{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1044,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 6.25e-07,
      "loss": 1.2271,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.25e-06,
      "loss": 1.2473,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 1.2508,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.5e-06,
      "loss": 1.2683,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.125e-06,
      "loss": 1.2469,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 1.1197,
      "step": 6
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 1.1028,
      "step": 7
    },
    {
      "epoch": 0.02,
      "learning_rate": 5e-06,
      "loss": 1.0605,
      "step": 8
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.625e-06,
      "loss": 0.9978,
      "step": 9
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.25e-06,
      "loss": 1.0445,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 6.875e-06,
      "loss": 1.0783,
      "step": 11
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.0416,
      "step": 12
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.125000000000001e-06,
      "loss": 1.0233,
      "step": 13
    },
    {
      "epoch": 0.04,
      "learning_rate": 8.750000000000001e-06,
      "loss": 1.0615,
      "step": 14
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.375000000000001e-06,
      "loss": 1.0445,
      "step": 15
    },
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 1.0157,
      "step": 16
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.0625e-05,
      "loss": 0.9636,
      "step": 17
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.125e-05,
      "loss": 1.0323,
      "step": 18
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.1875e-05,
      "loss": 0.9944,
      "step": 19
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.25e-05,
      "loss": 0.9794,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.3125e-05,
      "loss": 0.9469,
      "step": 21
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.375e-05,
      "loss": 0.9508,
      "step": 22
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.4375e-05,
      "loss": 0.9706,
      "step": 23
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.9545,
      "step": 24
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.5625e-05,
      "loss": 0.9334,
      "step": 25
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.6250000000000002e-05,
      "loss": 0.9593,
      "step": 26
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.6875e-05,
      "loss": 0.9383,
      "step": 27
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 0.9738,
      "step": 28
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.8125e-05,
      "loss": 0.9316,
      "step": 29
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.9257,
      "step": 30
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9375e-05,
      "loss": 0.9444,
      "step": 31
    },
    {
      "epoch": 0.09,
      "learning_rate": 2e-05,
      "loss": 0.959,
      "step": 32
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9999951815386923e-05,
      "loss": 0.9244,
      "step": 33
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9999807262012045e-05,
      "loss": 0.9203,
      "step": 34
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9999566341268413e-05,
      "loss": 0.9072,
      "step": 35
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9999229055477764e-05,
      "loss": 0.9388,
      "step": 36
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9998795407890486e-05,
      "loss": 0.9029,
      "step": 37
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.999826540268562e-05,
      "loss": 0.9409,
      "step": 38
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9997639044970785e-05,
      "loss": 0.9445,
      "step": 39
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.999691634078213e-05,
      "loss": 0.9054,
      "step": 40
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9996097297084306e-05,
      "loss": 0.8996,
      "step": 41
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.999518192177038e-05,
      "loss": 0.9194,
      "step": 42
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.999417022366174e-05,
      "loss": 0.8625,
      "step": 43
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9993062212508054e-05,
      "loss": 0.8853,
      "step": 44
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9991857898987136e-05,
      "loss": 0.9452,
      "step": 45
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9990557294704857e-05,
      "loss": 0.9043,
      "step": 46
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9989160412195047e-05,
      "loss": 0.9117,
      "step": 47
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.998766726491935e-05,
      "loss": 0.9192,
      "step": 48
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9986077867267113e-05,
      "loss": 0.9096,
      "step": 49
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.998439223455524e-05,
      "loss": 0.8721,
      "step": 50
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9982610383028036e-05,
      "loss": 0.8902,
      "step": 51
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9980732329857077e-05,
      "loss": 0.9104,
      "step": 52
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9978758093141004e-05,
      "loss": 0.9263,
      "step": 53
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9976687691905394e-05,
      "loss": 0.8895,
      "step": 54
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9974521146102535e-05,
      "loss": 0.8907,
      "step": 55
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.997225847661127e-05,
      "loss": 0.8979,
      "step": 56
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9969899705236763e-05,
      "loss": 0.8571,
      "step": 57
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9967444854710313e-05,
      "loss": 0.8903,
      "step": 58
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9964893948689122e-05,
      "loss": 0.8911,
      "step": 59
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.996224701175608e-05,
      "loss": 0.8906,
      "step": 60
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9959504069419512e-05,
      "loss": 0.9162,
      "step": 61
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9956665148112936e-05,
      "loss": 0.8527,
      "step": 62
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.995373027519482e-05,
      "loss": 0.9398,
      "step": 63
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.995069947894831e-05,
      "loss": 0.8867,
      "step": 64
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.994757278858095e-05,
      "loss": 0.8729,
      "step": 65
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9944350234224415e-05,
      "loss": 0.8645,
      "step": 66
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9941031846934213e-05,
      "loss": 0.9029,
      "step": 67
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9937617658689385e-05,
      "loss": 0.8651,
      "step": 68
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.99341077023922e-05,
      "loss": 0.8021,
      "step": 69
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.993050201186783e-05,
      "loss": 0.8731,
      "step": 70
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.992680062186404e-05,
      "loss": 0.8846,
      "step": 71
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9923003568050846e-05,
      "loss": 0.9393,
      "step": 72
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9919110887020147e-05,
      "loss": 0.9075,
      "step": 73
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.991512261628542e-05,
      "loss": 0.8125,
      "step": 74
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.991103879428131e-05,
      "loss": 0.895,
      "step": 75
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9906859460363307e-05,
      "loss": 0.882,
      "step": 76
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9902584654807325e-05,
      "loss": 0.8903,
      "step": 77
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.989821441880933e-05,
      "loss": 0.8958,
      "step": 78
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9893748794484948e-05,
      "loss": 0.8557,
      "step": 79
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.988918782486906e-05,
      "loss": 0.8739,
      "step": 80
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9884531553915374e-05,
      "loss": 0.8728,
      "step": 81
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.987978002649601e-05,
      "loss": 0.8755,
      "step": 82
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.987493328840108e-05,
      "loss": 0.8883,
      "step": 83
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.986999138633821e-05,
      "loss": 0.8865,
      "step": 84
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.986495436793214e-05,
      "loss": 0.9034,
      "step": 85
