{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.0,
  "eval_steps": 500,
  "global_step": 399,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.422,
      "step": 1
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.5154,
      "step": 2
    },
    {
      "epoch": 0.05,
      "learning_rate": 5e-06,
      "loss": 0.3768,
      "step": 3
    },
    {
      "epoch": 0.07,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.356,
      "step": 4
    },
    {
      "epoch": 0.09,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.2775,
      "step": 5
    },
    {
      "epoch": 0.11,
      "learning_rate": 1e-05,
      "loss": 0.1761,
      "step": 6
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 0.218,
      "step": 7
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.2161,
      "step": 8
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.2634,
      "step": 9
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.2294,
      "step": 10
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 0.1694,
      "step": 11
    },
    {
      "epoch": 0.21,
      "learning_rate": 2e-05,
      "loss": 0.1751,
      "step": 12
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9999670507574945e-05,
      "loss": 0.1785,
      "step": 13
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.999868205201284e-05,
      "loss": 0.1852,
      "step": 14
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9997034698451396e-05,
      "loss": 0.1508,
      "step": 15
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9994728555448723e-05,
      "loss": 0.1505,
      "step": 16
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.999176377497616e-05,
      "loss": 0.1586,
      "step": 17
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.998814055240823e-05,
      "loss": 0.1438,
      "step": 18
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9983859126509827e-05,
      "loss": 0.1479,
      "step": 19
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9978919779420427e-05,
      "loss": 0.1612,
      "step": 20
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9973322836635517e-05,
      "loss": 0.1547,
      "step": 21
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.996706866698515e-05,
      "loss": 0.1871,
      "step": 22
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9960157682609634e-05,
      "loss": 0.136,
      "step": 23
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.995259033893236e-05,
      "loss": 0.1541,
      "step": 24
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.994436713462982e-05,
      "loss": 0.1466,
      "step": 25
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9935488611598716e-05,
      "loss": 0.1409,
      "step": 26
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9925955354920265e-05,
      "loss": 0.1455,
      "step": 27
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9915767992821642e-05,
      "loss": 0.1687,
      "step": 28
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9904927196634572e-05,
      "loss": 0.1488,
      "step": 29
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.9893433680751105e-05,
      "loss": 0.1545,
      "step": 30
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.9881288202576518e-05,
      "loss": 0.1533,
      "step": 31
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.986849156247943e-05,
      "loss": 0.158,
      "step": 32
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.985504460373903e-05,
      "loss": 0.1545,
      "step": 33
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.9840948212489527e-05,
      "loss": 0.1597,
      "step": 34
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.9826203317661758e-05,
      "loss": 0.1556,
      "step": 35
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.9810810890921943e-05,
      "loss": 0.1581,
      "step": 36
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.979477194660769e-05,
      "loss": 0.1625,
      "step": 37
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.977808754166113e-05,
      "loss": 0.1562,
      "step": 38
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.9760758775559275e-05,
      "loss": 0.1503,
      "step": 39
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.9742786790241548e-05,
      "loss": 0.1549,
      "step": 40
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.9724172770034566e-05,
      "loss": 0.1736,
      "step": 41
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.9704917941574053e-05,
      "loss": 0.1537,
      "step": 42
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.9685023573724036e-05,
      "loss": 0.1483,
      "step": 43
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.9664490977493223e-05,
      "loss": 0.1675,
      "step": 44
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.9643321505948588e-05,
      "loss": 0.1436,
      "step": 45
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.962151655412624e-05,
      "loss": 0.1536,
      "step": 46
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.9599077558939468e-05,
      "loss": 0.146,
      "step": 47
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.957600599908406e-05,
      "loss": 0.1559,
      "step": 48
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.9552303394940862e-05,
      "loss": 0.1508,
      "step": 49
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.952797130847557e-05,
      "loss": 0.1468,
      "step": 50
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.9503011343135828e-05,
      "loss": 0.1562,
      "step": 51
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.947742514374553e-05,
      "loss": 0.174,
      "step": 52
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.9451214396396453e-05,
      "loss": 0.1482,
      "step": 53
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.9424380828337146e-05,
