{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.8959,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.8883,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.8056,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.5905,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.7238,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.8706,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.7291,
      "step": 7
    },
    {
      "epoch": 0.02,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.8379,
      "step": 8
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.7678,
      "step": 9
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.5016,
      "step": 10
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.8e-06,
      "loss": 0.5994,
      "step": 11
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.6028,
      "step": 12
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.04e-05,
      "loss": 0.6321,
      "step": 13
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.3967,
      "step": 14
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.2e-05,
      "loss": 0.4705,
      "step": 15
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 0.5506,
      "step": 16
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 0.3876,
      "step": 17
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 0.4482,
      "step": 18
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 0.5656,
      "step": 19
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.7643,
      "step": 20
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.3924,
      "step": 21
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.76e-05,
      "loss": 0.4228,
      "step": 22
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 0.3103,
      "step": 23
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.5515,
      "step": 24
    },
    {
      "epoch": 0.05,
      "learning_rate": 2e-05,
      "loss": 0.374,
      "step": 25
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.9999781283802247e-05,
      "loss": 0.174,
      "step": 26
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.999912514477634e-05,
      "loss": 0.2516,
      "step": 27
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.999803161162393e-05,
      "loss": 0.4822,
      "step": 28
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9996500732179695e-05,
      "loss": 0.2868,
      "step": 29
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.999453257340926e-05,
      "loss": 0.3294,
      "step": 30
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9992127221406276e-05,
      "loss": 0.4017,
      "step": 31
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9989284781388617e-05,
      "loss": 0.3511,
      "step": 32
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9986005377693826e-05,
      "loss": 0.3472,
      "step": 33
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9982289153773648e-05,
      "loss": 0.145,
      "step": 34
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9978136272187745e-05,
      "loss": 0.2914,
      "step": 35
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9973546914596622e-05,
      "loss": 0.403,
      "step": 36
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9968521281753642e-05,
      "loss": 0.2962,
      "step": 37
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.996305959349627e-05,
      "loss": 0.3813,
      "step": 38
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.995716208873644e-05,
      "loss": 0.2893,
      "step": 39
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9950829025450116e-05,
      "loss": 0.317,
      "step": 40
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9944060680666e-05,
      "loss": 0.2615,
      "step": 41
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.993685735045343e-05,
      "loss": 0.3313,
      "step": 42
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9929219349909393e-05,
      "loss": 0.258,
      "step": 43
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9921147013144782e-05,
      "loss": 0.213,
      "step": 44
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9912640693269754e-05,
      "loss": 0.3592,
      "step": 45
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9903700762378303e-05,
      "loss": 0.2213,
      "step": 46
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.989432761153196e-05,
      "loss": 0.3759,
      "step": 47
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9884521650742718e-05,
      "loss": 0.1658,
      "step": 48
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9874283308955058e-05,
      "loss": 0.1732,
      "step": 49
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9863613034027224e-05,
      "loss": 0.2279,
      "step": 50
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.985251129271161e-05,
      "loss": 0.1859,
      "step": 51
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9840978570634338e-05,
      "loss": 0.5561,
      "step": 52
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.982901537227404e-05,
      "loss": 0.1094,
      "step": 53
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9816622220939762e-05,
      "loss": 0.16,
      "step": 54
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9803799658748096e-05,
      "loss": 0.2063,
      "step": 55
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9790548246599447e-05,
      "loss": 0.2317,
      "step": 56
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9776868564153517e-05,
      "loss": 0.2567,
      "step": 57
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.976276120980393e-05,
      "loss": 0.1225,
      "step": 58
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9748226800652062e-05,
      "loss": 0.1664,
      "step": 59
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.973326597248006e-05,
      "loss": 0.2462,
      "step": 60
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9717879379723012e-05,
      "loss": 0.1357,
      "step": 61
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9702067695440333e-05,
      "loss": 0.2895,
      "step": 62
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9685831611286312e-05,
      "loss": 0.1432,
      "step": 63
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.966917183747987e-05,
      "loss": 0.9037,
      "step": 64
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9652089102773487e-05,
      "loss": 0.1122,
      "step": 65
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9634584154421316e-05,
      "loss": 0.3454,
      "step": 66
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9616657758146503e-05,
      "loss": 0.2993,
      "step": 67
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.95983106981077e-05,
      "loss": 0.294,
      "step": 68
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.957954377686475e-05,
      "loss": 0.2309,
      "step": 69
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9560357815343577e-05,
      "loss": 0.4046,
      "step": 70