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9859822281724218e-05,
      "loss": 0.8764,
      "step": 86
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9854595177171968e-05,
      "loss": 0.8963,
      "step": 87
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9849273104648592e-05,
      "loss": 0.8765,
      "step": 88
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9843856115442484e-05,
      "loss": 0.8733,
      "step": 89
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9838344261756755e-05,
      "loss": 0.9034,
      "step": 90
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9832737596708713e-05,
      "loss": 0.8787,
      "step": 91
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9827036174329355e-05,
      "loss": 0.8045,
      "step": 92
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9821240049562844e-05,
      "loss": 0.8915,
      "step": 93
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.981534927826599e-05,
      "loss": 0.8684,
      "step": 94
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9809363917207695e-05,
      "loss": 0.8887,
      "step": 95
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9803284024068428e-05,
      "loss": 0.899,
      "step": 96
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.979710965743964e-05,
      "loss": 0.8874,
      "step": 97
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.979084087682323e-05,
      "loss": 0.9121,
      "step": 98
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9784477742630953e-05,
      "loss": 0.8808,
      "step": 99
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9778020316183834e-05,
      "loss": 0.8384,
      "step": 100
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9771468659711595e-05,
      "loss": 0.8682,
      "step": 101
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9764822836352046e-05,
      "loss": 0.9177,
      "step": 102
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9758082910150472e-05,
      "loss": 0.8694,
      "step": 103
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9751248946059016e-05,
      "loss": 0.8712,
      "step": 104
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9744321009936063e-05,
      "loss": 0.8674,
      "step": 105
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.97372991685456e-05,
      "loss": 0.8819,
      "step": 106
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9730183489556563e-05,
      "loss": 0.8952,
      "step": 107
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9722974041542205e-05,
      "loss": 0.8733,
      "step": 108
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9715670893979416e-05,
      "loss": 0.8581,
      "step": 109
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9708274117248067e-05,
      "loss": 0.8365,
      "step": 110
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9700783782630318e-05,
      "loss": 0.8541,
      "step": 111
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.969319996230995e-05,
      "loss": 0.9049,
      "step": 112
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9685522729371648e-05,
      "loss": 0.8978,
      "step": 113
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9677752157800316e-05,
      "loss": 0.8297,
      "step": 114
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9669888322480344e-05,
      "loss": 0.8915,
      "step": 115
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.966193129919491e-05,
      "loss": 0.8426,
      "step": 116
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9653881164625234e-05,
      "loss": 0.8143,
      "step": 117
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9645737996349828e-05,
      "loss": 0.8841,
      "step": 118
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.963750187284379e-05,
      "loss": 0.8307,
      "step": 119
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9629172873477995e-05,
      "loss": 0.8737,
      "step": 120
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9620751078518368e-05,
      "loss": 0.8872,
      "step": 121
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9612236569125093e-05,
      "loss": 0.8487,
      "step": 122
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9603629427351845e-05,
      "loss": 0.8398,
      "step": 123
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9594929736144978e-05,
      "loss": 0.8553,
      "step": 124
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.958613757934274e-05,
      "loss": 0.9013,
      "step": 125
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.957725304167447e-05,
      "loss": 0.8568,
      "step": 126
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9568276208759772e-05,
      "loss": 0.8238,
      "step": 127
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9559207167107684e-05,
      "loss": 0.8481,
      "step": 128
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9550046004115862e-05,
      "loss": 0.876,
      "step": 129
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9540792808069726e-05,
      "loss": 0.8283,
      "step": 130
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.953144766814161e-05,
      "loss": 0.8722,
      "step": 131
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.95220106743899e-05,
      "loss": 0.8941,
      "step": 132
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.951248191775818e-05,
      "loss": 0.9044,
      "step": 133
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.950286149007434e-05,
      "loss": 0.8664,
      "step": 134
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9493149484049692e-05,
      "loss": 0.8634,
      "step": 135
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.948334599327809e-05,
      "loss": 0.827,
      "step": 136
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.947345111223502e-05,
      "loss": 0.89,
      "step": 137
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9463464936276676e-05,
      "loss": 0.8089,
      "step": 138
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.945338756163907e-05,
      "loss": 0.8678,
      "step": 139
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.944321908543708e-05,
      "loss": 0.83,
      "step": 140
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.943295960566352e-05,
      "loss": 0.8832,
      "step": 141
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9422609221188208e-05,
      "loss": 0.8491,
      "step": 142
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9412168031756993e-05,
      "loss": 0.83,
      "step": 143
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9401636137990814e-05,
      "loss": 0.8981,
      "step": 144
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.939101364138472e-05,
      "loss": 0.8268,
      "step": 145
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.938030064430688e-05,
      "loss": 0.8785,
      "step": 146
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.936949724999762e-05,
      "loss": 0.8416,
      "step": 147
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9358603562568417e-05,
      "loss": 0.8523,
      "step": 148
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9347619687000894e-05,
      "loss": 0.8484,
      "step": 149
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.933654572914581e-05,
      "loss": 0.8295,
      "step": 150
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9325381795722037e-05,
      "loss": 0.8911,
      "step": 151
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.931412799431554e-05,
      "loss": 0.8552,
      "step": 152
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9302784433378333e-05,
      "loss": 0.8467,
      "step": 153
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.929135122222743e-05,
      "loss": 0.8873,
      "step": 154
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9279828471043812e-05,
      "loss": 0.8854,
      "step": 155
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.926821629087133e-05,
      "loss": 0.8758,
      "step": 156
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9256514793615674e-05,
      "loss": 0.8801,
      "step": 157
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.924472409204326e-05,
      "loss": 0.8478,
      "step": 158
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.923284429978017e-05,
      "loss": 0.8111,
      "step": 159
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9220875531311044e-05,
      "loss": 0.888,
      "step": 160
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9208817901977977e-05,
      "loss": 0.855,
      "step": 161
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9196671527979407e-05,
      "loss": 0.8465,
      "step": 162
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9184436526369005e-05,
      "loss": 0.8401,
      "step": 163
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.917211301505453e-05,
      "loss": 0.8287,
      "step": 164
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9159701112796713e-05,
      "loss": 0.827,
      "step": 165