      "loss": 0.1531,
      "step": 54
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 0.1744,
      "step": 55
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.9368852344180168e-05,
      "loss": 0.1439,
      "step": 56
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.9340161087325483e-05,
      "loss": 0.1617,
      "step": 57
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.9310854328005383e-05,
      "loss": 0.1442,
      "step": 58
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.9280933997490912e-05,
      "loss": 0.1461,
      "step": 59
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.9250402067486523e-05,
      "loss": 0.1375,
      "step": 60
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.9219260550000144e-05,
      "loss": 0.1385,
      "step": 61
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.91875114972106e-05,
      "loss": 0.1347,
      "step": 62
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.9155157001332374e-05,
      "loss": 0.1273,
      "step": 63
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.9122199194477723e-05,
      "loss": 0.1539,
      "step": 64
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.9088640248516185e-05,
      "loss": 0.1375,
      "step": 65
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.905448237493147e-05,
      "loss": 0.1694,
      "step": 66
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.9019727824675686e-05,
      "loss": 0.1813,
      "step": 67
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.8984378888021045e-05,
      "loss": 0.1409,
      "step": 68
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.894843789440892e-05,
      "loss": 0.1333,
      "step": 69
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.8911907212296343e-05,
      "loss": 0.1657,
      "step": 70
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.8874789248999915e-05,
      "loss": 0.1627,
      "step": 71
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.8837086450537195e-05,
      "loss": 0.14,
      "step": 72
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.879880130146547e-05,
      "loss": 0.1388,
      "step": 73
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.8759936324718068e-05,
      "loss": 0.1353,
      "step": 74
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.872049408143808e-05,
      "loss": 0.1495,
      "step": 75
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.8680477170809573e-05,
      "loss": 0.1539,
      "step": 76
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.8639888229886342e-05,
      "loss": 0.1406,
      "step": 77
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.8598729933418102e-05,
      "loss": 0.1413,
      "step": 78
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.855700499367423e-05,
      "loss": 0.1384,
      "step": 79
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.851471616026505e-05,
      "loss": 0.1638,
      "step": 80
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.8471866219960604e-05,
      "loss": 0.1602,
      "step": 81
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.8428457996507053e-05,
      "loss": 0.1592,
      "step": 82
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.8384494350440553e-05,
      "loss": 0.1455,
      "step": 83
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.833997817889878e-05,
      "loss": 0.1575,
      "step": 84
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.8294912415429995e-05,
      "loss": 0.1343,
      "step": 85
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.8249300029799735e-05,
      "loss": 0.1668,
      "step": 86
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.820314402779511e-05,
      "loss": 0.1696,
      "step": 87
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.815644745102673e-05,
      "loss": 0.1436,
      "step": 88
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.810921337672826e-05,
      "loss": 0.1473,
      "step": 89
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.806144491755363e-05,
      "loss": 0.145,
      "step": 90
    },
    {
      "epoch": 1.6,
      "learning_rate": 1.8013145221371937e-05,
      "loss": 0.136,
      "step": 91
    },
    {
      "epoch": 1.61,
      "learning_rate": 1.7964317471059982e-05,
      "loss": 0.142,
      "step": 92
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.7914964884292543e-05,
      "loss": 0.1548,
      "step": 93
    },
    {
      "epoch": 1.65,
      "learning_rate": 1.7865090713330313e-05,
      "loss": 0.1357,
      "step": 94
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.7814698244805605e-05,
      "loss": 0.1433,
      "step": 95
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.7763790799505746e-05,
      "loss": 0.1339,
      "step": 96
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.771237173215426e-05,
      "loss": 0.1215,
      "step": 97
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.766044443118978e-05,
      "loss": 0.1359,
      "step": 98
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.760801231854278e-05,
      "loss": 0.136,
      "step": 99
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.7555078849410044e-05,
      "loss": 0.1477,
      "step": 100
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.7501647512026996e-05,
      "loss": 0.137,
      "step": 101
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.744772182743782e-05,
      "loss": 0.1322,
      "step": 102
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.7393305349263433e-05,
      "loss": 0.1319,
      "step": 103
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.733840166346731e-05,
      "loss": 0.139,
      "step": 104
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.728301438811916e-05,
      "loss": 0.1355,
      "step": 105
