22369747227456252,
      "learning_rate": 1.7837768737959937e-06,
      "loss": 0.0028,
      "step": 1527
    },
    {
      "epoch": 2.4448,
      "grad_norm": 0.19174459869622357,
      "learning_rate": 1.773938710748706e-06,
      "loss": 0.002,
      "step": 1528
    },
    {
      "epoch": 2.4464,
      "grad_norm": 0.3870861370113197,
      "learning_rate": 1.7641251119690505e-06,
      "loss": 0.0057,
      "step": 1529
    },
    {
      "epoch": 2.448,
      "grad_norm": 0.40104713267802156,
      "learning_rate": 1.7543361067619269e-06,
      "loss": 0.0106,
      "step": 1530
    },
    {
      "epoch": 2.4496,
      "grad_norm": 0.33739972760230735,
      "learning_rate": 1.7445717243587889e-06,
      "loss": 0.0032,
      "step": 1531
    },
    {
      "epoch": 2.4512,
      "grad_norm": 0.3016213026287297,
      "learning_rate": 1.734831993917564e-06,
      "loss": 0.0027,
      "step": 1532
    },
    {
      "epoch": 2.4528,
      "grad_norm": 0.1849836736276717,
      "learning_rate": 1.7251169445225658e-06,
      "loss": 0.0014,
      "step": 1533
    },
    {
      "epoch": 2.4544,
      "grad_norm": 0.3882689943555003,
      "learning_rate": 1.715426605184407e-06,
      "loss": 0.0068,
      "step": 1534
    },
    {
      "epoch": 2.456,
      "grad_norm": 0.3403722116164076,
      "learning_rate": 1.705761004839911e-06,
      "loss": 0.0041,
      "step": 1535
    },
    {
      "epoch": 2.4576000000000002,
      "grad_norm": 0.3656401880020418,
      "learning_rate": 1.6961201723520248e-06,
      "loss": 0.0055,
      "step": 1536
    },
    {
      "epoch": 2.4592,
      "grad_norm": 0.35146384136990116,
      "learning_rate": 1.6865041365097434e-06,
      "loss": 0.003,
      "step": 1537
    },
    {
      "epoch": 2.4608,
      "grad_norm": 0.6592852769443976,
      "learning_rate": 1.676912926028007e-06,
      "loss": 0.0079,
      "step": 1538
    },
    {
      "epoch": 2.4624,
      "grad_norm": 0.5307332952929912,
      "learning_rate": 1.6673465695476233e-06,
      "loss": 0.0062,
      "step": 1539
    },
    {
      "epoch": 2.464,
      "grad_norm": 0.18065197635713737,
      "learning_rate": 1.6578050956351887e-06,
      "loss": 0.0024,
      "step": 1540
    },
    {
      "epoch": 2.4656000000000002,
      "grad_norm": 0.25899370366356406,
      "learning_rate": 1.6482885327829912e-06,
      "loss": 0.0033,
      "step": 1541
    },
    {
      "epoch": 2.4672,
      "grad_norm": 0.45720416015892273,
      "learning_rate": 1.6387969094089318e-06,
      "loss": 0.005,
      "step": 1542
    },
    {
      "epoch": 2.4688,
      "grad_norm": 0.14764278603001516,
      "learning_rate": 1.6293302538564381e-06,
      "loss": 0.0018,
      "step": 1543
    },
    {
      "epoch": 2.4704,
      "grad_norm": 0.41132663149020565,
      "learning_rate": 1.619888594394382e-06,
      "loss": 0.0068,
      "step": 1544
    },
    {
      "epoch": 2.472,
      "grad_norm": 0.5567151556235029,
      "learning_rate": 1.6104719592169905e-06,
      "loss": 0.0032,
      "step": 1545
    },
    {
      "epoch": 2.4736000000000002,
      "grad_norm": 0.3841223040297522,
      "learning_rate": 1.6010803764437633e-06,
      "loss": 0.007,
      "step": 1546
    },
    {
      "epoch": 2.4752,
      "grad_norm": 0.2533837036175254,
      "learning_rate": 1.5917138741193972e-06,
      "loss": 0.0035,
      "step": 1547
    },
    {
      "epoch": 2.4768,
      "grad_norm": 0.41208188391712325,
      "learning_rate": 1.5823724802136863e-06,
      "loss": 0.0057,
      "step": 1548
    },
    {
      "epoch": 2.4784,
      "grad_norm": 0.333231114609289,
      "learning_rate": 1.5730562226214529e-06,
      "loss": 0.0035,
      "step": 1549
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.3460601045673392,
      "learning_rate": 1.5637651291624522e-06,
      "loss": 0.0048,
      "step": 1550
    },
    {
      "epoch": 2.4816,
      "grad_norm": 0.4561451554139878,
      "learning_rate": 1.5544992275813053e-06,
      "loss": 0.006,
      "step": 1551
    },
    {
      "epoch": 2.4832,
      "grad_norm": 0.2600217071914026,
      "learning_rate": 1.545258545547398e-06,
      "loss": 0.0033,
      "step": 1552
    },
    {
      "epoch": 2.4848,
      "grad_norm": 0.26522188048489387,
      "learning_rate": 1.536043110654809e-06,
      "loss": 0.0046,
      "step": 1553
    },
    {
      "epoch": 2.4864,
      "grad_norm": 0.26254243975243546,
      "learning_rate": 1.5268529504222262e-06,
      "loss": 0.0029,
      "step": 1554
    },
    {
      "epoch": 2.488,
      "grad_norm": 0.3105340678608335,
      "learning_rate": 1.5176880922928615e-06,
      "loss": 0.0046,
      "step": 1555
    },
    {
      "epoch": 2.4896,
      "grad_norm": 0.3592223753091298,
      "learning_rate": 1.5085485636343755e-06,
      "loss": 0.004,
      "step": 1556
    },
    {
      "epoch": 2.4912,
      "grad_norm": 0.27150401965803533,
      "learning_rate": 1.4994343917387854e-06,
      "loss": 0.0037,
      "step": 1557
    },
    {
      "epoch": 2.4928,
      "grad_norm": 0.5551493870366635,
      "learning_rate": 1.4903456038223941e-06,
      "loss": 0.0132,
      "step": 1558
    },
    {
      "epoch": 2.4944,
      "grad_norm": 0.34588012387295763,
      "learning_rate": 1.481282227025701e-06,
      "loss": 0.0064,
      "step": 1559
    },
    {
      "epoch": 2.496,
      "grad_norm": 0.23905683065560415,
      "learning_rate": 1.4722442884133214e-06,
      "loss": 0.0025,
      "step": 1560
    },
    {
      "epoch": 2.4976,
      "grad_norm": 0.397704021336595,
      "learning_rate": 1.4632318149739177e-06,
      "loss": 0.0049,
      "step": 1561
    },
    {
      "epoch": 2.4992,
      "grad_norm": 0.3374482968206889,
      "learning_rate": 1.4542448336201021e-06,
      "loss": 0.0033,
      "step": 1562
    },
    {
      "epoch": 2.5008,
      "grad_norm": 0.34762474001471755,
      "learning_rate": 1.4452833711883629e-06,
      "loss": 0.0087,
      "step": 1563
    },
    {
      "epoch": 2.5023999999999997,
      "grad_norm": 0.2282977695631175,
      "learning_rate": 1.4363474544389876e-06,
      "loss": 0.0026,
      "step": 1564
    },
    {
      "epoch": 2.504,
      "grad_norm": 0.42541342818139793,
      "learning_rate": 1.4274371100559792e-06,
      "loss": 0.0048,
      "step": 1565
    },
    {
      "epoch": 2.5056000000000003,
      "grad_norm": 0.4863614084250172,
      "learning_rate": 1.4185523646469822e-06,
      "loss": 0.0044,
      "step": 1566
    },
    {
      "epoch": 2.5072,
      "grad_norm": 0.12498832532587154,
      "learning_rate": 1.409693244743192e-06,
      "loss": 0.0019,
      "step": 1567
    },
    {
      "epoch": 2.5088,
      "grad_norm": 0.4225063695409753,
      "learning_rate": 1.4008597767992872e-06,
      "loss": 0.0047,
      "step": 1568
    },
    {
      "epoch": 2.5103999999999997,
      "grad_norm": 0.230938913837018,
      "learning_rate": 1.3920519871933425e-06,
      "loss": 0.0026,
      "step": 1569
    },
    {
      "epoch": 2.512,
      "grad_norm": 0.2688014101429334,
      "learning_rate": 1.3832699022267516e-06,
      "loss": 0.0047,
      "step": 1570
    },
    {