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9147200939208088e-05,
      "loss": 0.8369,
      "step": 166
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9134612614751865e-05,
      "loss": 0.9019,
      "step": 167
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9121936260740752e-05,
      "loss": 0.856,
      "step": 168
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9109171999335794e-05,
      "loss": 0.8381,
      "step": 169
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9096319953545186e-05,
      "loss": 0.7973,
      "step": 170
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.90833802472231e-05,
      "loss": 0.8209,
      "step": 171
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9070353005068484e-05,
      "loss": 0.8138,
      "step": 172
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9057238352623866e-05,
      "loss": 0.8544,
      "step": 173
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9044036416274133e-05,
      "loss": 0.8309,
      "step": 174
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.903074732324533e-05,
      "loss": 0.9077,
      "step": 175
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.901737120160341e-05,
      "loss": 0.8706,
      "step": 176
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9003908180253027e-05,
      "loss": 0.88,
      "step": 177
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.899035838893627e-05,
      "loss": 0.8544,
      "step": 178
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.8976721958231438e-05,
      "loss": 0.8696,
      "step": 179
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.8962999019551755e-05,
      "loss": 0.8376,
      "step": 180
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.894918970514412e-05,
      "loss": 0.8021,
      "step": 181
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.8935294148087822e-05,
      "loss": 0.8437,
      "step": 182
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.8921312482293273e-05,
      "loss": 0.8247,
      "step": 183
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.8907244842500706e-05,
      "loss": 0.8732,
      "step": 184
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.8893091364278875e-05,
      "loss": 0.886,
      "step": 185
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.8878852184023754e-05,
      "loss": 0.8693,
      "step": 186
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8864527438957223e-05,
      "loss": 0.8485,
      "step": 187
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.885011726712574e-05,
      "loss": 0.8348,
      "step": 188
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8835621807399016e-05,
      "loss": 0.8526,
      "step": 189
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8821041199468677e-05,
      "loss": 0.8779,
      "step": 190
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.880637558384691e-05,
      "loss": 0.8438,
      "step": 191
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8791625101865117e-05,
      "loss": 0.8827,
      "step": 192
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8776789895672557e-05,
      "loss": 0.8309,
      "step": 193
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.876187010823496e-05,
      "loss": 0.8135,
      "step": 194
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8746865883333164e-05,
      "loss": 0.7809,
      "step": 195
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8731777365561723e-05,
      "loss": 0.8226,
      "step": 196
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8716604700327516e-05,
      "loss": 0.8534,
      "step": 197
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.870134803384834e-05,
      "loss": 0.8615,
      "step": 198
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8686007513151516e-05,
      "loss": 0.8501,
      "step": 199
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8670583286072443e-05,
      "loss": 0.8773,
      "step": 200
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.865507550125322e-05,
      "loss": 0.8891,
      "step": 201
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8639484308141157e-05,
      "loss": 0.848,
      "step": 202
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.862380985698738e-05,
      "loss": 0.8228,
      "step": 203
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.860805229884536e-05,
      "loss": 0.8261,
      "step": 204
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.859221178556947e-05,
      "loss": 0.8762,
      "step": 205
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.8576288469813506e-05,
      "loss": 0.8023,
      "step": 206
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.856028250502923e-05,
      "loss": 0.8548,
      "step": 207
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8544194045464888e-05,
      "loss": 0.8398,
      "step": 208
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8528023246163718e-05,
      "loss": 0.8303,
      "step": 209
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8511770262962462e-05,
      "loss": 0.8278,
      "step": 210
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.849543525248986e-05,
      "loss": 0.8889,
      "step": 211
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8479018372165148e-05,
      "loss": 0.8089,
      "step": 212
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8462519780196532e-05,
      "loss": 0.8227,
      "step": 213
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8445939635579658e-05,
      "loss": 0.8681,
      "step": 214
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.84292780980961e-05,
      "loss": 0.8286,
      "step": 215
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8412535328311813e-05,
      "loss": 0.8632,
      "step": 216
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8395711487575564e-05,
      "loss": 0.8518,
      "step": 217
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.837880673801741e-05,
      "loss": 0.8549,
      "step": 218
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.836182124254711e-05,
      "loss": 0.8226,
      "step": 219
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8344755164852572e-05,
      "loss": 0.8434,
      "step": 220
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8327608669398265e-05,
      "loss": 0.8192,
      "step": 221
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8310381921423644e-05,
      "loss": 0.8609,
      "step": 222
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.829307508694154e-05,
      "loss": 0.8501,
      "step": 223
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8275688332736577e-05,
      "loss": 0.8059,
      "step": 224
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8258221826363563e-05,
      "loss": 0.8188,
      "step": 225
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8240675736145866e-05,
      "loss": 0.8365,
      "step": 226
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8223050231173802e-05,
      "loss": 0.8561,
      "step": 227
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8205345481303e-05,
      "loss": 0.8405,
      "step": 228
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.818756165715276e-05,
      "loss": 0.8013,
      "step": 229
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.816969893010442e-05,
      "loss": 0.8371,
      "step": 230
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8151757472299703e-05,
      "loss": 0.8386,
      "step": 231
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8133737456639043e-05,
      "loss": 0.8295,
      "step": 232
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8115639056779942e-05,
      "loss": 0.8214,
      "step": 233
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8097462447135273e-05,
      "loss": 0.8302,
      "step": 234
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.807920780287162e-05,
      "loss": 0.8582,
      "step": 235
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.806087529990758e-05,
      "loss": 0.8378,
      "step": 236
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.804246511491206e-05,
      "loss": 0.8319,
      "step": 237
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.802397742530259e-05,
      "loss": 0.9003,
      "step": 238
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8005412409243604e-05,
      "loss": 0.7941,
      "step": 239
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.7986770245644728e-05,
      "loss": 0.8162,
      "step": 240
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.7968051114159046e-05,
      "loss": 0.8004,
      "step": 241
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7949255195181387e-05,
      "loss": 0.8761,
      "step": 242
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7930382669846554e-05,
      "loss": 0.8232,
      "step": 243
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7911433720027624e-05,
      "loss": 0.8329,
      "step": 244
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.7892408528334155e-05,