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.7227147173156525e-05,
      "loss": 0.1202,
      "step": 106
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.7170803700144227e-05,
      "loss": 0.1146,
      "step": 107
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.711398768203178e-05,
      "loss": 0.1274,
      "step": 108
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.7056702862908704e-05,
      "loss": 0.1246,
      "step": 109
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.6998953017757787e-05,
      "loss": 0.1471,
      "step": 110
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.6940741952206342e-05,
      "loss": 0.1569,
      "step": 111
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.6882073502275394e-05,
      "loss": 0.1115,
      "step": 112
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.682295153412691e-05,
      "loss": 0.1191,
      "step": 113
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.676337994380903e-05,
      "loss": 0.1409,
      "step": 114
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.67033626569993e-05,
      "loss": 0.1032,
      "step": 115
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.6642903628745995e-05,
      "loss": 0.1077,
      "step": 116
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.658200684320748e-05,
      "loss": 0.1096,
      "step": 117
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.652067631338967e-05,
      "loss": 0.1263,
      "step": 118
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.6458916080881566e-05,
      "loss": 0.1081,
      "step": 119
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.6396730215588913e-05,
      "loss": 0.1276,
      "step": 120
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.6334122815466035e-05,
      "loss": 0.1161,
      "step": 121
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.6271098006245742e-05,
      "loss": 0.1141,
      "step": 122
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.6207659941167485e-05,
      "loss": 0.121,
      "step": 123
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.6143812800703646e-05,
      "loss": 0.093,
      "step": 124
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.6079560792284046e-05,
      "loss": 0.131,
      "step": 125
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.6014908150018703e-05,
      "loss": 0.0955,
      "step": 126
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5949859134418798e-05,
      "loss": 0.1054,
      "step": 127
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.5884418032115907e-05,
      "loss": 0.1262,
      "step": 128
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.581858915557953e-05,
      "loss": 0.1168,
      "step": 129
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.57523768428329e-05,
      "loss": 0.1003,
      "step": 130
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.5685785457167114e-05,
      "loss": 0.111,
      "step": 131
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.5618819386853607e-05,
      "loss": 0.1134,
      "step": 132
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.5551483044854954e-05,
      "loss": 0.1193,
      "step": 133
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.5483780868534087e-05,
      "loss": 0.1217,
      "step": 134
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.541571731936185e-05,
      "loss": 0.1028,
      "step": 135
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.534729688262302e-05,
      "loss": 0.115,
      "step": 136
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.527852406712072e-05,
      "loss": 0.1047,
      "step": 137
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.5209403404879305e-05,
      "loss": 0.1238,
      "step": 138
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.5139939450845699e-05,
      "loss": 0.1031,
      "step": 139
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.5070136782589236e-05,
      "loss": 0.0983,
      "step": 140
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.1298,
      "step": 141
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.4929533724985712e-05,
      "loss": 0.1263,
      "step": 142
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.485874260116714e-05,
      "loss": 0.1331,
      "step": 143
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.4787631293572094e-05,
      "loss": 0.1134,
      "step": 144
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.4716204488328007e-05,
      "loss": 0.125,
      "step": 145
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.4644466892353142e-05,
      "loss": 0.1047,
      "step": 146
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.4572423233046386e-05,
      "loss": 0.0993,
      "step": 147
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.4500078257975748e-05,
      "loss": 0.1071,
      "step": 148
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.4427436734565475e-05,
      "loss": 0.1047,
      "step": 149
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.4354503449781914e-05,
      "loss": 0.1008,
      "step": 150
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.428128320981804e-05,
      "loss": 0.1173,
      "step": 151
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4207780839776735e-05,
      "loss": 0.1268,
      "step": 152
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.4134001183352833e-05,
      "loss": 0.119,
      "step": 153
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.4059949102513914e-05,
      "loss": 0.1138,
      "step": 154
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.3985629477179915e-05,
      "loss": 0.125,
      "step": 155
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.391104720490156e-05,
      "loss": 0.1024,
      "step": 156
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.3836207200537597e-05,