      "epoch": 2.5136,
      "grad_norm": 0.5189916763612203,
      "learning_rate": 1.3745135481241602e-06,
      "loss": 0.0056,
      "step": 1571
    },
    {
      "epoch": 2.5152,
      "grad_norm": 0.2823365163678064,
      "learning_rate": 1.3657829510333653e-06,
      "loss": 0.0043,
      "step": 1572
    },
    {
      "epoch": 2.5168,
      "grad_norm": 0.2323500911953732,
      "learning_rate": 1.3570781370252584e-06,
      "loss": 0.0026,
      "step": 1573
    },
    {
      "epoch": 2.5183999999999997,
      "grad_norm": 0.26700063611671265,
      "learning_rate": 1.3483991320937307e-06,
      "loss": 0.0031,
      "step": 1574
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.2532235049322369,
      "learning_rate": 1.339745962155613e-06,
      "loss": 0.0035,
      "step": 1575
    },
    {
      "epoch": 2.5216,
      "grad_norm": 0.3169413186063335,
      "learning_rate": 1.3311186530505838e-06,
      "loss": 0.0045,
      "step": 1576
    },
    {
      "epoch": 2.5232,
      "grad_norm": 0.25250982744404826,
      "learning_rate": 1.322517230541096e-06,
      "loss": 0.0051,
      "step": 1577
    },
    {
      "epoch": 2.5248,
      "grad_norm": 0.35861267443225797,
      "learning_rate": 1.313941720312303e-06,
      "loss": 0.0047,
      "step": 1578
    },
    {
      "epoch": 2.5263999999999998,
      "grad_norm": 0.33250067415436657,
      "learning_rate": 1.30539214797198e-06,
      "loss": 0.003,
      "step": 1579
    },
    {
      "epoch": 2.528,
      "grad_norm": 0.26320164189665274,
      "learning_rate": 1.2968685390504465e-06,
      "loss": 0.0029,
      "step": 1580
    },
    {
      "epoch": 2.5296,
      "grad_norm": 0.33401526006983745,
      "learning_rate": 1.2883709190004956e-06,
      "loss": 0.0039,
      "step": 1581
    },
    {
      "epoch": 2.5312,
      "grad_norm": 0.30630453343895253,
      "learning_rate": 1.2798993131973093e-06,
      "loss": 0.0037,
      "step": 1582
    },
    {
      "epoch": 2.5328,
      "grad_norm": 0.1999894111806715,
      "learning_rate": 1.2714537469383858e-06,
      "loss": 0.002,
      "step": 1583
    },
    {
      "epoch": 2.5343999999999998,
      "grad_norm": 0.32881800262495675,
      "learning_rate": 1.263034245443473e-06,
      "loss": 0.0037,
      "step": 1584
    },
    {
      "epoch": 2.536,
      "grad_norm": 0.21174416417483852,
      "learning_rate": 1.254640833854477e-06,
      "loss": 0.0037,
      "step": 1585
    },
    {
      "epoch": 2.5376,
      "grad_norm": 0.08888564725084476,
      "learning_rate": 1.2462735372353996e-06,
      "loss": 0.0013,
      "step": 1586
    },
    {
      "epoch": 2.5392,
      "grad_norm": 0.38680815178140715,
      "learning_rate": 1.2379323805722575e-06,
      "loss": 0.0075,
      "step": 1587
    },
    {
      "epoch": 2.5408,
      "grad_norm": 0.2271389553814667,
      "learning_rate": 1.2296173887730122e-06,
      "loss": 0.0034,
      "step": 1588
    },
    {
      "epoch": 2.5423999999999998,
      "grad_norm": 0.43773514969335675,
      "learning_rate": 1.2213285866674908e-06,
      "loss": 0.0049,
      "step": 1589
    },
    {
      "epoch": 2.544,
      "grad_norm": 0.29155850382909776,
      "learning_rate": 1.2130659990073146e-06,
      "loss": 0.0029,
      "step": 1590
    },
    {
      "epoch": 2.5456,
      "grad_norm": 0.2653316292247236,
      "learning_rate": 1.2048296504658208e-06,
      "loss": 0.003,
      "step": 1591
    },
    {
      "epoch": 2.5472,
      "grad_norm": 0.36255738638210777,
      "learning_rate": 1.196619565638003e-06,
      "loss": 0.0033,
      "step": 1592
    },
    {
      "epoch": 2.5488,
      "grad_norm": 0.28846739314243985,
      "learning_rate": 1.1884357690404157e-06,
      "loss": 0.0042,
      "step": 1593
    },
    {
      "epoch": 2.5504,
      "grad_norm": 0.28643151052329563,
      "learning_rate": 1.1802782851111206e-06,
      "loss": 0.0034,
      "step": 1594
    },
    {
      "epoch": 2.552,
      "grad_norm": 0.3566538575233225,
      "learning_rate": 1.1721471382096028e-06,
      "loss": 0.0068,
      "step": 1595
    },
    {
      "epoch": 2.5536,
      "grad_norm": 0.6056236455845999,
      "learning_rate": 1.1640423526166987e-06,
      "loss": 0.0052,
      "step": 1596
    },
    {
      "epoch": 2.5552,
      "grad_norm": 0.3432641368965244,
      "learning_rate": 1.1559639525345313e-06,
      "loss": 0.0034,
      "step": 1597
    },
    {
      "epoch": 2.5568,
      "grad_norm": 0.314694297610856,
      "learning_rate": 1.1479119620864277e-06,
      "loss": 0.0063,
      "step": 1598
    },
    {
      "epoch": 2.5584,
      "grad_norm": 0.16891172033085503,
      "learning_rate": 1.1398864053168534e-06,
      "loss": 0.0021,
      "step": 1599
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.28250641261356624,
      "learning_rate": 1.1318873061913405e-06,
      "loss": 0.004,
      "step": 1600
    },
    {
      "epoch": 2.5616,
      "grad_norm": 0.3870674371758313,
      "learning_rate": 1.123914688596409e-06,
      "loss": 0.0042,
      "step": 1601
    },
    {
      "epoch": 2.5632,
      "grad_norm": 0.1787916187396734,
      "learning_rate": 1.1159685763395113e-06,
      "loss": 0.0021,
      "step": 1602
    },
    {
      "epoch": 2.5648,
      "grad_norm": 0.09943497923667746,
      "learning_rate": 1.108048993148939e-06,
      "loss": 0.0011,
      "step": 1603
    },
    {
      "epoch": 2.5664,
      "grad_norm": 0.3028880170947417,
      "learning_rate": 1.1001559626737757e-06,
      "loss": 0.0032,
      "step": 1604
    },
    {
      "epoch": 2.568,
      "grad_norm": 0.27090689294158915,
      "learning_rate": 1.0922895084838036e-06,
      "loss": 0.0039,
      "step": 1605
    },
    {
      "epoch": 2.5696,
      "grad_norm": 0.5076518218659594,
      "learning_rate": 1.0844496540694515e-06,
      "loss": 0.0056,
      "step": 1606
    },
    {
      "epoch": 2.5712,
      "grad_norm": 0.5329209114392545,
      "learning_rate": 1.0766364228417148e-06,
      "loss": 0.0085,
      "step": 1607
    },
    {
      "epoch": 2.5728,
      "grad_norm": 0.37775231134155207,
      "learning_rate": 1.0688498381320855e-06,
      "loss": 0.0051,
      "step": 1608
    },
    {
      "epoch": 2.5744,
      "grad_norm": 0.27772476593874784,
      "learning_rate": 1.0610899231924887e-06,
      "loss": 0.003,
      "step": 1609
    },
    {
      "epoch": 2.576,
      "grad_norm": 0.19838487249698816,
      "learning_rate": 1.0533567011952094e-06,
      "loss": 0.0027,
      "step": 1610
    },
    {
      "epoch": 2.5776,
      "grad_norm": 0.25085948073206776,
      "learning_rate": 1.0456501952328191e-06,
      "loss": 0.0034,
      "step": 1611
    },
    {
      "epoch": 2.5792,
      "grad_norm": 0.39908536394921706,
      "learning_rate": 1.037970428318118e-06,
      "loss": 0.0041,
      "step": 1612
    },
    {
      "epoch": 2.5808,
      "grad_norm": 0.6323558270607047,
      "learning_rate": 1.0303174233840529e-06,
      "loss": 0.0064,
      "step": 1613
    },
    {
      "epoch": 2.5824,
      "grad_norm": 0.18981437980393318,
      "learning_rate": 1.022691203283661e-06,
      "loss": 0.0029,
      "step": 1614