      "loss": 0.8168,
      "step": 245
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.787330727811045e-05,
      "loss": 0.8542,
      "step": 246
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.7854130153433785e-05,
      "loss": 0.827,
      "step": 247
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.7834877339112613e-05,
      "loss": 0.8665,
      "step": 248
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7815549020684826e-05,
      "loss": 0.8343,
      "step": 249
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.779614538441593e-05,
      "loss": 0.8466,
      "step": 250
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7776666617297262e-05,
      "loss": 0.8017,
      "step": 251
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.77571129070442e-05,
      "loss": 0.8056,
      "step": 252
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7737484442094332e-05,
      "loss": 0.8191,
      "step": 253
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.771778141160566e-05,
      "loss": 0.7749,
      "step": 254
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.769800400545476e-05,
      "loss": 0.789,
      "step": 255
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.767815241423497e-05,
      "loss": 0.7964,
      "step": 256
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.765822682925453e-05,
      "loss": 0.8225,
      "step": 257
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.763822744253477e-05,
      "loss": 0.7917,
      "step": 258
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.761815444680822e-05,
      "loss": 0.8081,
      "step": 259
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.75980080355168e-05,
      "loss": 0.8124,
      "step": 260
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.757778840280991e-05,
      "loss": 0.8173,
      "step": 261
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.7557495743542586e-05,
      "loss": 0.8249,
      "step": 262
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7537130253273613e-05,
      "loss": 0.8237,
      "step": 263
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7516692128263647e-05,
      "loss": 0.8059,
      "step": 264
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.749618156547332e-05,
      "loss": 0.8218,
      "step": 265
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7475598762561333e-05,
      "loss": 0.8635,
      "step": 266
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.745494391788257e-05,
      "loss": 0.8143,
      "step": 267
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7434217230486162e-05,
      "loss": 0.8001,
      "step": 268
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7413418900113606e-05,
      "loss": 0.8182,
      "step": 269
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.739254912719679e-05,
      "loss": 0.8301,
      "step": 270
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7371608112856102e-05,
      "loss": 0.8144,
      "step": 271
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7350596058898482e-05,
      "loss": 0.7961,
      "step": 272
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7329513167815465e-05,
      "loss": 0.8609,
      "step": 273
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.730835964278124e-05,
      "loss": 0.8234,
      "step": 274
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.7287135687650696e-05,
      "loss": 0.8427,
      "step": 275
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.726584150695744e-05,
      "loss": 0.8402,
      "step": 276
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7244477305911845e-05,
      "loss": 0.7912,
      "step": 277
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7223043290399065e-05,
      "loss": 0.7565,
      "step": 278
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7201539666977044e-05,
      "loss": 0.8978,
      "step": 279
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7179966642874543e-05,
      "loss": 0.8126,
      "step": 280
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7158324425989126e-05,
      "loss": 0.8668,
      "step": 281
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7136613224885157e-05,
      "loss": 0.8223,
      "step": 282
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7114833248791803e-05,
      "loss": 0.82,
      "step": 283
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7092984707601007e-05,
      "loss": 0.8593,
      "step": 284
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7071067811865477e-05,
      "loss": 0.8276,
      "step": 285
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7049082772796635e-05,
      "loss": 0.8344,
      "step": 286
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7027029802262598e-05,
      "loss": 0.8584,
      "step": 287
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.7004909112786142e-05,
      "loss": 0.7941,
      "step": 288
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.698272091754264e-05,
      "loss": 0.82,
      "step": 289
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.696046543035801e-05,
      "loss": 0.8247,
      "step": 290
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.6938142865706657e-05,
      "loss": 0.861,
      "step": 291
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.6915753438709415e-05,
      "loss": 0.8362,
      "step": 292
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.689329736513146e-05,
      "loss": 0.7785,
      "step": 293
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.687077486138023e-05,
      "loss": 0.7973,
      "step": 294
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.684818614450335e-05,
      "loss": 0.765,
      "step": 295
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.6825531432186545e-05,
      "loss": 0.8097,
      "step": 296
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.6802810942751514e-05,
      "loss": 0.8379,
      "step": 297
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.6780024895153862e-05,
      "loss": 0.8151,
      "step": 298
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.6757173508980965e-05,
      "loss": 0.8456,
      "step": 299
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.6734257004449863e-05,
      "loss": 0.8175,
      "step": 300
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.6711275602405138e-05,
      "loss": 0.8013,
      "step": 301
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.6688229524316782e-05,
      "loss": 0.8527,
      "step": 302
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.6665118992278064e-05,
      "loss": 0.8717,
      "step": 303
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.6641944229003395e-05,
      "loss": 0.8115,
      "step": 304
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.6618705457826178e-05,
      "loss": 0.815,
      "step": 305
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.6595402902696648e-05,
      "loss": 0.8107,
      "step": 306
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.6572036788179728e-05,
      "loss": 0.7771,
      "step": 307
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6548607339452853e-05,
      "loss": 0.8271,
      "step": 308
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6525114782303808e-05,
      "loss": 0.8265,
      "step": 309
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6501559343128553e-05,
      "loss": 0.7926,
      "step": 310
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6477941248929027e-05,
      "loss": 0.812,
      "step": 311
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.645426072731098e-05,
      "loss": 0.8699,
      "step": 312
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.643051800648176e-05,
      "loss": 0.8545,
      "step": 313
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.640671331524814e-05,
      "loss": 0.8527,
      "step": 314
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6382846883014078e-05,
      "loss": 0.8058,
      "step": 315
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6358918939778538e-05,
      "loss": 0.7999,
      "step": 316
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.633492971613326e-05,
      "loss": 0.8356,
      "step": 317
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.631087944326053e-05,
      "loss": 0.7557,
      "step": 318
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6286768352930972e-05,
      "loss": 0.8536,
      "step": 319
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6262596677501297e-05,
      "loss": 0.866,
      "step": 320
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.623836464991207e-05,
      "loss": 0.797,
      "step": 321
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.621407250368546e-05,
      "loss": 0.8338,
      "step": 322
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.618972047292301e-05,
      "loss": 0.8398,
      "step": 323
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.616530879230335e-05,
      "loss": 0.841,