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.95407536528003e-05,
      "loss": 0.5102,
      "step": 71
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.95407536528003e-05,
      "loss": 0.302,
      "step": 72
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9520732146784493e-05,
      "loss": 0.1975,
      "step": 73
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9500294173101687e-05,
      "loss": 0.5097,
      "step": 74
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.947944062577507e-05,
      "loss": 0.1714,
      "step": 75
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9458172417006347e-05,
      "loss": 0.3065,
      "step": 76
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9436490477135877e-05,
      "loss": 0.3434,
      "step": 77
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.941439575460195e-05,
      "loss": 0.2067,
      "step": 78
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.93918892158993e-05,
      "loss": 0.1366,
      "step": 79
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9368971845536844e-05,
      "loss": 0.2233,
      "step": 80
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.934564464599461e-05,
      "loss": 0.1644,
      "step": 81
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9321908637679868e-05,
      "loss": 0.3666,
      "step": 82
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9297764858882516e-05,
      "loss": 0.184,
      "step": 83
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9273214365729655e-05,
      "loss": 0.6584,
      "step": 84
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.924825823213939e-05,
      "loss": 0.2562,
      "step": 85
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.922289754977385e-05,
      "loss": 0.0571,
      "step": 86
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9197133427991437e-05,
      "loss": 0.1578,
      "step": 87
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9170966993798302e-05,
      "loss": 0.1214,
      "step": 88
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9144399391799043e-05,
      "loss": 0.1772,
      "step": 89
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.911743178414665e-05,
      "loss": 0.249,
      "step": 90
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.909006535049163e-05,
      "loss": 0.1092,
      "step": 91
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9062301287930448e-05,
      "loss": 0.1524,
      "step": 92
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.903414081095315e-05,
      "loss": 0.1331,
      "step": 93
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9005585151390224e-05,
      "loss": 0.4028,
      "step": 94
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.897663555835872e-05,
      "loss": 0.1218,
      "step": 95
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8947293298207637e-05,
      "loss": 0.1068,
      "step": 96
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.8917559654462474e-05,
      "loss": 0.1773,
      "step": 97
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8887435927769137e-05,
      "loss": 0.1191,
      "step": 98
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8856923435837024e-05,
      "loss": 0.5611,
      "step": 99
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8826023513381372e-05,
      "loss": 0.1207,
      "step": 100
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.879473751206489e-05,
      "loss": 0.1215,
      "step": 101
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.8763066800438638e-05,
      "loss": 0.2272,
      "step": 102
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8731012763882132e-05,
      "loss": 0.2555,
      "step": 103
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8698576804542775e-05,
      "loss": 0.0687,
      "step": 104
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.8665760341274505e-05,
      "loss": 0.2113,
      "step": 105
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.863256480957574e-05,
      "loss": 0.2414,
      "step": 106
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.859899166152657e-05,
      "loss": 0.2,
      "step": 107
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.856504236572526e-05,
      "loss": 0.2686,
      "step": 108
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8530718407223976e-05,
      "loss": 0.1561,
      "step": 109
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.849602128746387e-05,
      "loss": 0.1295,
      "step": 110
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8460952524209355e-05,
      "loss": 0.1129,
      "step": 111
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8425513651481748e-05,
      "loss": 0.1517,
      "step": 112
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8389706219492147e-05,
      "loss": 0.2548,
      "step": 113
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8353531794573623e-05,
      "loss": 0.1433,
      "step": 114
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.831699195911272e-05,
      "loss": 0.3476,
      "step": 115
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8280088311480203e-05,
      "loss": 0.3045,
      "step": 116
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8242822465961177e-05,
      "loss": 0.3731,
      "step": 117
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.8205196052684445e-05,
      "loss": 0.0941,
      "step": 118
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.8167210717551224e-05,
      "loss": 0.2446,
      "step": 119
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.8128868122163125e-05,
      "loss": 0.4302,
      "step": 120
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 0.1842,
      "step": 121
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.8051117875093974e-05,
      "loss": 0.2067,
      "step": 122
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.8011713624460608e-05,
      "loss": 0.3245,
      "step": 123
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.797195891551896e-05,
      "loss": 0.215,
      "step": 124
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.793185548726878e-05,
      "loss": 0.1254,
      "step": 125
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.789140509396394e-05,
      "loss": 0.2598,
      "step": 126
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.785060950503568e-05,
      "loss": 0.1237,
      "step": 127
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.780947050501522e-05,
      "loss": 0.1599,
      "step": 128
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7767989893455696e-05,
      "loss": 0.3742,
      "step": 129
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7726169484853438e-05,
      "loss": 0.2924,
      "step": 130
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7684011108568593e-05,
      "loss": 0.1479,
      "step": 131
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.7641516608745114e-05,
      "loss": 0.1032,
      "step": 132
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.759868784423009e-05,
      "loss": 0.3311,