      "loss": 0.0989,
      "step": 157
    },
    {
      "epoch": 2.77,
      "learning_rate": 1.376111439593093e-05,
      "loss": 0.1017,
      "step": 158
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.368577373958362e-05,
      "loss": 0.1113,
      "step": 159
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.3610190196330777e-05,
      "loss": 0.1196,
      "step": 160
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.3534368747013396e-05,
      "loss": 0.1314,
      "step": 161
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.3458314388150115e-05,
      "loss": 0.1279,
      "step": 162
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.3382032131607967e-05,
      "loss": 0.1216,
      "step": 163
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.3305527004272087e-05,
      "loss": 0.1212,
      "step": 164
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.3228804047714462e-05,
      "loss": 0.1388,
      "step": 165
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.31518683178617e-05,
      "loss": 0.1193,
      "step": 166
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.3074724884661833e-05,
      "loss": 0.113,
      "step": 167
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.2997378831750242e-05,
      "loss": 0.1115,
      "step": 168
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.2919835256114639e-05,
      "loss": 0.1232,
      "step": 169
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.2842099267759176e-05,
      "loss": 0.118,
      "step": 170
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.2764175989367717e-05,
      "loss": 0.1046,
      "step": 171
    },
    {
      "epoch": 3.02,
      "learning_rate": 1.2686070555966255e-05,
      "loss": 0.1047,
      "step": 172
    },
    {
      "epoch": 3.04,
      "learning_rate": 1.2607788114584522e-05,
      "loss": 0.1192,
      "step": 173
    },
    {
      "epoch": 3.05,
      "learning_rate": 1.2529333823916807e-05,
      "loss": 0.1008,
      "step": 174
    },
    {
      "epoch": 3.07,
      "learning_rate": 1.2450712853982014e-05,
      "loss": 0.0812,
      "step": 175
    },
    {
      "epoch": 3.09,
      "learning_rate": 1.2371930385782944e-05,
      "loss": 0.0961,
      "step": 176
    },
    {
      "epoch": 3.11,
      "learning_rate": 1.2292991610964902e-05,
      "loss": 0.0839,
      "step": 177
    },
    {
      "epoch": 3.12,
      "learning_rate": 1.2213901731473555e-05,
      "loss": 0.0851,
      "step": 178
    },
    {
      "epoch": 3.14,
      "learning_rate": 1.2134665959212138e-05,
      "loss": 0.0851,
      "step": 179
    },
    {
      "epoch": 3.16,
      "learning_rate": 1.2055289515698008e-05,
      "loss": 0.0709,
      "step": 180
    },
    {
      "epoch": 3.18,
      "learning_rate": 1.1975777631718533e-05,
      "loss": 0.0732,
      "step": 181
    },
    {
      "epoch": 3.19,
      "learning_rate": 1.189613554698641e-05,
      "loss": 0.1016,
      "step": 182
    },
    {
      "epoch": 3.21,
      "learning_rate": 1.1816368509794365e-05,
      "loss": 0.0832,
      "step": 183
    },
    {
      "epoch": 3.23,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 0.073,
      "step": 184
    },
    {
      "epoch": 3.25,
      "learning_rate": 1.1656480612025914e-05,
      "loss": 0.0744,
      "step": 185
    },
    {
      "epoch": 3.26,
      "learning_rate": 1.1576370287819737e-05,
      "loss": 0.0743,
      "step": 186
    },
    {
      "epoch": 3.28,
      "learning_rate": 1.149615608319978e-05,
      "loss": 0.0573,
      "step": 187
    },
    {
      "epoch": 3.3,
      "learning_rate": 1.14158432841606e-05,
      "loss": 0.0829,
      "step": 188
    },
    {
      "epoch": 3.32,
      "learning_rate": 1.133543718319398e-05,
      "loss": 0.0564,
      "step": 189
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.1254943078940161e-05,
      "loss": 0.1077,
      "step": 190
    },
    {
      "epoch": 3.35,
      "learning_rate": 1.1174366275838664e-05,
      "loss": 0.0831,
      "step": 191
    },
    {
      "epoch": 3.37,
      "learning_rate": 1.1093712083778748e-05,
      "loss": 0.079,
      "step": 192
    },
    {
      "epoch": 3.39,
      "learning_rate": 1.1012985817749465e-05,
      "loss": 0.0762,
      "step": 193
    },
    {
      "epoch": 3.4,
      "learning_rate": 1.093219279748946e-05,
      "loss": 0.0594,
      "step": 194
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.0851338347136358e-05,
      "loss": 0.0734,
      "step": 195
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.077042779487595e-05,
      "loss": 0.0868,
      "step": 196
    },
    {
      "epoch": 3.46,
      "learning_rate": 1.068946647259105e-05,
      "loss": 0.0958,
      "step": 197
    },
    {
      "epoch": 3.47,
      "learning_rate": 1.060845971551014e-05,
      "loss": 0.0836,
      "step": 198
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.052741286185579e-05,
      "loss": 0.1155,
      "step": 199
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.0446331252492864e-05,
      "loss": 0.0799,
      "step": 200
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.0365220230576592e-05,
      "loss": 0.0935,
      "step": 201
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.0284085141200424e-05,
      "loss": 0.0853,
      "step": 202
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.0202931331043839e-05,
      "loss": 0.0703,
      "step": 203
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.0121764148019977e-05,
      "loss": 0.0951,
      "step": 204
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.004058894092323e-05,
      "loss": 0.0804,
      "step": 205
    },
    {
      "epoch": 3.61,
      "learning_rate": 9.95941105907677e-06,
      "loss": 0.0963,
      "step": 206
    },
    {
      "epoch": 3.63,