      "step": 324
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.614083769707996e-05,
      "loss": 0.832,
      "step": 325
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.611630742307889e-05,
      "loss": 0.823,
      "step": 326
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.609171820669649e-05,
      "loss": 0.8725,
      "step": 327
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.606707028489714e-05,
      "loss": 0.8493,
      "step": 328
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6042363895210948e-05,
      "loss": 0.8412,
      "step": 329
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6017599275731482e-05,
      "loss": 0.8021,
      "step": 330
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.599277666511347e-05,
      "loss": 0.8083,
      "step": 331
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.5967896302570485e-05,
      "loss": 0.784,
      "step": 332
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.5942958427872654e-05,
      "loss": 0.8156,
      "step": 333
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.5917963281344346e-05,
      "loss": 0.841,
      "step": 334
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.5892911103861857e-05,
      "loss": 0.8654,
      "step": 335
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.5867802136851078e-05,
      "loss": 0.8117,
      "step": 336
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.5842636622285187e-05,
      "loss": 0.7831,
      "step": 337
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.5817414802682292e-05,
      "loss": 0.7931,
      "step": 338
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.5792136921103125e-05,
      "loss": 0.7862,
      "step": 339
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.5766803221148676e-05,
      "loss": 0.8173,
      "step": 340
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.5741413946957838e-05,
      "loss": 0.7993,
      "step": 341
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.5715969343205094e-05,
      "loss": 0.8151,
      "step": 342
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.569046965509812e-05,
      "loss": 0.8136,
      "step": 343
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.566491512837543e-05,
      "loss": 0.8148,
      "step": 344
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.563930600930403e-05,
      "loss": 0.7932,
      "step": 345
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.561364254467701e-05,
      "loss": 0.8425,
      "step": 346
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5587924981811196e-05,
      "loss": 0.7843,
      "step": 347
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5562153568544753e-05,
      "loss": 0.7302,
      "step": 348
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5536328553234795e-05,
      "loss": 0.5562,
      "step": 349
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5510450184754996e-05,
      "loss": 0.5453,
      "step": 350
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5484518712493188e-05,
      "loss": 0.5287,
      "step": 351
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5458534386348967e-05,
      "loss": 0.5338,
      "step": 352
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.543249745673127e-05,
      "loss": 0.517,
      "step": 353
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5406408174555978e-05,
      "loss": 0.5533,
      "step": 354
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5380266791243477e-05,
      "loss": 0.5235,
      "step": 355
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.535407355871626e-05,
      "loss": 0.5263,
      "step": 356
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5327828729396482e-05,
      "loss": 0.4674,
      "step": 357
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5301532556203524e-05,
      "loss": 0.5238,
      "step": 358
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5275185292551585e-05,
      "loss": 0.4859,
      "step": 359
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5248787192347196e-05,
      "loss": 0.4826,
      "step": 360
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.5222338509986813e-05,
      "loss": 0.4805,
      "step": 361
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.5195839500354337e-05,
      "loss": 0.4844,
      "step": 362
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.5169290418818675e-05,
      "loss": 0.4988,
      "step": 363
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5142691521231267e-05,
      "loss": 0.5306,
      "step": 364
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5116043063923638e-05,
      "loss": 0.506,
      "step": 365
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5089345303704901e-05,
      "loss": 0.5145,
      "step": 366
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.506259849785931e-05,
      "loss": 0.4746,
      "step": 367
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.503580290414376e-05,
      "loss": 0.5125,
      "step": 368
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5008958780785321e-05,
      "loss": 0.4949,
      "step": 369
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.4982066386478727e-05,
      "loss": 0.479,
      "step": 370
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.49551259803839e-05,
      "loss": 0.4942,
      "step": 371
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.4928137822123451e-05,
      "loss": 0.5008,
      "step": 372
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.4901102171780175e-05,
      "loss": 0.518,
      "step": 373
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.4874019289894536e-05,
      "loss": 0.5137,
      "step": 374
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.4846889437462175e-05,
      "loss": 0.4984,
      "step": 375
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.481971287593138e-05,
      "loss": 0.5068,
      "step": 376
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.479248986720057e-05,
      "loss": 0.4791,
      "step": 377
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.4765220673615774e-05,
      "loss": 0.4947,
      "step": 378
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.4737905557968104e-05,
      "loss": 0.4964,
      "step": 379
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.471054478349121e-05,
      "loss": 0.4973,
      "step": 380
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.4683138613858762e-05,
      "loss": 0.5099,
      "step": 381
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.4655687313181896e-05,
      "loss": 0.5259,
      "step": 382
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.4628191146006671e-05,
      "loss": 0.4938,
      "step": 383
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.4600650377311523e-05,
      "loss": 0.4793,
      "step": 384
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4573065272504708e-05,
      "loss": 0.4899,
      "step": 385
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4545436097421745e-05,
      "loss": 0.4926,
      "step": 386
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4517763118322861e-05,
      "loss": 0.4704,
      "step": 387
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4490046601890405e-05,
      "loss": 0.521,
      "step": 388
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4462286815226313e-05,
      "loss": 0.5006,
      "step": 389
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4434484025849494e-05,
      "loss": 0.5092,
      "step": 390
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4406638501693277e-05,
      "loss": 0.5035,
      "step": 391
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4378750511102826e-05,
      "loss": 0.4653,
      "step": 392
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4350820322832548e-05,
      "loss": 0.4987,
      "step": 393
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4322848206043505e-05,
      "loss": 0.4857,
      "step": 394
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4294834430300822e-05,
      "loss": 0.4825,
      "step": 395
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4266779265571087e-05,
      "loss": 0.5194,
      "step": 396
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4238682982219753e-05,
      "loss": 0.4864,
      "step": 397
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4210545851008529e-05,
      "loss": 0.4895,
      "step": 398
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4182368143092768e-05,
      "loss": 0.4735,
      "step": 399
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4154150130018867e-05,
      "loss": 0.4859,
      "step": 400
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4125892083721625e-05,
      "loss": 0.5107,
      "step": 401
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.4097594276521658e-05,
      "loss": 0.5188,
      "step": 402
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.4069256981122736e-05,
      "loss": 0.4966,
      "step": 403