      "step": 133
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7555526688492418e-05,
      "loss": 0.1297,
      "step": 134
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7512035029540887e-05,
      "loss": 0.2778,
      "step": 135
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.7468214769841542e-05,
      "loss": 0.2048,
      "step": 136
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.74240678262345e-05,
      "loss": 0.2699,
      "step": 137
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.7379596129850098e-05,
      "loss": 0.3385,
      "step": 138
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.73348016260244e-05,
      "loss": 0.281,
      "step": 139
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.7289686274214116e-05,
      "loss": 0.1776,
      "step": 140
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.7244252047910893e-05,
      "loss": 0.1584,
      "step": 141
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.7198500934554966e-05,
      "loss": 0.1673,
      "step": 142
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.7152434935448257e-05,
      "loss": 0.1588,
      "step": 143
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.7106056065666793e-05,
      "loss": 0.3462,
      "step": 144
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.705936635397259e-05,
      "loss": 0.2498,
      "step": 145
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.7012367842724887e-05,
      "loss": 0.3301,
      "step": 146
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.6965062587790823e-05,
      "loss": 0.2263,
      "step": 147
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6917452658455496e-05,
      "loss": 0.1131,
      "step": 148
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6869540137331445e-05,
      "loss": 0.2109,
      "step": 149
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6821327120267567e-05,
      "loss": 0.1655,
      "step": 150
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6772815716257414e-05,
      "loss": 0.3201,
      "step": 151
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.6724008047346946e-05,
      "loss": 0.1607,
      "step": 152
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.667490624854173e-05,
      "loss": 0.2095,
      "step": 153
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.66255124677135e-05,
      "loss": 0.5483,
      "step": 154
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6575828865506246e-05,
      "loss": 0.18,
      "step": 155
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.6525857615241686e-05,
      "loss": 0.0545,
      "step": 156
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.647560090282419e-05,
      "loss": 0.3574,
      "step": 157
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.6425060926645168e-05,
      "loss": 0.3601,
      "step": 158
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.63742398974869e-05,
      "loss": 0.278,
      "step": 159
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.6323140038425842e-05,
      "loss": 0.3027,
      "step": 160
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.6271763584735373e-05,
      "loss": 0.4648,
      "step": 161
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.622011278378801e-05,
      "loss": 0.0562,
      "step": 162
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.616818989495711e-05,
      "loss": 0.1884,
      "step": 163
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.6115997189518043e-05,
      "loss": 0.1693,
      "step": 164
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.6063536950548825e-05,
      "loss": 0.1806,
      "step": 165
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.6010811472830253e-05,
      "loss": 0.4201,
      "step": 166
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.595782306274553e-05,
      "loss": 0.2366,
      "step": 167
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5904574038179372e-05,
      "loss": 0.0898,
      "step": 168
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5851066728416617e-05,
      "loss": 0.0674,
      "step": 169
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5797303474040332e-05,
      "loss": 0.1518,
      "step": 170
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5743286626829437e-05,
      "loss": 0.6562,
      "step": 171
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.5689018549655815e-05,
      "loss": 0.1482,
      "step": 172
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5634501616380967e-05,
      "loss": 0.396,
      "step": 173
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5579738211752165e-05,
      "loss": 0.0992,
      "step": 174
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5524730731298136e-05,
      "loss": 0.1566,
      "step": 175
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.5469481581224274e-05,
      "loss": 0.186,
      "step": 176
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.541399317830738e-05,
      "loss": 0.372,
      "step": 177
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.5358267949789968e-05,
      "loss": 0.0843,
      "step": 178
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.530230833327405e-05,
      "loss": 0.2175,
      "step": 179
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.524611677661454e-05,
      "loss": 0.4303,
      "step": 180
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.5189695737812153e-05,
      "loss": 0.3197,
      "step": 181
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.5133047684905916e-05,
      "loss": 0.0897,
      "step": 182
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.5076175095865171e-05,
      "loss": 0.2646,
      "step": 183
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.5019080458481203e-05,
      "loss": 0.1229,
      "step": 184
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4961766270258422e-05,
      "loss": 0.1209,
      "step": 185
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4904235038305084e-05,
      "loss": 0.1069,
      "step": 186
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.4846489279223653e-05,
      "loss": 0.3356,
      "step": 187
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4788531519000696e-05,
      "loss": 0.1611,
      "step": 188
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.473036429289641e-05,
      "loss": 0.3134,
      "step": 189
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4671990145333697e-05,
      "loss": 0.4353,
      "step": 190
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.461341162978688e-05,
      "loss": 0.2424,
      "step": 191
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.4554631308669993e-05,
      "loss": 0.1279,
      "step": 192
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4495651753224706e-05,
      "loss": 0.1231,
      "step": 193
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4436475543407843e-05,
      "loss": 0.1808,
      "step": 194
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.437710526777852e-05,