      "learning_rate": 9.878235851980027e-06,
      "loss": 0.0955,
      "step": 207
    },
    {
      "epoch": 3.65,
      "learning_rate": 9.797068668956163e-06,
      "loss": 0.0862,
      "step": 208
    },
    {
      "epoch": 3.67,
      "learning_rate": 9.715914858799576e-06,
      "loss": 0.0737,
      "step": 209
    },
    {
      "epoch": 3.68,
      "learning_rate": 9.634779769423412e-06,
      "loss": 0.1326,
      "step": 210
    },
    {
      "epoch": 3.7,
      "learning_rate": 9.553668747507139e-06,
      "loss": 0.079,
      "step": 211
    },
    {
      "epoch": 3.72,
      "learning_rate": 9.472587138144215e-06,
      "loss": 0.0948,
      "step": 212
    },
    {
      "epoch": 3.74,
      "learning_rate": 9.391540284489862e-06,
      "loss": 0.0698,
      "step": 213
    },
    {
      "epoch": 3.75,
      "learning_rate": 9.310533527408953e-06,
      "loss": 0.0785,
      "step": 214
    },
    {
      "epoch": 3.77,
      "learning_rate": 9.229572205124052e-06,
      "loss": 0.078,
      "step": 215
    },
    {
      "epoch": 3.79,
      "learning_rate": 9.148661652863644e-06,
      "loss": 0.0861,
      "step": 216
    },
    {
      "epoch": 3.81,
      "learning_rate": 9.067807202510543e-06,
      "loss": 0.0847,
      "step": 217
    },
    {
      "epoch": 3.82,
      "learning_rate": 8.987014182250538e-06,
      "loss": 0.073,
      "step": 218
    },
    {
      "epoch": 3.84,
      "learning_rate": 8.906287916221259e-06,
      "loss": 0.0897,
      "step": 219
    },
    {
      "epoch": 3.86,
      "learning_rate": 8.825633724161336e-06,
      "loss": 0.0642,
      "step": 220
    },
    {
      "epoch": 3.88,
      "learning_rate": 8.74505692105984e-06,
      "loss": 0.0862,
      "step": 221
    },
    {
      "epoch": 3.89,
      "learning_rate": 8.664562816806022e-06,
      "loss": 0.0767,
      "step": 222
    },
    {
      "epoch": 3.91,
      "learning_rate": 8.584156715839402e-06,
      "loss": 0.0593,
      "step": 223
    },
    {
      "epoch": 3.93,
      "learning_rate": 8.503843916800221e-06,
      "loss": 0.0947,
      "step": 224
    },
    {
      "epoch": 3.95,
      "learning_rate": 8.423629712180265e-06,
      "loss": 0.0717,
      "step": 225
    },
    {
      "epoch": 3.96,
      "learning_rate": 8.343519387974091e-06,
      "loss": 0.0909,
      "step": 226
    },
    {
      "epoch": 3.98,
      "learning_rate": 8.263518223330698e-06,
      "loss": 0.0657,
      "step": 227
    },
    {
      "epoch": 4.0,
      "learning_rate": 8.183631490205636e-06,
      "loss": 0.0593,
      "step": 228
    },
    {
      "epoch": 4.02,
      "learning_rate": 8.103864453013593e-06,
      "loss": 0.0517,
      "step": 229
    },
    {
      "epoch": 4.04,
      "learning_rate": 8.024222368281469e-06,
      "loss": 0.0422,
      "step": 230
    },
    {
      "epoch": 4.05,
      "learning_rate": 7.944710484301995e-06,
      "loss": 0.0432,
      "step": 231
    },
    {
      "epoch": 4.07,
      "learning_rate": 7.865334040787866e-06,
      "loss": 0.0531,
      "step": 232
    },
    {
      "epoch": 4.09,
      "learning_rate": 7.786098268526447e-06,
      "loss": 0.043,
      "step": 233
    },
    {
      "epoch": 4.11,
      "learning_rate": 7.707008389035102e-06,
      "loss": 0.0303,
      "step": 234
    },
    {
      "epoch": 4.12,
      "learning_rate": 7.628069614217058e-06,
      "loss": 0.0582,
      "step": 235
    },
    {
      "epoch": 4.14,
      "learning_rate": 7.54928714601799e-06,
      "loss": 0.0372,
      "step": 236
    },
    {
      "epoch": 4.16,
      "learning_rate": 7.470666176083193e-06,
      "loss": 0.0364,
      "step": 237
    },
    {
      "epoch": 4.18,
      "learning_rate": 7.3922118854154815e-06,
      "loss": 0.0444,
      "step": 238
    },
    {
      "epoch": 4.19,
      "learning_rate": 7.31392944403375e-06,
      "loss": 0.0577,
      "step": 239
    },
    {
      "epoch": 4.21,
      "learning_rate": 7.235824010632284e-06,
      "loss": 0.0252,
      "step": 240
    },
    {
      "epoch": 4.23,
      "learning_rate": 7.157900732240828e-06,
      "loss": 0.0406,
      "step": 241
    },
    {
      "epoch": 4.25,
      "learning_rate": 7.080164743885364e-06,
      "loss": 0.0426,
      "step": 242
    },
    {
      "epoch": 4.26,
      "learning_rate": 7.002621168249759e-06,
      "loss": 0.0229,
      "step": 243
    },
    {
      "epoch": 4.28,
      "learning_rate": 6.925275115338168e-06,
      "loss": 0.0299,
      "step": 244
    },
    {
      "epoch": 4.3,
      "learning_rate": 6.848131682138304e-06,
      "loss": 0.0474,
      "step": 245
    },
    {
      "epoch": 4.32,
      "learning_rate": 6.771195952285541e-06,
      "loss": 0.0421,
      "step": 246
    },
    {
      "epoch": 4.33,
      "learning_rate": 6.694472995727914e-06,
      "loss": 0.031,
      "step": 247
    },
    {
      "epoch": 4.35,
      "learning_rate": 6.617967868392036e-06,
      "loss": 0.0392,
      "step": 248
    },
    {
      "epoch": 4.37,
      "learning_rate": 6.5416856118498874e-06,
      "loss": 0.0357,
      "step": 249
    },
    {
      "epoch": 4.39,
      "learning_rate": 6.4656312529866086e-06,
      "loss": 0.0436,
      "step": 250
    },
    {
      "epoch": 4.4,
      "learning_rate": 6.389809803669226e-06,
      "loss": 0.0355,
      "step": 251
    },
    {
      "epoch": 4.42,
      "learning_rate": 6.314226260416383e-06,
      "loss": 0.0492,
      "step": 252
    },
    {
      "epoch": 4.44,
      "learning_rate": 6.238885604069075e-06,
      "loss": 0.016,
      "step": 253
    },
    {
      "epoch": 4.46,
      "learning_rate": 6.163792799462404e-06,
      "loss": 0.0327,
      "step": 254
    },
    {
      "epoch": 4.47,
      "learning_rate": 6.088952795098442e-06,
      "loss": 0.0511,
      "step": 255
    },
    {
      "epoch": 4.49,
      "learning_rate": 6.014370522820085e-06,