    },
    {
      "epoch": 2.584,
      "grad_norm": 0.22420449384561897,
      "learning_rate": 1.0150917907899926e-06,
      "loss": 0.0028,
      "step": 1615
    },
    {
      "epoch": 2.5856,
      "grad_norm": 0.22817326728027199,
      "learning_rate": 1.0075192085960451e-06,
      "loss": 0.0023,
      "step": 1616
    },
    {
      "epoch": 2.5872,
      "grad_norm": 0.24630772747490712,
      "learning_rate": 9.999734793146998e-07,
      "loss": 0.0038,
      "step": 1617
    },
    {
      "epoch": 2.5888,
      "grad_norm": 0.31054400155061074,
      "learning_rate": 9.924546254786493e-07,
      "loss": 0.0034,
      "step": 1618
    },
    {
      "epoch": 2.5904,
      "grad_norm": 0.15441061679808213,
      "learning_rate": 9.849626695403326e-07,
      "loss": 0.0029,
      "step": 1619
    },
    {
      "epoch": 2.592,
      "grad_norm": 0.28853886186684,
      "learning_rate": 9.77497633871868e-07,
      "loss": 0.0031,
      "step": 1620
    },
    {
      "epoch": 2.5936,
      "grad_norm": 0.13359951082571278,
      "learning_rate": 9.700595407649805e-07,
      "loss": 0.0013,
      "step": 1621
    },
    {
      "epoch": 2.5952,
      "grad_norm": 0.25004584541730634,
      "learning_rate": 9.62648412430951e-07,
      "loss": 0.004,
      "step": 1622
    },
    {
      "epoch": 2.5968,
      "grad_norm": 0.39526327537043676,
      "learning_rate": 9.5526427100053e-07,
      "loss": 0.0035,
      "step": 1623
    },
    {
      "epoch": 2.5984,
      "grad_norm": 0.24772152883498377,
      "learning_rate": 9.479071385238892e-07,
      "loss": 0.0033,
      "step": 1624
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.3254513376690307,
      "learning_rate": 9.40577036970538e-07,
      "loss": 0.0029,
      "step": 1625
    },
    {
      "epoch": 2.6016,
      "grad_norm": 0.34472744884953915,
      "learning_rate": 9.332739882292752e-07,
      "loss": 0.0035,
      "step": 1626
    },
    {
      "epoch": 2.6032,
      "grad_norm": 0.37230591882644276,
      "learning_rate": 9.259980141081115e-07,
      "loss": 0.0033,
      "step": 1627
    },
    {
      "epoch": 2.6048,
      "grad_norm": 0.3720842210650768,
      "learning_rate": 9.187491363342094e-07,
      "loss": 0.0046,
      "step": 1628
    },
    {
      "epoch": 2.6064,
      "grad_norm": 0.4182576528710122,
      "learning_rate": 9.115273765538202e-07,
      "loss": 0.0089,
      "step": 1629
    },
    {
      "epoch": 2.608,
      "grad_norm": 0.5352513477811385,
      "learning_rate": 9.043327563322113e-07,
      "loss": 0.0076,
      "step": 1630
    },
    {
      "epoch": 2.6096,
      "grad_norm": 0.316514214802117,
      "learning_rate": 8.971652971536149e-07,
      "loss": 0.0046,
      "step": 1631
    },
    {
      "epoch": 2.6112,
      "grad_norm": 0.23125374471918014,
      "learning_rate": 8.900250204211513e-07,
      "loss": 0.0034,
      "step": 1632
    },
    {
      "epoch": 2.6128,
      "grad_norm": 0.290009493417907,
      "learning_rate": 8.829119474567672e-07,
      "loss": 0.0036,
      "step": 1633
    },
    {
      "epoch": 2.6144,
      "grad_norm": 0.22741470652410545,
      "learning_rate": 8.758260995011825e-07,
      "loss": 0.0022,
      "step": 1634
    },
    {
      "epoch": 2.616,
      "grad_norm": 0.44988819712172906,
      "learning_rate": 8.687674977138116e-07,
      "loss": 0.0067,
      "step": 1635
    },
    {
      "epoch": 2.6176,
      "grad_norm": 0.40645278584597383,
      "learning_rate": 8.617361631727139e-07,
      "loss": 0.0051,
      "step": 1636
    },
    {
      "epoch": 2.6192,
      "grad_norm": 0.39458796436952626,
      "learning_rate": 8.547321168745192e-07,
      "loss": 0.004,
      "step": 1637
    },
    {
      "epoch": 2.6208,
      "grad_norm": 0.21891152487583124,
      "learning_rate": 8.477553797343729e-07,
      "loss": 0.0027,
      "step": 1638
    },
    {
      "epoch": 2.6224,
      "grad_norm": 0.39473639258139465,
      "learning_rate": 8.40805972585872e-07,
      "loss": 0.0034,
      "step": 1639
    },
    {
      "epoch": 2.624,
      "grad_norm": 0.4043861845152941,
      "learning_rate": 8.338839161809997e-07,
      "loss": 0.0028,
      "step": 1640
    },
    {
      "epoch": 2.6256,
      "grad_norm": 0.27864258890756977,
      "learning_rate": 8.269892311900696e-07,
      "loss": 0.0031,
      "step": 1641
    },
    {
      "epoch": 2.6272,
      "grad_norm": 0.26557343737536915,
      "learning_rate": 8.201219382016556e-07,
      "loss": 0.0027,
      "step": 1642
    },
    {
      "epoch": 2.6288,
      "grad_norm": 0.24467885630190223,
      "learning_rate": 8.132820577225386e-07,
      "loss": 0.0034,
      "step": 1643
    },
    {
      "epoch": 2.6304,
      "grad_norm": 0.3069297710916183,
      "learning_rate": 8.06469610177636e-07,
      "loss": 0.004,
      "step": 1644
    },
    {
      "epoch": 2.632,
      "grad_norm": 0.2345896763838344,
      "learning_rate": 7.996846159099558e-07,
      "loss": 0.0026,
      "step": 1645
    },
    {
      "epoch": 2.6336,
      "grad_norm": 0.2687014085198747,
      "learning_rate": 7.92927095180518e-07,
      "loss": 0.0023,
      "step": 1646
    },
    {
      "epoch": 2.6352,
      "grad_norm": 0.2509517187635115,
      "learning_rate": 7.861970681683051e-07,
      "loss": 0.0029,
      "step": 1647
    },
    {
      "epoch": 2.6368,
      "grad_norm": 0.3224143429555162,
      "learning_rate": 7.794945549701993e-07,
      "loss": 0.003,
      "step": 1648
    },
    {
      "epoch": 2.6384,
      "grad_norm": 0.34475283717693306,
      "learning_rate": 7.728195756009204e-07,
      "loss": 0.0028,
      "step": 1649
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.4283927293710006,
      "learning_rate": 7.661721499929753e-07,
      "loss": 0.0114,
      "step": 1650
    },
    {
      "epoch": 2.6416,
      "grad_norm": 0.23359871179380523,
      "learning_rate": 7.595522979965819e-07,
      "loss": 0.0024,
      "step": 1651
    },
    {
      "epoch": 2.6432,
      "grad_norm": 0.28872326379249535,
      "learning_rate": 7.529600393796232e-07,
      "loss": 0.0027,
      "step": 1652
    },
    {
      "epoch": 2.6448,
      "grad_norm": 0.20751657884631444,
      "learning_rate": 7.463953938275859e-07,
      "loss": 0.0039,
      "step": 1653
    },
    {
      "epoch": 2.6464,
      "grad_norm": 0.1548793088215341,
      "learning_rate": 7.398583809434944e-07,
      "loss": 0.0017,
      "step": 1654
    },
    {
      "epoch": 2.648,
      "grad_norm": 0.18302776241554822,
      "learning_rate": 7.333490202478666e-07,
      "loss": 0.0029,
      "step": 1655
    },
    {
      "epoch": 2.6496,
      "grad_norm": 0.2665263815329152,
      "learning_rate": 7.268673311786378e-07,
      "loss": 0.003,
      "step": 1656
    },
    {
      "epoch": 2.6512000000000002,
      "grad_norm": 0.2852673473046866,
      "learning_rate": 7.204133330911179e-07,
      "loss": 0.0026,
      "step": 1657
    },
    {
      "epoch": 2.6528,
      "grad_norm": 0.4359148892423722,
      "learning_rate": 7.1398704525792e-07,
      "loss": 0.0042,
      "step": 1658
    },
    {
      "epoch": 2.6544,
      "grad_norm": 0.22052120673606213,