      "loss": 0.1154,
      "step": 195
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4317543523384928e-05,
      "loss": 0.2613,
      "step": 196
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.4257792915650728e-05,
      "loss": 0.1583,
      "step": 197
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.4197856058261059e-05,
      "loss": 0.2394,
      "step": 198
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.4137735573048232e-05,
      "loss": 0.0783,
      "step": 199
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.4077434089877038e-05,
      "loss": 0.1302,
      "step": 200
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.4016954246529697e-05,
      "loss": 0.1186,
      "step": 201
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.3956298688590484e-05,
      "loss": 0.0853,
      "step": 202
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3895470069330003e-05,
      "loss": 0.2058,
      "step": 203
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3834471049589117e-05,
      "loss": 0.2419,
      "step": 204
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.377330429766256e-05,
      "loss": 0.2448,
      "step": 205
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3711972489182208e-05,
      "loss": 0.0812,
      "step": 206
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.3650478307000059e-05,
      "loss": 0.0932,
      "step": 207
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3588824441070852e-05,
      "loss": 0.2578,
      "step": 208
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3527013588334415e-05,
      "loss": 0.1267,
      "step": 209
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3465048452597682e-05,
      "loss": 0.1509,
      "step": 210
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3402931744416432e-05,
      "loss": 0.2086,
      "step": 211
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.3340666180976713e-05,
      "loss": 0.1317,
      "step": 212
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.3340666180976713e-05,
      "loss": 0.096,
      "step": 213
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.3278254485975977e-05,
      "loss": 0.3035,
      "step": 214
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.3215699389503956e-05,
      "loss": 0.3848,
      "step": 215
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.3153003627923217e-05,
      "loss": 0.2722,
      "step": 216
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 0.2982,
      "step": 217
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.3027201085531633e-05,
      "loss": 0.0712,
      "step": 218
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2964099807731539e-05,
      "loss": 0.0559,
      "step": 219
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2900868870603502e-05,
      "loss": 0.1888,
      "step": 220
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.283751104007355e-05,
      "loss": 0.1306,
      "step": 221
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.2774029087618448e-05,
      "loss": 0.1362,
      "step": 222
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2710425790144445e-05,
      "loss": 0.0681,
      "step": 223
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2646703929865817e-05,
      "loss": 0.1985,
      "step": 224
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2582866294183167e-05,
      "loss": 0.3362,
      "step": 225
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2518915675561482e-05,
      "loss": 0.266,
      "step": 226
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.2454854871407993e-05,
      "loss": 0.2089,
      "step": 227
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.2390686683949799e-05,
      "loss": 0.2582,
      "step": 228
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.2326413920111304e-05,
      "loss": 0.2865,
      "step": 229
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.2262039391391405e-05,
      "loss": 0.1277,
      "step": 230
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.2197565913740531e-05,
      "loss": 0.2532,
      "step": 231
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.213299630743747e-05,
      "loss": 0.1533,
      "step": 232
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.2068333396965968e-05,
      "loss": 0.0608,
      "step": 233
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.2003580010891214e-05,
      "loss": 0.2867,
      "step": 234
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1938738981736084e-05,
      "loss": 0.2761,
      "step": 235
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.187381314585725e-05,
      "loss": 0.2299,
      "step": 236
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.1808805343321102e-05,
      "loss": 0.4514,
      "step": 237
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1743718417779518e-05,
      "loss": 0.2327,
      "step": 238
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1678555216345478e-05,
      "loss": 0.5068,
      "step": 239
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1613318589468512e-05,
      "loss": 0.183,
      "step": 240
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.1548011390810016e-05,
      "loss": 0.4153,
      "step": 241
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.148263647711842e-05,
      "loss": 0.1517,
      "step": 242
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.1417196708104244e-05,
      "loss": 0.2708,
      "step": 243
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.135169494631497e-05,
      "loss": 0.0495,
      "step": 244
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.1286134057009862e-05,
      "loss": 0.0739,
      "step": 245
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.1220516908034602e-05,
      "loss": 0.3282,
      "step": 246
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.1154846369695864e-05,
      "loss": 0.0478,
      "step": 247
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.1089125314635727e-05,
      "loss": 0.2256,
      "step": 248
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.1023356617706051e-05,
      "loss": 0.107,
      "step": 249
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0957543155842703e-05,
      "loss": 0.0797,
      "step": 250
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0891687807939707e-05,
      "loss": 0.07,
      "step": 251
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.0825793454723325e-05,
      "loss": 0.3117,
      "step": 252
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0759862978626032e-05,
      "loss": 0.2769,
      "step": 253
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0693899263660442e-05,
      "loss": 0.1839,
      "step": 254
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0627905195293135e-05,
      "loss": 0.0428,
      "step": 255
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0561883660318456e-05,
      "loss": 0.0903,
      "step": 256
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.0495837546732224e-05,