      "loss": 0.0609,
      "step": 256
    },
    {
      "epoch": 4.51,
      "learning_rate": 5.9400508974860895e-06,
      "loss": 0.0286,
      "step": 257
    },
    {
      "epoch": 4.53,
      "learning_rate": 5.8659988166471715e-06,
      "loss": 0.0436,
      "step": 258
    },
    {
      "epoch": 4.54,
      "learning_rate": 5.792219160223268e-06,
      "loss": 0.0556,
      "step": 259
    },
    {
      "epoch": 4.56,
      "learning_rate": 5.7187167901819665e-06,
      "loss": 0.0761,
      "step": 260
    },
    {
      "epoch": 4.58,
      "learning_rate": 5.645496550218089e-06,
      "loss": 0.046,
      "step": 261
    },
    {
      "epoch": 4.6,
      "learning_rate": 5.572563265434528e-06,
      "loss": 0.0292,
      "step": 262
    },
    {
      "epoch": 4.61,
      "learning_rate": 5.499921742024258e-06,
      "loss": 0.0283,
      "step": 263
    },
    {
      "epoch": 4.63,
      "learning_rate": 5.427576766953615e-06,
      "loss": 0.0351,
      "step": 264
    },
    {
      "epoch": 4.65,
      "learning_rate": 5.3555331076468585e-06,
      "loss": 0.0391,
      "step": 265
    },
    {
      "epoch": 4.67,
      "learning_rate": 5.283795511671994e-06,
      "loss": 0.0524,
      "step": 266
    },
    {
      "epoch": 4.68,
      "learning_rate": 5.212368706427913e-06,
      "loss": 0.0389,
      "step": 267
    },
    {
      "epoch": 4.7,
      "learning_rate": 5.141257398832863e-06,
      "loss": 0.0379,
      "step": 268
    },
    {
      "epoch": 4.72,
      "learning_rate": 5.0704662750142875e-06,
      "loss": 0.045,
      "step": 269
    },
    {
      "epoch": 4.74,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.0348,
      "step": 270
    },
    {
      "epoch": 4.75,
      "learning_rate": 4.929863217410768e-06,
      "loss": 0.0548,
      "step": 271
    },
    {
      "epoch": 4.77,
      "learning_rate": 4.860060549154302e-06,
      "loss": 0.0467,
      "step": 272
    },
    {
      "epoch": 4.79,
      "learning_rate": 4.790596595120699e-06,
      "loss": 0.0423,
      "step": 273
    },
    {
      "epoch": 4.81,
      "learning_rate": 4.721475932879283e-06,
      "loss": 0.0434,
      "step": 274
    },
    {
      "epoch": 4.82,
      "learning_rate": 4.652703117376987e-06,
      "loss": 0.0383,
      "step": 275
    },
    {
      "epoch": 4.84,
      "learning_rate": 4.584282680638155e-06,
      "loss": 0.0431,
      "step": 276
    },
    {
      "epoch": 4.86,
      "learning_rate": 4.51621913146592e-06,
      "loss": 0.0346,
      "step": 277
    },
    {
      "epoch": 4.88,
      "learning_rate": 4.448516955145048e-06,
      "loss": 0.0323,
      "step": 278
    },
    {
      "epoch": 4.89,
      "learning_rate": 4.381180613146396e-06,
      "loss": 0.032,
      "step": 279
    },
    {
      "epoch": 4.91,
      "learning_rate": 4.314214542832889e-06,
      "loss": 0.051,
      "step": 280
    },
    {
      "epoch": 4.93,
      "learning_rate": 4.2476231571671025e-06,
      "loss": 0.05,
      "step": 281
    },
    {
      "epoch": 4.95,
      "learning_rate": 4.181410844420473e-06,
      "loss": 0.0302,
      "step": 282
    },
    {
      "epoch": 4.96,
      "learning_rate": 4.1155819678840935e-06,
      "loss": 0.0471,
      "step": 283
    },
    {
      "epoch": 4.98,
      "learning_rate": 4.050140865581205e-06,
      "loss": 0.0379,
      "step": 284
    },
    {
      "epoch": 5.0,
      "learning_rate": 3.9850918499812976e-06,
      "loss": 0.0414,
      "step": 285
    },
    {
      "epoch": 5.02,
      "learning_rate": 3.920439207715955e-06,
      "loss": 0.032,
      "step": 286
    },
    {
      "epoch": 5.04,
      "learning_rate": 3.8561871992963585e-06,
      "loss": 0.0191,
      "step": 287
    },
    {
      "epoch": 5.05,
      "learning_rate": 3.7923400588325156e-06,
      "loss": 0.0167,
      "step": 288
    },
    {
      "epoch": 5.07,
      "learning_rate": 3.7289019937542603e-06,
      "loss": 0.0157,
      "step": 289
    },
    {
      "epoch": 5.09,
      "learning_rate": 3.665877184533968e-06,
      "loss": 0.0247,
      "step": 290
    },
    {
      "epoch": 5.11,
      "learning_rate": 3.6032697844110896e-06,
      "loss": 0.0128,
      "step": 291
    },
    {
      "epoch": 5.12,
      "learning_rate": 3.5410839191184386e-06,
      "loss": 0.0147,
      "step": 292
    },
    {
      "epoch": 5.14,
      "learning_rate": 3.4793236866103296e-06,
      "loss": 0.019,
      "step": 293
    },
    {
      "epoch": 5.16,
      "learning_rate": 3.4179931567925216e-06,
      "loss": 0.0067,
      "step": 294
    },
    {
      "epoch": 5.18,
      "learning_rate": 3.3570963712540083e-06,
      "loss": 0.0141,
      "step": 295
    },
    {
      "epoch": 5.19,
      "learning_rate": 3.2966373430007047e-06,
      "loss": 0.0192,
      "step": 296
    },
    {
      "epoch": 5.21,
      "learning_rate": 3.236620056190972e-06,
      "loss": 0.0114,
      "step": 297
    },
    {
      "epoch": 5.23,
      "learning_rate": 3.1770484658730896e-06,
      "loss": 0.0092,
      "step": 298
    },
    {
      "epoch": 5.25,
      "learning_rate": 3.1179264977246106e-06,
      "loss": 0.0263,
      "step": 299
    },
    {
      "epoch": 5.26,
      "learning_rate": 3.0592580477936606e-06,
      "loss": 0.0119,
      "step": 300
    },
    {
      "epoch": 5.28,
      "learning_rate": 3.001046982242216e-06,
      "loss": 0.0164,
      "step": 301
    },
    {
      "epoch": 5.3,
      "learning_rate": 2.9432971370913e-06,
      "loss": 0.0086,
      "step": 302
    },
    {
      "epoch": 5.32,
      "learning_rate": 2.8860123179682244e-06,
      "loss": 0.0173,
      "step": 303
    },
    {
      "epoch": 5.33,
      "learning_rate": 2.8291962998557754e-06,
      "loss": 0.0255,
      "step": 304
    },
    {
      "epoch": 5.35,
      "learning_rate": 2.772852826843476e-06,