      "learning_rate": 7.07588486868922e-07,
      "loss": 0.0024,
      "step": 1659
    },
    {
      "epoch": 2.656,
      "grad_norm": 0.1597290951375419,
      "learning_rate": 7.012176770311863e-07,
      "loss": 0.002,
      "step": 1660
    },
    {
      "epoch": 2.6576,
      "grad_norm": 0.17961582222811928,
      "learning_rate": 6.948746347689184e-07,
      "loss": 0.0018,
      "step": 1661
    },
    {
      "epoch": 2.6592000000000002,
      "grad_norm": 0.27243437680930604,
      "learning_rate": 6.885593790234057e-07,
      "loss": 0.0019,
      "step": 1662
    },
    {
      "epoch": 2.6608,
      "grad_norm": 0.18828994729054635,
      "learning_rate": 6.8227192865296e-07,
      "loss": 0.0051,
      "step": 1663
    },
    {
      "epoch": 2.6624,
      "grad_norm": 0.2779927024355395,
      "learning_rate": 6.760123024328624e-07,
      "loss": 0.0031,
      "step": 1664
    },
    {
      "epoch": 2.664,
      "grad_norm": 0.1688514914160169,
      "learning_rate": 6.697805190553086e-07,
      "loss": 0.002,
      "step": 1665
    },
    {
      "epoch": 2.6656,
      "grad_norm": 0.28104834926529215,
      "learning_rate": 6.635765971293484e-07,
      "loss": 0.0033,
      "step": 1666
    },
    {
      "epoch": 2.6672000000000002,
      "grad_norm": 0.3832135221159943,
      "learning_rate": 6.574005551808338e-07,
      "loss": 0.0058,
      "step": 1667
    },
    {
      "epoch": 2.6688,
      "grad_norm": 0.18272617424296622,
      "learning_rate": 6.512524116523633e-07,
      "loss": 0.0025,
      "step": 1668
    },
    {
      "epoch": 2.6704,
      "grad_norm": 0.2412288002713958,
      "learning_rate": 6.451321849032289e-07,
      "loss": 0.0026,
      "step": 1669
    },
    {
      "epoch": 2.672,
      "grad_norm": 0.4097149189369184,
      "learning_rate": 6.390398932093555e-07,
      "loss": 0.0043,
      "step": 1670
    },
    {
      "epoch": 2.6736,
      "grad_norm": 0.27431070134136,
      "learning_rate": 6.329755547632499e-07,
      "loss": 0.0033,
      "step": 1671
    },
    {
      "epoch": 2.6752000000000002,
      "grad_norm": 0.23379952328211157,
      "learning_rate": 6.269391876739494e-07,
      "loss": 0.002,
      "step": 1672
    },
    {
      "epoch": 2.6768,
      "grad_norm": 0.4225055935864826,
      "learning_rate": 6.209308099669598e-07,
      "loss": 0.0069,
      "step": 1673
    },
    {
      "epoch": 2.6784,
      "grad_norm": 0.17709261279424496,
      "learning_rate": 6.149504395842087e-07,
      "loss": 0.0016,
      "step": 1674
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.310095475370071,
      "learning_rate": 6.089980943839924e-07,
      "loss": 0.0035,
      "step": 1675
    },
    {
      "epoch": 2.6816,
      "grad_norm": 0.6092157006990844,
      "learning_rate": 6.030737921409169e-07,
      "loss": 0.0043,
      "step": 1676
    },
    {
      "epoch": 2.6832000000000003,
      "grad_norm": 0.4616010603664959,
      "learning_rate": 5.971775505458444e-07,
      "loss": 0.005,
      "step": 1677
    },
    {
      "epoch": 2.6848,
      "grad_norm": 0.3561329460517029,
      "learning_rate": 5.913093872058528e-07,
      "loss": 0.0053,
      "step": 1678
    },
    {
      "epoch": 2.6864,
      "grad_norm": 0.2757271718152053,
      "learning_rate": 5.854693196441641e-07,
      "loss": 0.0034,
      "step": 1679
    },
    {
      "epoch": 2.6879999999999997,
      "grad_norm": 0.3106683728908216,
      "learning_rate": 5.796573653001091e-07,
      "loss": 0.0039,
      "step": 1680
    },
    {
      "epoch": 2.6896,
      "grad_norm": 0.3744773202118949,
      "learning_rate": 5.738735415290642e-07,
      "loss": 0.0037,
      "step": 1681
    },
    {
      "epoch": 2.6912000000000003,
      "grad_norm": 0.42079205797207647,
      "learning_rate": 5.681178656024055e-07,
      "loss": 0.0043,
      "step": 1682
    },
    {
      "epoch": 2.6928,
      "grad_norm": 0.22350666390277044,
      "learning_rate": 5.62390354707455e-07,
      "loss": 0.0022,
      "step": 1683
    },
    {
      "epoch": 2.6944,
      "grad_norm": 0.2899058545688104,
      "learning_rate": 5.56691025947429e-07,
      "loss": 0.0027,
      "step": 1684
    },
    {
      "epoch": 2.6959999999999997,
      "grad_norm": 0.2199050514310044,
      "learning_rate": 5.510198963413882e-07,
      "loss": 0.0033,
      "step": 1685
    },
    {
      "epoch": 2.6976,
      "grad_norm": 0.23732426830995915,
      "learning_rate": 5.453769828241872e-07,
      "loss": 0.0025,
      "step": 1686
    },
    {
      "epoch": 2.6992000000000003,
      "grad_norm": 0.4124371839877955,
      "learning_rate": 5.397623022464227e-07,
      "loss": 0.004,
      "step": 1687
    },
    {
      "epoch": 2.7008,
      "grad_norm": 0.334224332684429,
      "learning_rate": 5.341758713743828e-07,
      "loss": 0.004,
      "step": 1688
    },
    {
      "epoch": 2.7024,
      "grad_norm": 0.420696843510133,
      "learning_rate": 5.286177068899989e-07,
      "loss": 0.0047,
      "step": 1689
    },
    {
      "epoch": 2.7039999999999997,
      "grad_norm": 0.4228631696533493,
      "learning_rate": 5.230878253907911e-07,
      "loss": 0.0056,
      "step": 1690
    },
    {
      "epoch": 2.7056,
      "grad_norm": 0.2897142564004242,
      "learning_rate": 5.175862433898282e-07,
      "loss": 0.004,
      "step": 1691
    },
    {
      "epoch": 2.7072000000000003,
      "grad_norm": 0.3487682996311858,
      "learning_rate": 5.121129773156663e-07,
      "loss": 0.0048,
      "step": 1692
    },
    {
      "epoch": 2.7088,
      "grad_norm": 0.13490441372806583,
      "learning_rate": 5.066680435123106e-07,
      "loss": 0.0019,
      "step": 1693
    },
    {
      "epoch": 2.7104,
      "grad_norm": 0.15332938659980572,
      "learning_rate": 5.012514582391592e-07,
      "loss": 0.0017,
      "step": 1694
    },
    {
      "epoch": 2.7119999999999997,
      "grad_norm": 0.47657162395392066,
      "learning_rate": 4.95863237670956e-07,
      "loss": 0.0045,
      "step": 1695
    },
    {
      "epoch": 2.7136,
      "grad_norm": 0.3755987585052613,
      "learning_rate": 4.905033978977492e-07,
      "loss": 0.0045,
      "step": 1696
    },
    {
      "epoch": 2.7152,
      "grad_norm": 0.34581872345178366,
      "learning_rate": 4.851719549248301e-07,
      "loss": 0.0033,
      "step": 1697
    },
    {
      "epoch": 2.7168,
      "grad_norm": 0.3155072867452202,
      "learning_rate": 4.798689246727006e-07,
      "loss": 0.0033,
      "step": 1698
    },
    {
      "epoch": 2.7184,
      "grad_norm": 0.3187100892007427,
      "learning_rate": 4.7459432297701224e-07,
      "loss": 0.0059,
      "step": 1699
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 0.2505391599969973,
      "learning_rate": 4.693481655885257e-07,
      "loss": 0.0026,
      "step": 1700
    },
    {
      "epoch": 2.7216,
      "grad_norm": 0.2971684776745727,
      "learning_rate": 4.6413046817306404e-07,
      "loss": 0.0022,
      "step": 1701
    },
    {
      "epoch": 2.7232,
      "grad_norm": 0.4266222875524458,
      "learning_rate": 4.58941246311464e-07,
      "loss": 0.0042,
      "step": 1702