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.4040880470609186e-05,
      "loss": 0.5021,
      "step": 404
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.401246501844324e-05,
      "loss": 0.4906,
      "step": 405
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.3984010898462417e-05,
      "loss": 0.4709,
      "step": 406
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.3955518384876863e-05,
      "loss": 0.4852,
      "step": 407
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.3926987752266733e-05,
      "loss": 0.5086,
      "step": 408
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.3898419275579523e-05,
      "loss": 0.5089,
      "step": 409
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.3869813230127432e-05,
      "loss": 0.4699,
      "step": 410
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.3841169891584704e-05,
      "loss": 0.5047,
      "step": 411
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.381248953598498e-05,
      "loss": 0.5126,
      "step": 412
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.3783772439718628e-05,
      "loss": 0.52,
      "step": 413
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.3755018879530075e-05,
      "loss": 0.4685,
      "step": 414
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.3726229132515166e-05,
      "loss": 0.4864,
      "step": 415
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.3697403476118453e-05,
      "loss": 0.4919,
      "step": 416
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.3668542188130567e-05,
      "loss": 0.5058,
      "step": 417
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.36396455466855e-05,
      "loss": 0.5109,
      "step": 418
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.3610713830257955e-05,
      "loss": 0.494,
      "step": 419
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3581747317660641e-05,
      "loss": 0.4921,
      "step": 420
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3552746288041598e-05,
      "loss": 0.4629,
      "step": 421
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3523711020881504e-05,
      "loss": 0.487,
      "step": 422
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3494641795990986e-05,
      "loss": 0.5203,
      "step": 423
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3465538893507908e-05,
      "loss": 0.4933,
      "step": 424
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3436402593894696e-05,
      "loss": 0.5071,
      "step": 425
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3407233177935608e-05,
      "loss": 0.5591,
      "step": 426
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3378030926734052e-05,
      "loss": 0.4702,
      "step": 427
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3348796121709862e-05,
      "loss": 0.4922,
      "step": 428
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3319529044596592e-05,
      "loss": 0.5129,
      "step": 429
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3290229977438802e-05,
      "loss": 0.4791,
      "step": 430
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3260899202589335e-05,
      "loss": 0.4556,
      "step": 431
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3231537002706594e-05,
      "loss": 0.5269,
      "step": 432
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3202143660751828e-05,
      "loss": 0.4962,
      "step": 433
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.31727194599864e-05,
      "loss": 0.4685,
      "step": 434
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3143264683969052e-05,
      "loss": 0.5229,
      "step": 435
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3113779616553189e-05,
      "loss": 0.5242,
      "step": 436
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.3084264541884118e-05,
      "loss": 0.478,
      "step": 437
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.3054719744396333e-05,
      "loss": 0.5011,
      "step": 438
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.302514550881076e-05,
      "loss": 0.5483,
      "step": 439
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.2995542120132016e-05,
      "loss": 0.4824,
      "step": 440
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.2965909863645671e-05,
      "loss": 0.5189,
      "step": 441
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.2936249024915487e-05,
      "loss": 0.516,
      "step": 442
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.2906559889780668e-05,
      "loss": 0.4944,
      "step": 443
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.2876842744353113e-05,
      "loss": 0.5199,
      "step": 444
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.2847097875014656e-05,
      "loss": 0.4984,
      "step": 445
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.2817325568414299e-05,
      "loss": 0.4996,
      "step": 446
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.2787526111465453e-05,
      "loss": 0.4989,
      "step": 447
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2757699791343188e-05,
      "loss": 0.552,
      "step": 448
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2727846895481436e-05,
      "loss": 0.5002,
      "step": 449
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2697967711570243e-05,
      "loss": 0.5042,
      "step": 450
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2668062527552998e-05,
      "loss": 0.5072,
      "step": 451
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2638131631623641e-05,
      "loss": 0.488,
      "step": 452
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.26081753122239e-05,
      "loss": 0.5164,
      "step": 453
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2578193858040508e-05,
      "loss": 0.5082,
      "step": 454
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.2548187558002418e-05,
      "loss": 0.507,
      "step": 455
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.251815670127802e-05,
      "loss": 0.46,
      "step": 456
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.248810157727236e-05,
      "loss": 0.5017,
      "step": 457
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2458022475624343e-05,
      "loss": 0.5125,
      "step": 458
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2427919686203939e-05,
      "loss": 0.4661,
      "step": 459
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2397793499109403e-05,
      "loss": 0.485,
      "step": 460
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2367644204664468e-05,
      "loss": 0.5335,
      "step": 461
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2337472093415554e-05,
      "loss": 0.5066,
      "step": 462
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2307277456128958e-05,
      "loss": 0.467,
      "step": 463
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2277060583788065e-05,
      "loss": 0.5245,
      "step": 464
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2246821767590537e-05,
      "loss": 0.5152,
      "step": 465
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2216561298945502e-05,
      "loss": 0.5245,
      "step": 466
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2186279469470757e-05,
      "loss": 0.4941,
      "step": 467
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2155976570989949e-05,
      "loss": 0.525,
      "step": 468
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.2125652895529766e-05,
      "loss": 0.4866,
      "step": 469
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.2095308735317126e-05,
      "loss": 0.4614,
      "step": 470
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.2064944382776344e-05,
      "loss": 0.487,
      "step": 471
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2034560130526341e-05,
      "loss": 0.5255,
      "step": 472
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2004156271377802e-05,
      "loss": 0.5372,
      "step": 473
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.197373309833037e-05,
      "loss": 0.4938,
      "step": 474
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.19432909045698e-05,
      "loss": 0.5137,
      "step": 475
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.1912829983465169e-05,
      "loss": 0.4791,
      "step": 476
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.1882350628566008e-05,
      "loss": 0.4934,
      "step": 477
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.1851853133599507e-05,
      "loss": 0.514,
      "step": 478
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.182133779246766e-05,
      "loss": 0.4839,
      "step": 479
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1790804899244453e-05,
      "loss": 0.5187,
      "step": 480
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.176025474817301e-05,
      "loss": 0.5094,
      "step": 481
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1729687633662776e-05,
      "loss": 0.4961,
      "step": 482
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1699103850286668e-05,