      "loss": 0.0195,
      "step": 305
    },
    {
      "epoch": 5.37,
      "learning_rate": 2.7169856118808414e-06,
      "loss": 0.0115,
      "step": 306
    },
    {
      "epoch": 5.39,
      "learning_rate": 2.6615983365326926e-06,
      "loss": 0.0176,
      "step": 307
    },
    {
      "epoch": 5.4,
      "learning_rate": 2.60669465073657e-06,
      "loss": 0.0088,
      "step": 308
    },
    {
      "epoch": 5.42,
      "learning_rate": 2.5522781725621814e-06,
      "loss": 0.0313,
      "step": 309
    },
    {
      "epoch": 5.44,
      "learning_rate": 2.498352487973007e-06,
      "loss": 0.0221,
      "step": 310
    },
    {
      "epoch": 5.46,
      "learning_rate": 2.444921150589957e-06,
      "loss": 0.0201,
      "step": 311
    },
    {
      "epoch": 5.47,
      "learning_rate": 2.3919876814572197e-06,
      "loss": 0.018,
      "step": 312
    },
    {
      "epoch": 5.49,
      "learning_rate": 2.339555568810221e-06,
      "loss": 0.0125,
      "step": 313
    },
    {
      "epoch": 5.51,
      "learning_rate": 2.287628267845744e-06,
      "loss": 0.0087,
      "step": 314
    },
    {
      "epoch": 5.53,
      "learning_rate": 2.2362092004942583e-06,
      "loss": 0.0098,
      "step": 315
    },
    {
      "epoch": 5.54,
      "learning_rate": 2.185301755194399e-06,
      "loss": 0.0135,
      "step": 316
    },
    {
      "epoch": 5.56,
      "learning_rate": 2.1349092866696906e-06,
      "loss": 0.005,
      "step": 317
    },
    {
      "epoch": 5.58,
      "learning_rate": 2.08503511570746e-06,
      "loss": 0.0176,
      "step": 318
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.0356825289400185e-06,
      "loss": 0.0114,
      "step": 319
    },
    {
      "epoch": 5.61,
      "learning_rate": 1.986854778628067e-06,
      "loss": 0.0074,
      "step": 320
    },
    {
      "epoch": 5.63,
      "learning_rate": 1.9385550824463727e-06,
      "loss": 0.0171,
      "step": 321
    },
    {
      "epoch": 5.65,
      "learning_rate": 1.890786623271743e-06,
      "loss": 0.0166,
      "step": 322
    },
    {
      "epoch": 5.67,
      "learning_rate": 1.843552548973272e-06,
      "loss": 0.0227,
      "step": 323
    },
    {
      "epoch": 5.68,
      "learning_rate": 1.7968559722048906e-06,
      "loss": 0.0134,
      "step": 324
    },
    {
      "epoch": 5.7,
      "learning_rate": 1.7506999702002682e-06,
      "loss": 0.0135,
      "step": 325
    },
    {
      "epoch": 5.72,
      "learning_rate": 1.7050875845700066e-06,
      "loss": 0.0126,
      "step": 326
    },
    {
      "epoch": 5.74,
      "learning_rate": 1.660021821101222e-06,
      "loss": 0.0065,
      "step": 327
    },
    {
      "epoch": 5.75,
      "learning_rate": 1.6155056495594467e-06,
      "loss": 0.0061,
      "step": 328
    },
    {
      "epoch": 5.77,
      "learning_rate": 1.5715420034929475e-06,
      "loss": 0.0069,
      "step": 329
    },
    {
      "epoch": 5.79,
      "learning_rate": 1.528133780039397e-06,
      "loss": 0.0384,
      "step": 330
    },
    {
      "epoch": 5.81,
      "learning_rate": 1.4852838397349545e-06,
      "loss": 0.0142,
      "step": 331
    },
    {
      "epoch": 5.82,
      "learning_rate": 1.4429950063257713e-06,
      "loss": 0.0043,
      "step": 332
    },
    {
      "epoch": 5.84,
      "learning_rate": 1.401270066581899e-06,
      "loss": 0.0065,
      "step": 333
    },
    {
      "epoch": 5.86,
      "learning_rate": 1.360111770113659e-06,
      "loss": 0.0038,
      "step": 334
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.3195228291904271e-06,
      "loss": 0.0084,
      "step": 335
    },
    {
      "epoch": 5.89,
      "learning_rate": 1.279505918561923e-06,
      "loss": 0.0068,
      "step": 336
    },
    {
      "epoch": 5.91,
      "learning_rate": 1.240063675281934e-06,
      "loss": 0.0193,
      "step": 337
    },
    {
      "epoch": 5.93,
      "learning_rate": 1.2011986985345313e-06,
      "loss": 0.0093,
      "step": 338
    },
    {
      "epoch": 5.95,
      "learning_rate": 1.1629135494628097e-06,
      "loss": 0.0127,
      "step": 339
    },
    {
      "epoch": 5.96,
      "learning_rate": 1.1252107510000843e-06,
      "loss": 0.0112,
      "step": 340
    },
    {
      "epoch": 5.98,
      "learning_rate": 1.0880927877036608e-06,
      "loss": 0.0128,
      "step": 341
    },
    {
      "epoch": 6.0,
      "learning_rate": 1.051562105591082e-06,
      "loss": 0.0144,
      "step": 342
    },
    {
      "epoch": 6.02,
      "learning_rate": 1.0156211119789582e-06,
      "loss": 0.0027,
      "step": 343
    },
    {
      "epoch": 6.04,
      "learning_rate": 9.802721753243193e-07,
      "loss": 0.003,
      "step": 344
    },
    {
      "epoch": 6.05,
      "learning_rate": 9.455176250685338e-07,
      "loss": 0.0044,
      "step": 345
    },
    {
      "epoch": 6.07,
      "learning_rate": 9.113597514838135e-07,
      "loss": 0.0108,
      "step": 346
    },
    {
      "epoch": 6.09,
      "learning_rate": 8.778008055222809e-07,
      "loss": 0.0014,
      "step": 347
    },
    {
      "epoch": 6.11,
      "learning_rate": 8.448429986676298e-07,
      "loss": 0.0029,
      "step": 348
    },
    {
      "epoch": 6.12,
      "learning_rate": 8.124885027894014e-07,
      "loss": 0.0044,
      "step": 349
    },
    {
      "epoch": 6.14,
      "learning_rate": 7.807394499998578e-07,
      "loss": 0.0037,
      "step": 350
    },
    {
      "epoch": 6.16,
      "learning_rate": 7.495979325134806e-07,
      "loss": 0.004,
      "step": 351
    },
    {
      "epoch": 6.18,
      "learning_rate": 7.19066002509089e-07,
      "loss": 0.0028,
      "step": 352
    },
    {
      "epoch": 6.19,
      "learning_rate": 6.891456719946188e-07,
      "loss": 0.004,
      "step": 353
    },
    {
      "epoch": 6.21,
      "learning_rate": 6.598389126745209e-07,