    },
    {
      "epoch": 2.7248,
      "grad_norm": 0.2588915506106927,
      "learning_rate": 4.5378051549952783e-07,
      "loss": 0.003,
      "step": 1703
    },
    {
      "epoch": 2.7264,
      "grad_norm": 0.29728353306995725,
      "learning_rate": 4.4864829114798394e-07,
      "loss": 0.003,
      "step": 1704
    },
    {
      "epoch": 2.7279999999999998,
      "grad_norm": 0.35270506873865676,
      "learning_rate": 4.4354458858242857e-07,
      "loss": 0.0042,
      "step": 1705
    },
    {
      "epoch": 2.7296,
      "grad_norm": 0.35164139940299416,
      "learning_rate": 4.384694230432984e-07,
      "loss": 0.005,
      "step": 1706
    },
    {
      "epoch": 2.7312,
      "grad_norm": 0.303702081177274,
      "learning_rate": 4.3342280968580287e-07,
      "loss": 0.0043,
      "step": 1707
    },
    {
      "epoch": 2.7328,
      "grad_norm": 0.4782687537873158,
      "learning_rate": 4.2840476357989825e-07,
      "loss": 0.0062,
      "step": 1708
    },
    {
      "epoch": 2.7344,
      "grad_norm": 0.22788466574501093,
      "learning_rate": 4.2341529971023253e-07,
      "loss": 0.0025,
      "step": 1709
    },
    {
      "epoch": 2.7359999999999998,
      "grad_norm": 0.3466322894758412,
      "learning_rate": 4.184544329761009e-07,
      "loss": 0.0036,
      "step": 1710
    },
    {
      "epoch": 2.7376,
      "grad_norm": 0.1948126746096502,
      "learning_rate": 4.1352217819140337e-07,
      "loss": 0.0023,
      "step": 1711
    },
    {
      "epoch": 2.7392,
      "grad_norm": 0.10559120452233281,
      "learning_rate": 4.0861855008460403e-07,
      "loss": 0.0013,
      "step": 1712
    },
    {
      "epoch": 2.7408,
      "grad_norm": 0.2509838023826191,
      "learning_rate": 4.037435632986786e-07,
      "loss": 0.0027,
      "step": 1713
    },
    {
      "epoch": 2.7424,
      "grad_norm": 0.432675046135986,
      "learning_rate": 3.988972323910778e-07,
      "loss": 0.0092,
      "step": 1714
    },
    {
      "epoch": 2.7439999999999998,
      "grad_norm": 0.4779228597178864,
      "learning_rate": 3.9407957183368093e-07,
      "loss": 0.0109,
      "step": 1715
    },
    {
      "epoch": 2.7456,
      "grad_norm": 0.1786314616239119,
      "learning_rate": 3.8929059601275463e-07,
      "loss": 0.0029,
      "step": 1716
    },
    {
      "epoch": 2.7472,
      "grad_norm": 0.31884951510899023,
      "learning_rate": 3.845303192289074e-07,
      "loss": 0.0035,
      "step": 1717
    },
    {
      "epoch": 2.7488,
      "grad_norm": 0.2333984870236776,
      "learning_rate": 3.797987556970495e-07,
      "loss": 0.0068,
      "step": 1718
    },
    {
      "epoch": 2.7504,
      "grad_norm": 0.22299594984545076,
      "learning_rate": 3.750959195463466e-07,
      "loss": 0.0028,
      "step": 1719
    },
    {
      "epoch": 2.752,
      "grad_norm": 0.3939060564745656,
      "learning_rate": 3.7042182482018074e-07,
      "loss": 0.0049,
      "step": 1720
    },
    {
      "epoch": 2.7536,
      "grad_norm": 0.47879957382724625,
      "learning_rate": 3.6577648547611033e-07,
      "loss": 0.0033,
      "step": 1721
    },
    {
      "epoch": 2.7552,
      "grad_norm": 0.24671324801670064,
      "learning_rate": 3.611599153858214e-07,
      "loss": 0.0029,
      "step": 1722
    },
    {
      "epoch": 2.7568,
      "grad_norm": 0.24115774832630246,
      "learning_rate": 3.5657212833509313e-07,
      "loss": 0.0023,
      "step": 1723
    },
    {
      "epoch": 2.7584,
      "grad_norm": 0.20611777074711515,
      "learning_rate": 3.520131380237546e-07,
      "loss": 0.0017,
      "step": 1724
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.2860716910321781,
      "learning_rate": 3.474829580656436e-07,
      "loss": 0.0032,
      "step": 1725
    },
    {
      "epoch": 2.7616,
      "grad_norm": 0.6103961839691567,
      "learning_rate": 3.429816019885657e-07,
      "loss": 0.0068,
      "step": 1726
    },
    {
      "epoch": 2.7632,
      "grad_norm": 0.18070944610210354,
      "learning_rate": 3.385090832342497e-07,
      "loss": 0.0015,
      "step": 1727
    },
    {
      "epoch": 2.7648,
      "grad_norm": 0.3073672440385467,
      "learning_rate": 3.3406541515832e-07,
      "loss": 0.0031,
      "step": 1728
    },
    {
      "epoch": 2.7664,
      "grad_norm": 0.23741522353964717,
      "learning_rate": 3.296506110302422e-07,
      "loss": 0.0026,
      "step": 1729
    },
    {
      "epoch": 2.768,
      "grad_norm": 0.29542755115257086,
      "learning_rate": 3.252646840332918e-07,
      "loss": 0.0037,
      "step": 1730
    },
    {
      "epoch": 2.7696,
      "grad_norm": 0.183430451355198,
      "learning_rate": 3.209076472645112e-07,
      "loss": 0.0018,
      "step": 1731
    },
    {
      "epoch": 2.7712,
      "grad_norm": 0.3426702018465107,
      "learning_rate": 3.16579513734675e-07,
      "loss": 0.003,
      "step": 1732
    },
    {
      "epoch": 2.7728,
      "grad_norm": 0.505331823250815,
      "learning_rate": 3.1228029636824477e-07,
      "loss": 0.0048,
      "step": 1733
    },
    {
      "epoch": 2.7744,
      "grad_norm": 0.19787976338126043,
      "learning_rate": 3.080100080033388e-07,
      "loss": 0.002,
      "step": 1734
    },
    {
      "epoch": 2.776,
      "grad_norm": 0.6953127204583504,
      "learning_rate": 3.037686613916857e-07,
      "loss": 0.0093,
      "step": 1735
    },
    {
      "epoch": 2.7776,
      "grad_norm": 0.32483452624924114,
      "learning_rate": 2.995562691985898e-07,
      "loss": 0.0028,
      "step": 1736
    },
    {
      "epoch": 2.7792,
      "grad_norm": 0.3534645641928358,
      "learning_rate": 2.9537284400289354e-07,
      "loss": 0.0033,
      "step": 1737
    },
    {
      "epoch": 2.7808,
      "grad_norm": 0.3465996865866739,
      "learning_rate": 2.9121839829693857e-07,
      "loss": 0.0067,
      "step": 1738
    },
    {
      "epoch": 2.7824,
      "grad_norm": 0.3650640452288819,
      "learning_rate": 2.8709294448653223e-07,
      "loss": 0.0041,
      "step": 1739
    },
    {
      "epoch": 2.784,
      "grad_norm": 0.27385479392048906,
      "learning_rate": 2.829964948909048e-07,
      "loss": 0.0037,
      "step": 1740
    },
    {
      "epoch": 2.7856,
      "grad_norm": 0.23809407069874067,
      "learning_rate": 2.7892906174267653e-07,
      "loss": 0.002,
      "step": 1741
    },
    {
      "epoch": 2.7872,
      "grad_norm": 0.20794968591921323,
      "learning_rate": 2.748906571878207e-07,
      "loss": 0.0024,
      "step": 1742
    },
    {
      "epoch": 2.7888,
      "grad_norm": 0.5666314290271737,
      "learning_rate": 2.708812932856253e-07,
      "loss": 0.0041,
      "step": 1743
    },
    {
      "epoch": 2.7904,
      "grad_norm": 0.4194997875269323,
      "learning_rate": 2.6690098200866097e-07,
      "loss": 0.0055,
      "step": 1744
    },
    {
      "epoch": 2.792,
      "grad_norm": 0.2758212918383284,
      "learning_rate": 2.6294973524274127e-07,
      "loss": 0.0028,
      "step": 1745
    },
    {
      "epoch": 2.7936,
      "grad_norm": 0.3627580297339781,
      "learning_rate": 2.5902756478688674e-07,
      "loss": 0.0051,
      "step": 1746
    },
    {