      "loss": 0.2216,
      "step": 257
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.0429769743605406e-05,
      "loss": 0.0941,
      "step": 258
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.0363683140957745e-05,
      "loss": 0.5135,
      "step": 259
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.0297580629631324e-05,
      "loss": 0.1504,
      "step": 260
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.023146510116414e-05,
      "loss": 0.1991,
      "step": 261
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.0165339447663586e-05,
      "loss": 0.0628,
      "step": 262
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.0099206561679964e-05,
      "loss": 0.298,
      "step": 263
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.0033069336079952e-05,
      "loss": 0.5587,
      "step": 264
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.96693066392005e-06,
      "loss": 0.0739,
      "step": 265
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.900793438320037e-06,
      "loss": 0.3523,
      "step": 266
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.834660552336415e-06,
      "loss": 0.1706,
      "step": 267
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.768534898835864e-06,
      "loss": 0.2783,
      "step": 268
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.702419370368677e-06,
      "loss": 0.2093,
      "step": 269
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.636316859042258e-06,
      "loss": 0.2205,
      "step": 270
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.570230256394595e-06,
      "loss": 0.1987,
      "step": 271
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.504162453267776e-06,
      "loss": 0.1426,
      "step": 272
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.438116339681546e-06,
      "loss": 0.0697,
      "step": 273
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.372094804706867e-06,
      "loss": 0.1873,
      "step": 274
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.30610073633956e-06,
      "loss": 0.3201,
      "step": 275
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.24013702137397e-06,
      "loss": 0.1485,
      "step": 276
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.174206545276678e-06,
      "loss": 0.0733,
      "step": 277
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.108312192060298e-06,
      "loss": 0.4001,
      "step": 278
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.042456844157299e-06,
      "loss": 0.1089,
      "step": 279
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.97664338229395e-06,
      "loss": 0.0931,
      "step": 280
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.910874685364275e-06,
      "loss": 0.5483,
      "step": 281
    },
    {
      "epoch": 0.56,
      "learning_rate": 8.84515363030414e-06,
      "loss": 0.071,
      "step": 282
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.7794830919654e-06,
      "loss": 0.2319,
      "step": 283
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.713865942990143e-06,
      "loss": 0.264,
      "step": 284
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.648305053685035e-06,
      "loss": 0.3833,
      "step": 285
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.582803291895758e-06,
      "loss": 0.3883,
      "step": 286
    },
    {
      "epoch": 0.57,
      "learning_rate": 8.51736352288158e-06,
      "loss": 0.1494,
      "step": 287
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.451988609189987e-06,
      "loss": 0.1005,
      "step": 288
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.386681410531493e-06,
      "loss": 0.1057,
      "step": 289
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.321444783654524e-06,
      "loss": 0.1763,
      "step": 290
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.256281582220486e-06,
      "loss": 0.2966,
      "step": 291
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.191194656678905e-06,
      "loss": 0.12,
      "step": 292
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.126186854142752e-06,
      "loss": 0.4565,
      "step": 293
    },
    {
      "epoch": 0.59,
      "learning_rate": 8.06126101826392e-06,
      "loss": 0.1963,
      "step": 294
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.99641998910879e-06,
      "loss": 0.0081,
      "step": 295
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.931666603034034e-06,
      "loss": 0.0224,
      "step": 296
    },
    {
      "epoch": 0.59,
      "learning_rate": 7.867003692562533e-06,
      "loss": 0.2567,
      "step": 297
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.80243408625947e-06,
      "loss": 0.0473,
      "step": 298
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.7379606086086e-06,
      "loss": 0.5308,
      "step": 299
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.673586079888699e-06,
      "loss": 0.0563,
      "step": 300
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.6093133160502e-06,
      "loss": 0.1527,
      "step": 301
    },
    {
      "epoch": 0.6,
      "learning_rate": 7.545145128592009e-06,
      "loss": 0.6643,
      "step": 302
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.481084324438521e-06,
      "loss": 0.162,
      "step": 303
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.4171337058168365e-06,
      "loss": 0.0417,
      "step": 304
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.353296070134186e-06,
      "loss": 0.361,
      "step": 305
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.28957420985556e-06,
      "loss": 0.2668,
      "step": 306
    },
    {
      "epoch": 0.61,
      "learning_rate": 7.225970912381557e-06,
      "loss": 0.1708,
      "step": 307
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.16248895992645e-06,
      "loss": 0.0891,
      "step": 308
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.099131129396501e-06,
      "loss": 0.2068,
      "step": 309
    },
    {
      "epoch": 0.62,
      "learning_rate": 7.035900192268464e-06,
      "loss": 0.2319,
      "step": 310
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.972798914468369e-06,
      "loss": 0.2967,
      "step": 311
    },
    {
      "epoch": 0.62,
      "learning_rate": 6.909830056250527e-06,
      "loss": 0.109,
      "step": 312
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.846996372076786e-06,
      "loss": 0.0556,
      "step": 313
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.784300610496049e-06,
      "loss": 0.2744,
      "step": 314
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.721745514024023e-06,
      "loss": 0.3622,
      "step": 315
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.659333819023291e-06,
      "loss": 0.1994,
      "step": 316
    },
    {
      "epoch": 0.63,
      "learning_rate": 6.59706825558357e-06,
      "loss": 0.4396,
      "step": 317
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.534951547402322e-06,
      "loss": 0.3304,
      "step": 318
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.47298641166559e-06,