      "loss": 0.5064,
      "step": 483
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1668503692778239e-05,
      "loss": 0.5315,
      "step": 484
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1637887456028837e-05,
      "loss": 0.4978,
      "step": 485
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1607255435084772e-05,
      "loss": 0.5583,
      "step": 486
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1576607925144456e-05,
      "loss": 0.4804,
      "step": 487
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1545945221555571e-05,
      "loss": 0.4719,
      "step": 488
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1515267619812213e-05,
      "loss": 0.5232,
      "step": 489
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1484575415552069e-05,
      "loss": 0.4982,
      "step": 490
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1453868904553525e-05,
      "loss": 0.4909,
      "step": 491
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1423148382732854e-05,
      "loss": 0.5239,
      "step": 492
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1392414146141348e-05,
      "loss": 0.4853,
      "step": 493
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1361666490962468e-05,
      "loss": 0.5296,
      "step": 494
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1330905713508986e-05,
      "loss": 0.4858,
      "step": 495
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1300132110220135e-05,
      "loss": 0.4851,
      "step": 496
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1269345977658747e-05,
      "loss": 0.5619,
      "step": 497
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.12385476125084e-05,
      "loss": 0.5036,
      "step": 498
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1207737311570559e-05,
      "loss": 0.498,
      "step": 499
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1176915371761702e-05,
      "loss": 0.4831,
      "step": 500
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1146082090110486e-05,
      "loss": 0.4823,
      "step": 501
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1115237763754856e-05,
      "loss": 0.5197,
      "step": 502
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1084382689939199e-05,
      "loss": 0.4954,
      "step": 503
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1053517166011471e-05,
      "loss": 0.4959,
      "step": 504
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1022641489420342e-05,
      "loss": 0.4871,
      "step": 505
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.0991755957712317e-05,
      "loss": 0.5253,
      "step": 506
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0960860868528872e-05,
      "loss": 0.475,
      "step": 507
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0929956519603595e-05,
      "loss": 0.5117,
      "step": 508
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0899043208759306e-05,
      "loss": 0.5035,
      "step": 509
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0868121233905183e-05,
      "loss": 0.5099,
      "step": 510
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0837190893033907e-05,
      "loss": 0.4704,
      "step": 511
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.080625248421878e-05,
      "loss": 0.5206,
      "step": 512
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0775306305610856e-05,
      "loss": 0.5057,
      "step": 513
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0744352655436058e-05,
      "loss": 0.5036,
      "step": 514
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0713391831992324e-05,
      "loss": 0.4985,
      "step": 515
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0682424133646712e-05,
      "loss": 0.4762,
      "step": 516
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.065144985883253e-05,
      "loss": 0.5061,
      "step": 517
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0620469306046473e-05,
      "loss": 0.5151,
      "step": 518
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0589482773845728e-05,
      "loss": 0.491,
      "step": 519
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0558490560845107e-05,
      "loss": 0.4809,
      "step": 520
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.052749296571417e-05,
      "loss": 0.49,
      "step": 521
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0496490287174343e-05,
      "loss": 0.49,
      "step": 522
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.046548282399604e-05,
      "loss": 0.5135,
      "step": 523
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0434470874995781e-05,
      "loss": 0.486,
      "step": 524
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0403454739033321e-05,
      "loss": 0.4926,
      "step": 525
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0372434715008763e-05,
      "loss": 0.5231,
      "step": 526
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.034141110185968e-05,
      "loss": 0.5341,
      "step": 527
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0310384198558226e-05,
      "loss": 0.4889,
      "step": 528
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0279354304108272e-05,
      "loss": 0.4909,
      "step": 529
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0248321717542505e-05,
      "loss": 0.4855,
      "step": 530
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.021728673791956e-05,
      "loss": 0.4895,
      "step": 531
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0186249664321139e-05,
      "loss": 0.5353,
      "step": 532
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0155210795849116e-05,
      "loss": 0.5182,
      "step": 533
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.012417043162266e-05,
      "loss": 0.5208,
      "step": 534
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.0093128870775365e-05,
      "loss": 0.5197,
      "step": 535
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.0062086412452353e-05,
      "loss": 0.4805,
      "step": 536
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.0031043355807386e-05,
      "loss": 0.4851,
      "step": 537
    },
    {
      "epoch": 1.55,
      "learning_rate": 1e-05,
      "loss": 0.4963,
      "step": 538
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.968956644192617e-06,
      "loss": 0.533,
      "step": 539
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.93791358754765e-06,
      "loss": 0.5255,
      "step": 540
    },
    {
      "epoch": 1.55,
      "learning_rate": 9.906871129224636e-06,
      "loss": 0.5108,
      "step": 541
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.875829568377343e-06,
      "loss": 0.4985,
      "step": 542
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.84478920415089e-06,
      "loss": 0.4939,
      "step": 543
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.813750335678866e-06,
      "loss": 0.4992,
      "step": 544
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.782713262080445e-06,
      "loss": 0.5088,
      "step": 545
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.751678282457502e-06,
      "loss": 0.4896,
      "step": 546
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.720645695891733e-06,
      "loss": 0.5212,
      "step": 547
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.689615801441776e-06,
      "loss": 0.4752,
      "step": 548
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.658588898140322e-06,
      "loss": 0.5023,
      "step": 549
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.627565284991237e-06,
      "loss": 0.5153,
      "step": 550
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.596545260966679e-06,
      "loss": 0.5053,
      "step": 551
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.56552912500422e-06,
      "loss": 0.5161,
      "step": 552
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.534517176003963e-06,
      "loss": 0.4752,
      "step": 553
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.50350971282566e-06,
      "loss": 0.492,
      "step": 554
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.472507034285832e-06,
      "loss": 0.5076,
      "step": 555
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.441509439154896e-06,
      "loss": 0.5067,
      "step": 556
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.410517226154276e-06,
      "loss": 0.5327,
      "step": 557
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.37953069395353e-06,
      "loss": 0.5016,
      "step": 558
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.348550141167472e-06,
      "loss": 0.4621,
      "step": 559
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.317575866353293e-06,
      "loss": 0.4801,
      "step": 560
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.286608168007678e-06,
      "loss": 0.5026,
      "step": 561
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.255647344563943e-06,
      "loss": 0.4799,
      "step": 562