      "epoch": 2.7952,
      "grad_norm": 0.3392983160782683,
      "learning_rate": 2.551344823532964e-07,
      "loss": 0.0036,
      "step": 1747
    },
    {
      "epoch": 2.7968,
      "grad_norm": 0.4412941334882733,
      "learning_rate": 2.5127049956730207e-07,
      "loss": 0.009,
      "step": 1748
    },
    {
      "epoch": 2.7984,
      "grad_norm": 0.28087146012137015,
      "learning_rate": 2.474356279673462e-07,
      "loss": 0.003,
      "step": 1749
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.23433563341090205,
      "learning_rate": 2.436298790049363e-07,
      "loss": 0.0022,
      "step": 1750
    },
    {
      "epoch": 2.8016,
      "grad_norm": 0.38528540201664624,
      "learning_rate": 2.398532640446161e-07,
      "loss": 0.0065,
      "step": 1751
    },
    {
      "epoch": 2.8032,
      "grad_norm": 0.3196802362593276,
      "learning_rate": 2.3610579436392999e-07,
      "loss": 0.004,
      "step": 1752
    },
    {
      "epoch": 2.8048,
      "grad_norm": 0.35865307250947054,
      "learning_rate": 2.3238748115339327e-07,
      "loss": 0.0035,
      "step": 1753
    },
    {
      "epoch": 2.8064,
      "grad_norm": 0.2520640853349629,
      "learning_rate": 2.2869833551645293e-07,
      "loss": 0.0024,
      "step": 1754
    },
    {
      "epoch": 2.808,
      "grad_norm": 0.31570000493873085,
      "learning_rate": 2.2503836846945792e-07,
      "loss": 0.0032,
      "step": 1755
    },
    {
      "epoch": 2.8096,
      "grad_norm": 0.2822847318578194,
      "learning_rate": 2.2140759094162468e-07,
      "loss": 0.0036,
      "step": 1756
    },
    {
      "epoch": 2.8112,
      "grad_norm": 0.2574367021470435,
      "learning_rate": 2.178060137750071e-07,
      "loss": 0.0038,
      "step": 1757
    },
    {
      "epoch": 2.8128,
      "grad_norm": 0.3104942552175079,
      "learning_rate": 2.1423364772445886e-07,
      "loss": 0.0033,
      "step": 1758
    },
    {
      "epoch": 2.8144,
      "grad_norm": 0.19123508252467655,
      "learning_rate": 2.106905034576112e-07,
      "loss": 0.0027,
      "step": 1759
    },
    {
      "epoch": 2.816,
      "grad_norm": 0.31883454012133566,
      "learning_rate": 2.071765915548274e-07,
      "loss": 0.0045,
      "step": 1760
    },
    {
      "epoch": 2.8176,
      "grad_norm": 0.1846107453424149,
      "learning_rate": 2.036919225091827e-07,
      "loss": 0.0031,
      "step": 1761
    },
    {
      "epoch": 2.8192,
      "grad_norm": 0.25595674556604037,
      "learning_rate": 2.002365067264289e-07,
      "loss": 0.0026,
      "step": 1762
    },
    {
      "epoch": 2.8208,
      "grad_norm": 0.2890342090918847,
      "learning_rate": 1.9681035452496112e-07,
      "loss": 0.0042,
      "step": 1763
    },
    {
      "epoch": 2.8224,
      "grad_norm": 0.23682918205933798,
      "learning_rate": 1.9341347613579086e-07,
      "loss": 0.0039,
      "step": 1764
    },
    {
      "epoch": 2.824,
      "grad_norm": 0.3958425374941724,
      "learning_rate": 1.900458817025097e-07,
      "loss": 0.0034,
      "step": 1765
    },
    {
      "epoch": 2.8256,
      "grad_norm": 0.16606697252502298,
      "learning_rate": 1.867075812812691e-07,
      "loss": 0.0019,
      "step": 1766
    },
    {
      "epoch": 2.8272,
      "grad_norm": 0.24363351954827045,
      "learning_rate": 1.8339858484073935e-07,
      "loss": 0.002,
      "step": 1767
    },
    {
      "epoch": 2.8288,
      "grad_norm": 0.44218824058715006,
      "learning_rate": 1.8011890226208527e-07,
      "loss": 0.0055,
      "step": 1768
    },
    {
      "epoch": 2.8304,
      "grad_norm": 0.5250979322757067,
      "learning_rate": 1.7686854333893833e-07,
      "loss": 0.0073,
      "step": 1769
    },
    {
      "epoch": 2.832,
      "grad_norm": 0.3967675739130118,
      "learning_rate": 1.7364751777736334e-07,
      "loss": 0.0043,
      "step": 1770
    },
    {
      "epoch": 2.8336,
      "grad_norm": 0.235277710668951,
      "learning_rate": 1.7045583519583075e-07,
      "loss": 0.0023,
      "step": 1771
    },
    {
      "epoch": 2.8352,
      "grad_norm": 0.24237221288449026,
      "learning_rate": 1.6729350512519006e-07,
      "loss": 0.0019,
      "step": 1772
    },
    {
      "epoch": 2.8368,
      "grad_norm": 0.21388220559863358,
      "learning_rate": 1.6416053700863965e-07,
      "loss": 0.0026,
      "step": 1773
    },
    {
      "epoch": 2.8384,
      "grad_norm": 0.22403363096859452,
      "learning_rate": 1.6105694020169594e-07,
      "loss": 0.003,
      "step": 1774
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.2578079621489086,
      "learning_rate": 1.5798272397217097e-07,
      "loss": 0.0025,
      "step": 1775
    },
    {
      "epoch": 2.8416,
      "grad_norm": 0.22062263793996525,
      "learning_rate": 1.5493789750014032e-07,
      "loss": 0.0019,
      "step": 1776
    },
    {
      "epoch": 2.8432,
      "grad_norm": 0.3146290994127335,
      "learning_rate": 1.519224698779198e-07,
      "loss": 0.004,
      "step": 1777
    },
    {
      "epoch": 2.8448,
      "grad_norm": 0.18928296544669773,
      "learning_rate": 1.489364501100332e-07,
      "loss": 0.0017,
      "step": 1778
    },
    {
      "epoch": 2.8464,
      "grad_norm": 0.45541486559528266,
      "learning_rate": 1.459798471131868e-07,
      "loss": 0.0095,
      "step": 1779
    },
    {
      "epoch": 2.848,
      "grad_norm": 0.3656356148384636,
      "learning_rate": 1.430526697162482e-07,
      "loss": 0.0036,
      "step": 1780
    },
    {
      "epoch": 2.8496,
      "grad_norm": 0.29965165814152733,
      "learning_rate": 1.4015492666021313e-07,
      "loss": 0.0058,
      "step": 1781
    },
    {
      "epoch": 2.8512,
      "grad_norm": 0.14044524917062876,
      "learning_rate": 1.3728662659818205e-07,
      "loss": 0.0013,
      "step": 1782
    },
    {
      "epoch": 2.8528000000000002,
      "grad_norm": 0.3106923022270456,
      "learning_rate": 1.344477780953346e-07,
      "loss": 0.0025,
      "step": 1783
    },
    {
      "epoch": 2.8544,
      "grad_norm": 0.4012983707274509,
      "learning_rate": 1.3163838962890196e-07,
      "loss": 0.0035,
      "step": 1784
    },
    {
      "epoch": 2.856,
      "grad_norm": 0.23270638672389216,
      "learning_rate": 1.2885846958814673e-07,
      "loss": 0.0027,
      "step": 1785
    },
    {
      "epoch": 2.8576,
      "grad_norm": 0.39470098855170954,
      "learning_rate": 1.2610802627432972e-07,
      "loss": 0.0051,
      "step": 1786
    },
    {
      "epoch": 2.8592,
      "grad_norm": 0.2895920263603846,
      "learning_rate": 1.2338706790069433e-07,
      "loss": 0.0041,
      "step": 1787
    },
    {
      "epoch": 2.8608000000000002,
      "grad_norm": 0.455922258530345,
      "learning_rate": 1.206956025924333e-07,
      "loss": 0.0078,
      "step": 1788
    },
    {
      "epoch": 2.8624,
      "grad_norm": 0.38409334305045495,
      "learning_rate": 1.1803363838667092e-07,
      "loss": 0.0031,
      "step": 1789
    },
    {
      "epoch": 2.864,
      "grad_norm": 0.3160169405210605,
      "learning_rate": 1.1540118323243866e-07,
      "loss": 0.0032,
      "step": 1790
    },
    {
      "epoch": 2.8656,