      "loss": 0.3576,
      "step": 319
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.411175558929152e-06,
      "loss": 0.1696,
      "step": 320
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.349521692999945e-06,
      "loss": 0.072,
      "step": 321
    },
    {
      "epoch": 0.64,
      "learning_rate": 6.2880275108177915e-06,
      "loss": 0.568,
      "step": 322
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.2266957023374434e-06,
      "loss": 0.3337,
      "step": 323
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.165528950410884e-06,
      "loss": 0.0888,
      "step": 324
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.10452993067e-06,
      "loss": 0.2292,
      "step": 325
    },
    {
      "epoch": 0.65,
      "learning_rate": 6.04370131140952e-06,
      "loss": 0.1889,
      "step": 326
    },
    {
      "epoch": 0.65,
      "learning_rate": 5.983045753470308e-06,
      "loss": 0.0938,
      "step": 327
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.922565910122967e-06,
      "loss": 0.3774,
      "step": 328
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.862264426951768e-06,
      "loss": 0.1941,
      "step": 329
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.802143941738945e-06,
      "loss": 0.1338,
      "step": 330
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.742207084349274e-06,
      "loss": 0.3389,
      "step": 331
    },
    {
      "epoch": 0.66,
      "learning_rate": 5.6824564766150724e-06,
      "loss": 0.1251,
      "step": 332
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.622894732221482e-06,
      "loss": 0.2994,
      "step": 333
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.563524456592163e-06,
      "loss": 0.1573,
      "step": 334
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.504348246775299e-06,
      "loss": 0.1989,
      "step": 335
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.445368691330008e-06,
      "loss": 0.1285,
      "step": 336
    },
    {
      "epoch": 0.67,
      "learning_rate": 5.386588370213124e-06,
      "loss": 0.0982,
      "step": 337
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.328009854666303e-06,
      "loss": 0.2538,
      "step": 338
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.269635707103593e-06,
      "loss": 0.0733,
      "step": 339
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.211468480999304e-06,
      "loss": 0.5152,
      "step": 340
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.153510720776354e-06,
      "loss": 0.1063,
      "step": 341
    },
    {
      "epoch": 0.68,
      "learning_rate": 5.095764961694923e-06,
      "loss": 0.4609,
      "step": 342
    },
    {
      "epoch": 0.69,
      "learning_rate": 5.0382337297415775e-06,
      "loss": 0.1152,
      "step": 343
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.980919541518796e-06,
      "loss": 0.2863,
      "step": 344
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.92382490413483e-06,
      "loss": 0.1122,
      "step": 345
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.866952315094088e-06,
      "loss": 0.0646,
      "step": 346
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.8103042621878515e-06,
      "loss": 0.1197,
      "step": 347
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.753883223385467e-06,
      "loss": 0.0571,
      "step": 348
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.6976916667259555e-06,
      "loss": 0.2951,
      "step": 349
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.641732050210032e-06,
      "loss": 0.1941,
      "step": 350
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.58600682169262e-06,
      "loss": 0.1809,
      "step": 351
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.530518418775734e-06,
      "loss": 0.4181,
      "step": 352
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.4752692687018685e-06,
      "loss": 0.0959,
      "step": 353
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.420261788247841e-06,
      "loss": 0.1603,
      "step": 354
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.365498383619036e-06,
      "loss": 0.361,
      "step": 355
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.3109814503441894e-06,
      "loss": 0.311,
      "step": 356
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.256713373170565e-06,
      "loss": 0.1056,
      "step": 357
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.202696525959667e-06,
      "loss": 0.1141,
      "step": 358
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.148933271583385e-06,
      "loss": 0.0853,
      "step": 359
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.0954259618206295e-06,
      "loss": 0.6156,
      "step": 360
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.042176937254474e-06,
      "loss": 0.1291,
      "step": 361
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.989188527169749e-06,
      "loss": 0.0533,
      "step": 362
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.936463049451179e-06,
      "loss": 0.1032,
      "step": 363
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.884002810481959e-06,
      "loss": 0.2597,
      "step": 364
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.83181010504289e-06,
      "loss": 0.2671,
      "step": 365
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.7798872162119948e-06,
      "loss": 0.2067,
      "step": 366
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.72823641526463e-06,
      "loss": 0.0933,
      "step": 367
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.676859961574162e-06,
      "loss": 0.0428,
      "step": 368
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.625760102513103e-06,
      "loss": 0.7957,
      "step": 369
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.5749390733548382e-06,
      "loss": 0.1478,
      "step": 370
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.5243990971758124e-06,
      "loss": 0.3287,
      "step": 371
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.4741423847583134e-06,
      "loss": 0.1473,
      "step": 372
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.4241711344937557e-06,
      "loss": 0.0655,
      "step": 373
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.3744875322865035e-06,
      "loss": 0.3421,
      "step": 374
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.3250937514582758e-06,
      "loss": 0.0579,
      "step": 375
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.2759919526530536e-06,
      "loss": 0.4689,
      "step": 376
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.2271842837425917e-06,
      "loss": 0.3456,
      "step": 377
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.178672879732435e-06,
      "loss": 0.2375,
      "step": 378
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.1304598626685544e-06,
      "loss": 0.0537,
      "step": 379
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.0825473415445073e-06,
      "loss": 0.368,
      "step": 380