      "loss": 0.0025,
      "step": 354
    },
    {
      "epoch": 6.23,
      "learning_rate": 6.311476558198337e-07,
      "loss": 0.0106,
      "step": 355
    },
    {
      "epoch": 6.25,
      "learning_rate": 6.030737921409169e-07,
      "loss": 0.0023,
      "step": 356
    },
    {
      "epoch": 6.26,
      "learning_rate": 5.756191716628556e-07,
      "loss": 0.0025,
      "step": 357
    },
    {
      "epoch": 6.28,
      "learning_rate": 5.487856036035488e-07,
      "loss": 0.0025,
      "step": 358
    },
    {
      "epoch": 6.3,
      "learning_rate": 5.225748562544741e-07,
      "loss": 0.0021,
      "step": 359
    },
    {
      "epoch": 6.32,
      "learning_rate": 4.969886568641757e-07,
      "loss": 0.0016,
      "step": 360
    },
    {
      "epoch": 6.33,
      "learning_rate": 4.7202869152443096e-07,
      "loss": 0.0103,
      "step": 361
    },
    {
      "epoch": 6.35,
      "learning_rate": 4.4769660505914136e-07,
      "loss": 0.0015,
      "step": 362
    },
    {
      "epoch": 6.37,
      "learning_rate": 4.2399400091594154e-07,
      "loss": 0.0015,
      "step": 363
    },
    {
      "epoch": 6.39,
      "learning_rate": 4.0092244106053393e-07,
      "loss": 0.0049,
      "step": 364
    },
    {
      "epoch": 6.4,
      "learning_rate": 3.7848344587376297e-07,
      "loss": 0.0031,
      "step": 365
    },
    {
      "epoch": 6.42,
      "learning_rate": 3.566784940514145e-07,
      "loss": 0.0028,
      "step": 366
    },
    {
      "epoch": 6.44,
      "learning_rate": 3.355090225067792e-07,
      "loss": 0.0023,
      "step": 367
    },
    {
      "epoch": 6.46,
      "learning_rate": 3.1497642627596247e-07,
      "loss": 0.0022,
      "step": 368
    },
    {
      "epoch": 6.47,
      "learning_rate": 2.9508205842594727e-07,
      "loss": 0.0014,
      "step": 369
    },
    {
      "epoch": 6.49,
      "learning_rate": 2.758272299654341e-07,
      "loss": 0.014,
      "step": 370
    },
    {
      "epoch": 6.51,
      "learning_rate": 2.572132097584523e-07,
      "loss": 0.0018,
      "step": 371
    },
    {
      "epoch": 6.53,
      "learning_rate": 2.392412244407294e-07,
      "loss": 0.0013,
      "step": 372
    },
    {
      "epoch": 6.54,
      "learning_rate": 2.2191245833886988e-07,
      "loss": 0.0055,
      "step": 373
    },
    {
      "epoch": 6.56,
      "learning_rate": 2.052280533923101e-07,
      "loss": 0.0022,
      "step": 374
    },
    {
      "epoch": 6.58,
      "learning_rate": 1.8918910907805733e-07,
      "loss": 0.0039,
      "step": 375
    },
    {
      "epoch": 6.6,
      "learning_rate": 1.737966823382442e-07,
      "loss": 0.0012,
      "step": 376
    },
    {
      "epoch": 6.61,
      "learning_rate": 1.5905178751047135e-07,
      "loss": 0.0032,
      "step": 377
    },
    {
      "epoch": 6.63,
      "learning_rate": 1.4495539626097289e-07,
      "loss": 0.0026,
      "step": 378
    },
    {
      "epoch": 6.65,
      "learning_rate": 1.3150843752057442e-07,
      "loss": 0.0017,
      "step": 379
    },
    {
      "epoch": 6.67,
      "learning_rate": 1.1871179742348416e-07,
      "loss": 0.0059,
      "step": 380
    },
    {
      "epoch": 6.68,
      "learning_rate": 1.0656631924889749e-07,
      "loss": 0.0023,
      "step": 381
    },
    {
      "epoch": 6.7,
      "learning_rate": 9.507280336542912e-08,
      "loss": 0.0049,
      "step": 382
    },
    {
      "epoch": 6.72,
      "learning_rate": 8.423200717835978e-08,
      "loss": 0.0133,
      "step": 383
    },
    {
      "epoch": 6.74,
      "learning_rate": 7.404464507973608e-08,
      "loss": 0.0018,
      "step": 384
    },
    {
      "epoch": 6.75,
      "learning_rate": 6.451138840128601e-08,
      "loss": 0.0014,
      "step": 385
    },
    {
      "epoch": 6.77,
      "learning_rate": 5.5632865370183196e-08,
      "loss": 0.0033,
      "step": 386
    },
    {
      "epoch": 6.79,
      "learning_rate": 4.740966106764222e-08,
      "loss": 0.0023,
      "step": 387
    },
    {
      "epoch": 6.81,
      "learning_rate": 3.984231739036948e-08,
      "loss": 0.0029,
      "step": 388
    },
    {
      "epoch": 6.82,
      "learning_rate": 3.293133301485063e-08,
      "loss": 0.0071,
      "step": 389
    },
    {
      "epoch": 6.84,
      "learning_rate": 2.667716336448356e-08,
      "loss": 0.0015,
      "step": 390
    },
    {
      "epoch": 6.86,
      "learning_rate": 2.1080220579573485e-08,
      "loss": 0.0034,
      "step": 391
    },
    {
      "epoch": 6.88,
      "learning_rate": 1.614087349017246e-08,
      "loss": 0.0016,
      "step": 392
    },
    {
      "epoch": 6.89,
      "learning_rate": 1.1859447591769934e-08,
      "loss": 0.0037,
      "step": 393
    },
    {
      "epoch": 6.91,
      "learning_rate": 8.236225023844357e-09,
      "loss": 0.0051,
      "step": 394
    },
    {
      "epoch": 6.93,
      "learning_rate": 5.271444551276928e-09,
      "loss": 0.0018,
      "step": 395
    },
    {
      "epoch": 6.95,
      "learning_rate": 2.9653015486064143e-09,
      "loss": 0.0055,
      "step": 396
    },
    {
      "epoch": 6.96,
      "learning_rate": 1.3179479871638923e-09,
      "loss": 0.002,
      "step": 397
    },
    {
      "epoch": 6.98,
      "learning_rate": 3.2949242505408987e-10,
      "loss": 0.003,
      "step": 398
    },
    {
      "epoch": 7.0,
      "learning_rate": 0.0,
      "loss": 0.002,
      "step": 399
    },
    {
      "epoch": 7.0,
      "step": 399,
      "total_flos": 9464738217984.0,
      "train_loss": 0.08329384368762635,
      "train_runtime": 745.7277,
      "train_samples_per_second": 34.074,
      "train_steps_per_second": 0.535
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 399,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 7,
  "save_steps": 800,
  "total_flos": 9464738217984.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}