      "grad_norm": 0.33398026700734806,
      "learning_rate": 1.1279824499064396e-07,
      "loss": 0.0052,
      "step": 1791
    },
    {
      "epoch": 2.8672,
      "grad_norm": 0.31053508960210263,
      "learning_rate": 1.1022483143405705e-07,
      "loss": 0.0041,
      "step": 1792
    },
    {
      "epoch": 2.8688000000000002,
      "grad_norm": 0.2165345269428243,
      "learning_rate": 1.0768095024728309e-07,
      "loss": 0.0023,
      "step": 1793
    },
    {
      "epoch": 2.8704,
      "grad_norm": 0.294364409308341,
      "learning_rate": 1.0516660902673448e-07,
      "loss": 0.0026,
      "step": 1794
    },
    {
      "epoch": 2.872,
      "grad_norm": 0.31868904443652213,
      "learning_rate": 1.0268181528061749e-07,
      "loss": 0.0036,
      "step": 1795
    },
    {
      "epoch": 2.8736,
      "grad_norm": 0.27411807331233073,
      "learning_rate": 1.0022657642890232e-07,
      "loss": 0.0021,
      "step": 1796
    },
    {
      "epoch": 2.8752,
      "grad_norm": 0.22469651822799042,
      "learning_rate": 9.780089980330643e-08,
      "loss": 0.002,
      "step": 1797
    },
    {
      "epoch": 2.8768000000000002,
      "grad_norm": 0.5504478236454433,
      "learning_rate": 9.540479264726676e-08,
      "loss": 0.0065,
      "step": 1798
    },
    {
      "epoch": 2.8784,
      "grad_norm": 0.1794435496363775,
      "learning_rate": 9.303826211592316e-08,
      "loss": 0.0016,
      "step": 1799
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.2654843469205127,
      "learning_rate": 9.070131527609604e-08,
      "loss": 0.0037,
      "step": 1800
    },
    {
      "epoch": 2.8816,
      "grad_norm": 0.24419522322704548,
      "learning_rate": 8.839395910626214e-08,
      "loss": 0.0025,
      "step": 1801
    },
    {
      "epoch": 2.8832,
      "grad_norm": 0.22986470580386867,
      "learning_rate": 8.61162004965388e-08,
      "loss": 0.0023,
      "step": 1802
    },
    {
      "epoch": 2.8848000000000003,
      "grad_norm": 0.16419135289067946,
      "learning_rate": 8.386804624865851e-08,
      "loss": 0.0017,
      "step": 1803
    },
    {
      "epoch": 2.8864,
      "grad_norm": 0.15006603379648195,
      "learning_rate": 8.16495030759501e-08,
      "loss": 0.002,
      "step": 1804
    },
    {
      "epoch": 2.888,
      "grad_norm": 0.2881652846675583,
      "learning_rate": 7.946057760332193e-08,
      "loss": 0.0037,
      "step": 1805
    },
    {
      "epoch": 2.8895999999999997,
      "grad_norm": 0.4238189600001012,
      "learning_rate": 7.730127636723539e-08,
      "loss": 0.008,
      "step": 1806
    },
    {
      "epoch": 2.8912,
      "grad_norm": 0.5138563588810117,
      "learning_rate": 7.517160581569371e-08,
      "loss": 0.0051,
      "step": 1807
    },
    {
      "epoch": 2.8928000000000003,
      "grad_norm": 0.5308687748429286,
      "learning_rate": 7.307157230821426e-08,
      "loss": 0.0081,
      "step": 1808
    },
    {
      "epoch": 2.8944,
      "grad_norm": 0.5921440444783744,
      "learning_rate": 7.100118211581852e-08,
      "loss": 0.0049,
      "step": 1809
    },
    {
      "epoch": 2.896,
      "grad_norm": 0.31771549916104036,
      "learning_rate": 6.896044142100433e-08,
      "loss": 0.0039,
      "step": 1810
    },
    {
      "epoch": 2.8975999999999997,
      "grad_norm": 0.3052563999741175,
      "learning_rate": 6.694935631773259e-08,
      "loss": 0.0035,
      "step": 1811
    },
    {
      "epoch": 2.8992,
      "grad_norm": 0.3725902993944454,
      "learning_rate": 6.496793281141056e-08,
      "loss": 0.0063,
      "step": 1812
    },
    {
      "epoch": 2.9008000000000003,
      "grad_norm": 0.3102151511513018,
      "learning_rate": 6.301617681886863e-08,
      "loss": 0.0038,
      "step": 1813
    },
    {
      "epoch": 2.9024,
      "grad_norm": 0.2182676130139072,
      "learning_rate": 6.109409416834689e-08,
      "loss": 0.0023,
      "step": 1814
    },
    {
      "epoch": 2.904,
      "grad_norm": 0.260922445485721,
      "learning_rate": 5.920169059947412e-08,
      "loss": 0.0028,
      "step": 1815
    },
    {
      "epoch": 2.9055999999999997,
      "grad_norm": 0.27559308235578495,
      "learning_rate": 5.7338971763256646e-08,
      "loss": 0.0038,
      "step": 1816
    },
    {
      "epoch": 2.9072,
      "grad_norm": 0.35145438615648,
      "learning_rate": 5.5505943222055046e-08,
      "loss": 0.005,
      "step": 1817
    },
    {
      "epoch": 2.9088000000000003,
      "grad_norm": 0.35854604442106486,
      "learning_rate": 5.37026104495697e-08,
      "loss": 0.0042,
      "step": 1818
    },
    {
      "epoch": 2.9104,
      "grad_norm": 0.37939974541440513,
      "learning_rate": 5.192897883082748e-08,
      "loss": 0.0044,
      "step": 1819
    },
    {
      "epoch": 2.912,
      "grad_norm": 0.21722654227864255,
      "learning_rate": 5.0185053662161756e-08,
      "loss": 0.0025,
      "step": 1820
    },
    {
      "epoch": 2.9135999999999997,
      "grad_norm": 0.35596655693883184,
      "learning_rate": 4.8470840151195745e-08,
      "loss": 0.0049,
      "step": 1821
    },
    {
      "epoch": 2.9152,
      "grad_norm": 0.25434248568517853,
      "learning_rate": 4.678634341683252e-08,
      "loss": 0.0025,
      "step": 1822
    },
    {
      "epoch": 2.9168,
      "grad_norm": 0.24723308450126122,
      "learning_rate": 4.513156848923616e-08,
      "loss": 0.0018,
      "step": 1823
    },
    {
      "epoch": 2.9184,
      "grad_norm": 0.3699984409531007,
      "learning_rate": 4.350652030981395e-08,
      "loss": 0.0046,
      "step": 1824
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.8110138980784226,
      "learning_rate": 4.19112037312075e-08,
      "loss": 0.0049,
      "step": 1825
    },
    {
      "epoch": 2.9215999999999998,
      "grad_norm": 0.4837641134180625,
      "learning_rate": 4.0345623517273894e-08,
      "loss": 0.0055,
      "step": 1826
    },
    {
      "epoch": 2.9232,
      "grad_norm": 0.21993881647963598,
      "learning_rate": 3.8809784343072364e-08,
      "loss": 0.0024,
      "step": 1827
    },
    {
      "epoch": 2.9248,
      "grad_norm": 0.36956005696857025,
      "learning_rate": 3.7303690794854296e-08,
      "loss": 0.0027,
      "step": 1828
    },
    {
      "epoch": 2.9264,
      "grad_norm": 0.2819053942346473,
      "learning_rate": 3.582734737004101e-08,
      "loss": 0.0042,
      "step": 1829
    },
    {
      "epoch": 2.928,
      "grad_norm": 0.2761653957431238,
      "learning_rate": 3.438075847721933e-08,
      "loss": 0.0037,
      "step": 1830
    },
    {
      "epoch": 2.9295999999999998,
      "grad_norm": 0.4670080908820644,
      "learning_rate": 3.2963928436122726e-08,
      "loss": 0.008,
      "step": 1831
    },
    {
      "epoch": 2.9312,
      "grad_norm": 0.2297603417125354,
      "learning_rate": 3.157686147762129e-08,
      "loss": 0.0045,
      "step": 1832
    },
    {
      "epoch": 2.9328,
      "grad_norm": 0.38004580779980107,
      "learning_rate": 3.0219561743707326e-08,
      "loss": 0.0027,
      "step": 1833
    },
    {
      "epoch": 2.9344,
      "grad_norm": 0.3030335150073986,
      "learning_rate": 2.8892033287484245e-08,