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.034937412209178e-06,
      "loss": 0.1693,
      "step": 381
    },
    {
      "epoch": 0.76,
      "learning_rate": 2.9876321572751143e-06,
      "loss": 0.0516,
      "step": 382
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.9406336460274144e-06,
      "loss": 0.2091,
      "step": 383
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.8939439343332086e-06,
      "loss": 0.0812,
      "step": 384
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.847565064551747e-06,
      "loss": 0.5318,
      "step": 385
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.8014990654450325e-06,
      "loss": 0.1424,
      "step": 386
    },
    {
      "epoch": 0.77,
      "learning_rate": 2.7557479520891104e-06,
      "loss": 0.2126,
      "step": 387
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.7103137257858867e-06,
      "loss": 0.0954,
      "step": 388
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.6651983739756026e-06,
      "loss": 0.2334,
      "step": 389
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.6204038701499056e-06,
      "loss": 0.2642,
      "step": 390
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.5759321737655017e-06,
      "loss": 0.2249,
      "step": 391
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.5317852301584642e-06,
      "loss": 0.1975,
      "step": 392
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.487964970459118e-06,
      "loss": 0.0917,
      "step": 393
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.4444733115075823e-06,
      "loss": 0.0956,
      "step": 394
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.401312155769916e-06,
      "loss": 0.0797,
      "step": 395
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.3584833912548887e-06,
      "loss": 0.3195,
      "step": 396
    },
    {
      "epoch": 0.79,
      "learning_rate": 2.315988891431412e-06,
      "loss": 0.3054,
      "step": 397
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.2738305151465646e-06,
      "loss": 0.1219,
      "step": 398
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.2320101065443055e-06,
      "loss": 0.1354,
      "step": 399
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.190529494984782e-06,
      "loss": 0.1462,
      "step": 400
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.149390494964323e-06,
      "loss": 0.072,
      "step": 401
    },
    {
      "epoch": 0.8,
      "learning_rate": 2.1085949060360654e-06,
      "loss": 0.4922,
      "step": 402
    },
    {
      "epoch": 0.81,
      "learning_rate": 2.0681445127312218e-06,
      "loss": 0.0753,
      "step": 403
    },
    {
      "epoch": 0.81,
      "learning_rate": 2.0280410844810426e-06,
      "loss": 0.109,
      "step": 404
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.988286375539391e-06,
      "loss": 0.1175,
      "step": 405
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.9488821249060297e-06,
      "loss": 0.0664,
      "step": 406
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 0.0725,
      "step": 407
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.8711318778368792e-06,
      "loss": 0.3384,
      "step": 408
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.8327892824487792e-06,
      "loss": 0.1832,
      "step": 409
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7948039473155553e-06,
      "loss": 0.4133,
      "step": 410
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7571775340388275e-06,
      "loss": 0.2031,
      "step": 411
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7199116885197996e-06,
      "loss": 0.0999,
      "step": 412
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.6830080408872852e-06,
      "loss": 0.1729,
      "step": 413
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.646468205426377e-06,
      "loss": 0.0638,
      "step": 414
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.6102937805078544e-06,
      "loss": 0.2655,
      "step": 415
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5744863485182537e-06,
      "loss": 0.1536,
      "step": 416
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.5390474757906449e-06,
      "loss": 0.3617,
      "step": 417
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.5039787125361327e-06,
      "loss": 0.3315,
      "step": 418
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4692815927760274e-06,
      "loss": 0.052,
      "step": 419
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4349576342747462e-06,
      "loss": 0.2473,
      "step": 420
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.4010083384734308e-06,
      "loss": 0.0813,
      "step": 421
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.367435190424261e-06,
      "loss": 0.1755,
      "step": 422
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.3342396587254957e-06,
      "loss": 0.1983,
      "step": 423
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.3014231954572287e-06,
      "loss": 0.1303,
      "step": 424
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.26898723611787e-06,
      "loss": 0.3482,
      "step": 425
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.2369331995613664e-06,
      "loss": 0.0799,
      "step": 426
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.2052624879351105e-06,
      "loss": 0.1366,
      "step": 427
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.1739764866186309e-06,
      "loss": 0.0649,
      "step": 428
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.143076564162977e-06,
      "loss": 0.073,
      "step": 429
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.112564072230863e-06,
      "loss": 0.1632,
      "step": 430
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0824403455375287e-06,
      "loss": 0.056,
      "step": 431
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.0527067017923654e-06,
      "loss": 0.161,
      "step": 432
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.0233644416412792e-06,
      "loss": 0.0551,
      "step": 433
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.944148486097793e-07,
      "loss": 0.0524,
      "step": 434
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.658591890468515e-07,
      "loss": 0.0555,
      "step": 435
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.376987120695547e-07,
      "loss": 0.4678,
      "step": 436
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.09934649508375e-07,
      "loss": 0.3581,
      "step": 437
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.825682158533555e-07,
      "loss": 0.13,
      "step": 438
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.55600608200956e-07,
      "loss": 0.3564,
      "step": 439
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.290330062017015e-07,
      "loss": 0.2288,
      "step": 440
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.02866572008566e-07,
      "loss": 0.5636,
      "step": 441
    },
    {
      "epoch": 0.88,
      "learning_rate": 7.771024502261526e-07,
      "loss": 0.3019,
      "step": 442