      "loss": 0.0025,
      "step": 1834
    },
    {
      "epoch": 2.936,
      "grad_norm": 0.28940707302564167,
      "learning_rate": 2.7594280073152123e-08,
      "loss": 0.0051,
      "step": 1835
    },
    {
      "epoch": 2.9375999999999998,
      "grad_norm": 0.2527949134917213,
      "learning_rate": 2.6326305976001054e-08,
      "loss": 0.0031,
      "step": 1836
    },
    {
      "epoch": 2.9392,
      "grad_norm": 0.4091815899064453,
      "learning_rate": 2.508811478239226e-08,
      "loss": 0.005,
      "step": 1837
    },
    {
      "epoch": 2.9408,
      "grad_norm": 0.28684768340282324,
      "learning_rate": 2.3879710189753657e-08,
      "loss": 0.0031,
      "step": 1838
    },
    {
      "epoch": 2.9424,
      "grad_norm": 0.3642277160058468,
      "learning_rate": 2.2701095806565432e-08,
      "loss": 0.0049,
      "step": 1839
    },
    {
      "epoch": 2.944,
      "grad_norm": 0.28489682761878427,
      "learning_rate": 2.1552275152346702e-08,
      "loss": 0.0026,
      "step": 1840
    },
    {
      "epoch": 2.9455999999999998,
      "grad_norm": 0.3919995396630444,
      "learning_rate": 2.0433251657653307e-08,
      "loss": 0.0077,
      "step": 1841
    },
    {
      "epoch": 2.9472,
      "grad_norm": 0.2654161369122685,
      "learning_rate": 1.9344028664056715e-08,
      "loss": 0.0026,
      "step": 1842
    },
    {
      "epoch": 2.9488,
      "grad_norm": 0.11667868222023793,
      "learning_rate": 1.8284609424142897e-08,
      "loss": 0.0012,
      "step": 1843
    },
    {
      "epoch": 2.9504,
      "grad_norm": 0.2708149125559501,
      "learning_rate": 1.7254997101500137e-08,
      "loss": 0.0058,
      "step": 1844
    },
    {
      "epoch": 2.952,
      "grad_norm": 0.22510298505002913,
      "learning_rate": 1.6255194770704586e-08,
      "loss": 0.0023,
      "step": 1845
    },
    {
      "epoch": 2.9536,
      "grad_norm": 0.47317677070689074,
      "learning_rate": 1.528520541731915e-08,
      "loss": 0.0109,
      "step": 1846
    },
    {
      "epoch": 2.9552,
      "grad_norm": 0.3557026006553359,
      "learning_rate": 1.4345031937879061e-08,
      "loss": 0.0032,
      "step": 1847
    },
    {
      "epoch": 2.9568,
      "grad_norm": 0.23021052365253492,
      "learning_rate": 1.3434677139885222e-08,
      "loss": 0.003,
      "step": 1848
    },
    {
      "epoch": 2.9584,
      "grad_norm": 0.35060407682918143,
      "learning_rate": 1.2554143741795311e-08,
      "loss": 0.0047,
      "step": 1849
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.21884835184223417,
      "learning_rate": 1.170343437301491e-08,
      "loss": 0.0025,
      "step": 1850
    },
    {
      "epoch": 2.9616,
      "grad_norm": 0.29393714548985944,
      "learning_rate": 1.0882551573891953e-08,
      "loss": 0.0072,
      "step": 1851
    },
    {
      "epoch": 2.9632,
      "grad_norm": 0.422476275688548,
      "learning_rate": 1.0091497795706728e-08,
      "loss": 0.005,
      "step": 1852
    },
    {
      "epoch": 2.9648,
      "grad_norm": 0.23338307459392,
      "learning_rate": 9.330275400666334e-09,
      "loss": 0.0031,
      "step": 1853
    },
    {
      "epoch": 2.9664,
      "grad_norm": 0.2891855387233942,
      "learning_rate": 8.59888666189579e-09,
      "loss": 0.0038,
      "step": 1854
    },
    {
      "epoch": 2.968,
      "grad_norm": 0.41158431954259306,
      "learning_rate": 7.897333763433601e-09,
      "loss": 0.0067,
      "step": 1855
    },
    {
      "epoch": 2.9696,
      "grad_norm": 0.4170226848139635,
      "learning_rate": 7.225618800222878e-09,
      "loss": 0.0028,
      "step": 1856
    },
    {
      "epoch": 2.9712,
      "grad_norm": 0.20532199418262084,
      "learning_rate": 6.583743778106888e-09,
      "loss": 0.0025,
      "step": 1857
    },
    {
      "epoch": 2.9728,
      "grad_norm": 0.4447779548881662,
      "learning_rate": 5.971710613821291e-09,
      "loss": 0.0048,
      "step": 1858
    },
    {
      "epoch": 2.9744,
      "grad_norm": 0.31124710271313133,
      "learning_rate": 5.3895211349896946e-09,
      "loss": 0.0027,
      "step": 1859
    },
    {
      "epoch": 2.976,
      "grad_norm": 0.342272098849646,
      "learning_rate": 4.837177080119215e-09,
      "loss": 0.005,
      "step": 1860
    },
    {
      "epoch": 2.9776,
      "grad_norm": 0.32407596246883896,
      "learning_rate": 4.314680098592705e-09,
      "loss": 0.0037,
      "step": 1861
    },
    {
      "epoch": 2.9792,
      "grad_norm": 0.2816315966016677,
      "learning_rate": 3.8220317506654226e-09,
      "loss": 0.0034,
      "step": 1862
    },
    {
      "epoch": 2.9808,
      "grad_norm": 0.2839306857760757,
      "learning_rate": 3.3592335074594805e-09,
      "loss": 0.003,
      "step": 1863
    },
    {
      "epoch": 2.9824,
      "grad_norm": 0.2245985260963881,
      "learning_rate": 2.9262867509605164e-09,
      "loss": 0.0028,
      "step": 1864
    },
    {
      "epoch": 2.984,
      "grad_norm": 0.17332393727737602,
      "learning_rate": 2.5231927740154705e-09,
      "loss": 0.0019,
      "step": 1865
    },
    {
      "epoch": 2.9856,
      "grad_norm": 0.30104904584694364,
      "learning_rate": 2.149952780321485e-09,
      "loss": 0.0026,
      "step": 1866
    },
    {
      "epoch": 2.9872,
      "grad_norm": 0.26045479648076253,
      "learning_rate": 1.8065678844314538e-09,
      "loss": 0.0026,
      "step": 1867
    },
    {
      "epoch": 2.9888,
      "grad_norm": 0.2864632662273618,
      "learning_rate": 1.4930391117451427e-09,
      "loss": 0.0033,
      "step": 1868
    },
    {
      "epoch": 2.9904,
      "grad_norm": 0.1659935085129691,
      "learning_rate": 1.209367398504746e-09,
      "loss": 0.0018,
      "step": 1869
    },
    {
      "epoch": 2.992,
      "grad_norm": 0.26353924511099674,
      "learning_rate": 9.555535917993297e-10,
      "loss": 0.0028,
      "step": 1870
    },
    {
      "epoch": 2.9936,
      "grad_norm": 0.21139113799467085,
      "learning_rate": 7.315984495548378e-10,
      "loss": 0.002,
      "step": 1871
    },
    {
      "epoch": 2.9952,
      "grad_norm": 0.4251269053335163,
      "learning_rate": 5.375026405352035e-10,
      "loss": 0.0039,
      "step": 1872
    },
    {
      "epoch": 2.9968,
      "grad_norm": 0.3840411035653194,
      "learning_rate": 3.732667443390181e-10,
      "loss": 0.0036,
      "step": 1873
    },
    {
      "epoch": 2.9984,
      "grad_norm": 0.3308770694702444,
      "learning_rate": 2.388912514017516e-10,
      "loss": 0.0037,
      "step": 1874
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.38326237286825415,
      "learning_rate": 1.3437656298687096e-10,
      "loss": 0.0038,
      "step": 1875
    },
    {
      "epoch": 3.0,
      "step": 1875,
      "total_flos": 75543577763840.0,
      "train_loss": 0.05507048086337745,
      "train_runtime": 6924.8564,
      "train_samples_per_second": 4.332,
      "train_steps_per_second": 0.271
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1875,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 50000,
  "total_flos": 75543577763840.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}