    },
    {
      "epoch": 0.89,
      "learning_rate": 7.517417678606121e-07,
      "loss": 0.4403,
      "step": 443
    },
    {
      "epoch": 0.89,
      "learning_rate": 7.267856342703461e-07,
      "loss": 0.1987,
      "step": 444
    },
    {
      "epoch": 0.89,
      "learning_rate": 7.022351411174866e-07,
      "loss": 0.1701,
      "step": 445
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.780913623201346e-07,
      "loss": 0.2195,
      "step": 446
    },
    {
      "epoch": 0.89,
      "learning_rate": 6.543553540053926e-07,
      "loss": 0.351,
      "step": 447
    },
    {
      "epoch": 0.9,
      "learning_rate": 6.310281544631547e-07,
      "loss": 0.0955,
      "step": 448
    },
    {
      "epoch": 0.9,
      "learning_rate": 6.081107841007006e-07,
      "loss": 0.2191,
      "step": 449
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.856042453980526e-07,
      "loss": 0.2032,
      "step": 450
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.63509522864123e-07,
      "loss": 0.3534,
      "step": 451
    },
    {
      "epoch": 0.9,
      "learning_rate": 5.418275829936537e-07,
      "loss": 0.1667,
      "step": 452
    },
    {
      "epoch": 0.91,
      "learning_rate": 5.205593742249326e-07,
      "loss": 0.1871,
      "step": 453
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.997058268983135e-07,
      "loss": 0.3943,
      "step": 454
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.792678532155115e-07,
      "loss": 0.4139,
      "step": 455
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.592463471997022e-07,
      "loss": 0.0439,
      "step": 456
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.396421846564236e-07,
      "loss": 0.3162,
      "step": 457
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.204562231352516e-07,
      "loss": 0.2204,
      "step": 458
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.016893018922996e-07,
      "loss": 0.0393,
      "step": 459
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.83342241853496e-07,
      "loss": 0.0726,
      "step": 460
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.6541584557868604e-07,
      "loss": 0.2546,
      "step": 461
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.4791089722651437e-07,
      "loss": 0.1004,
      "step": 462
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.3082816252012927e-07,
      "loss": 0.3242,
      "step": 463
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.1416838871368925e-07,
      "loss": 0.1316,
      "step": 464
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.9793230455966936e-07,
      "loss": 0.0682,
      "step": 465
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.8212062027698995e-07,
      "loss": 0.0692,
      "step": 466
    },
    {
      "epoch": 0.93,
      "learning_rate": 2.667340275199426e-07,
      "loss": 0.2308,
      "step": 467
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.5177319934793995e-07,
      "loss": 0.1639,
      "step": 468
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.3723879019607376e-07,
      "loss": 0.3444,
      "step": 469
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.2313143584648423e-07,
      "loss": 0.1779,
      "step": 470
    },
    {
      "epoch": 0.94,
      "learning_rate": 2.0945175340055356e-07,
      "loss": 0.3289,
      "step": 471
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.9620034125190645e-07,
      "loss": 0.3088,
      "step": 472
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.833777790602398e-07,
      "loss": 0.4307,
      "step": 473
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.7098462772596302e-07,
      "loss": 0.233,
      "step": 474
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.5902142936566333e-07,
      "loss": 0.2091,
      "step": 475
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.474887072883935e-07,
      "loss": 0.0338,
      "step": 476
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.3638696597277678e-07,
      "loss": 0.2794,
      "step": 477
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.2571669104494254e-07,
      "loss": 0.1669,
      "step": 478
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.154783492572853e-07,
      "loss": 0.2617,
      "step": 479
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.0567238846803995e-07,
      "loss": 0.1099,
      "step": 480
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.629923762170091e-08,
      "loss": 0.2098,
      "step": 481
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.735930673024806e-08,
      "loss": 0.13,
      "step": 482
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.885298685522235e-08,
      "loss": 0.1898,
      "step": 483
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.078065009060941e-08,
      "loss": 0.0568,
      "step": 484
    },
    {
      "epoch": 0.97,
      "learning_rate": 6.314264954657257e-08,
      "loss": 0.1637,
      "step": 485
    },
    {
      "epoch": 0.97,
      "learning_rate": 5.5939319333998546e-08,
      "loss": 0.16,
      "step": 486
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.9170974549885844e-08,
      "loss": 0.1184,
      "step": 487
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.2837911263562406e-08,
      "loss": 0.0857,
      "step": 488
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.694040650373154e-08,
      "loss": 0.2791,
      "step": 489
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.147871824635717e-08,
      "loss": 0.1298,
      "step": 490
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.645308540337843e-08,
      "loss": 0.1037,
      "step": 491
    },
    {
      "epoch": 0.98,
      "learning_rate": 2.1863727812254653e-08,
      "loss": 0.0524,
      "step": 492
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.7710846226355327e-08,
      "loss": 0.0859,
      "step": 493
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.3994622306173766e-08,
      "loss": 0.1147,
      "step": 494
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.0715218611384582e-08,
      "loss": 0.114,
      "step": 495
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.872778593728258e-09,
      "loss": 0.1548,
      "step": 496
    },
    {
      "epoch": 0.99,
      "learning_rate": 5.467426590739511e-09,
      "loss": 0.0877,
      "step": 497
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.499267820307184e-09,
      "loss": 0.1934,
      "step": 498
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.9683883760723834e-09,
      "loss": 0.1438,
      "step": 499
    },
    {
      "epoch": 1.0,
      "learning_rate": 8.748552236603758e-10,
      "loss": 0.0584,
      "step": 500
    },
    {
      "epoch": 1.0,
      "step": 500,
      "total_flos": 20389429714944.0,
      "train_loss": 0.23822226324118673,
      "train_runtime": 2064.6478,
      "train_samples_per_second": 0.969,
      "train_steps_per_second": 0.242
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 20389429714944.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}