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.224693694389149e-06,
      "loss": 0.5117,
      "step": 563
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.193747515781225e-06,
      "loss": 0.4816,
      "step": 564
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.162809106966098e-06,
      "loss": 0.5454,
      "step": 565
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.131878766094822e-06,
      "loss": 0.4895,
      "step": 566
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.100956791240699e-06,
      "loss": 0.5117,
      "step": 567
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.070043480396404e-06,
      "loss": 0.5072,
      "step": 568
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.039139131471128e-06,
      "loss": 0.5144,
      "step": 569
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.008244042287684e-06,
      "loss": 0.4847,
      "step": 570
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.977358510579658e-06,
      "loss": 0.5103,
      "step": 571
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.94648283398853e-06,
      "loss": 0.5282,
      "step": 572
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.915617310060803e-06,
      "loss": 0.5067,
      "step": 573
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.884762236245146e-06,
      "loss": 0.5232,
      "step": 574
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.853917909889516e-06,
      "loss": 0.4918,
      "step": 575
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.8230846282383e-06,
      "loss": 0.4871,
      "step": 576
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.792262688429445e-06,
      "loss": 0.4978,
      "step": 577
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.761452387491601e-06,
      "loss": 0.4887,
      "step": 578
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.730654022341256e-06,
      "loss": 0.4774,
      "step": 579
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.699867889779869e-06,
      "loss": 0.4775,
      "step": 580
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.669094286491017e-06,
      "loss": 0.4854,
      "step": 581
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.638333509037537e-06,
      "loss": 0.4997,
      "step": 582
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.607585853858655e-06,
      "loss": 0.4815,
      "step": 583
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.576851617267151e-06,
      "loss": 0.4988,
      "step": 584
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.54613109544648e-06,
      "loss": 0.5063,
      "step": 585
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.515424584447935e-06,
      "loss": 0.5214,
      "step": 586
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.484732380187785e-06,
      "loss": 0.4944,
      "step": 587
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.45405477844443e-06,
      "loss": 0.5081,
      "step": 588
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.423392074855545e-06,
      "loss": 0.4856,
      "step": 589
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.392744564915228e-06,
      "loss": 0.4949,
      "step": 590
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.362112543971162e-06,
      "loss": 0.4709,
      "step": 591
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.331496307221763e-06,
      "loss": 0.4794,
      "step": 592
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.300896149713334e-06,
      "loss": 0.4571,
      "step": 593
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.270312366337227e-06,
      "loss": 0.4835,
      "step": 594
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.239745251826993e-06,
      "loss": 0.514,
      "step": 595
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.209195100755552e-06,
      "loss": 0.4589,
      "step": 596
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.178662207532343e-06,
      "loss": 0.4544,
      "step": 597
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.148146866400498e-06,
      "loss": 0.474,
      "step": 598
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.117649371433994e-06,
      "loss": 0.4911,
      "step": 599
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.087170016534836e-06,
      "loss": 0.4967,
      "step": 600
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.056709095430201e-06,
      "loss": 0.4833,
      "step": 601
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.026266901669635e-06,
      "loss": 0.5223,
      "step": 602
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.995843728622202e-06,
      "loss": 0.4964,
      "step": 603
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.965439869473664e-06,
      "loss": 0.4647,
      "step": 604
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.93505561722366e-06,
      "loss": 0.5033,
      "step": 605
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.90469126468288e-06,
      "loss": 0.4783,
      "step": 606
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.874347104470234e-06,
      "loss": 0.4857,
      "step": 607
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.844023429010051e-06,
      "loss": 0.5099,
      "step": 608
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.813720530529245e-06,
      "loss": 0.5567,
      "step": 609
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.7834387010545e-06,
      "loss": 0.4991,
      "step": 610
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.753178232409466e-06,
      "loss": 0.4742,
      "step": 611
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.722939416211938e-06,
      "loss": 0.5084,
      "step": 612
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.692722543871045e-06,
      "loss": 0.5173,
      "step": 613
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.66252790658445e-06,
      "loss": 0.5057,
      "step": 614
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.632355795335533e-06,
      "loss": 0.5022,
      "step": 615
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.6022065008906e-06,
      "loss": 0.4676,
      "step": 616
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.572080313796064e-06,
      "loss": 0.5001,
      "step": 617
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.541977524375661e-06,
      "loss": 0.4822,
      "step": 618
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.5118984227276416e-06,
      "loss": 0.4821,
      "step": 619
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.481843298721982e-06,
      "loss": 0.4801,
      "step": 620
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.451812441997587e-06,
      "loss": 0.494,
      "step": 621
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.421806141959498e-06,
      "loss": 0.4816,
      "step": 622
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.3918246877761045e-06,
      "loss": 0.4981,
      "step": 623
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.361868368376364e-06,
      "loss": 0.5282,
      "step": 624
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.331937472447006e-06,
      "loss": 0.4948,
      "step": 625
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.3020322884297565e-06,
      "loss": 0.5087,
      "step": 626
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.272153104518567e-06,
      "loss": 0.4855,
      "step": 627
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.242300208656814e-06,
      "loss": 0.5081,
      "step": 628
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.212473888534546e-06,
      "loss": 0.4935,
      "step": 629
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.182674431585703e-06,
      "loss": 0.4855,
      "step": 630
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.152902124985346e-06,
      "loss": 0.4927,
      "step": 631
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.123157255646889e-06,
      "loss": 0.4672,
      "step": 632
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.093440110219336e-06,
      "loss": 0.4636,
      "step": 633
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.0637509750845175e-06,
      "loss": 0.5185,
      "step": 634
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.0340901363543324e-06,
      "loss": 0.4721,
      "step": 635
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.004457879867987e-06,
      "loss": 0.5067,
      "step": 636
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.974854491189243e-06,
      "loss": 0.4621,
      "step": 637
    },
    {
      "epoch": 1.83,
      "learning_rate": 6.9452802556036705e-06,
      "loss": 0.4545,
      "step": 638
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.9157354581158845e-06,
      "loss": 0.5077,
      "step": 639
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.886220383446814e-06,
      "loss": 0.5089,
      "step": 640
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.856735316030951e-06,
      "loss": 0.4659,
      "step": 641
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.827280540013607e-06,
      "loss": 0.4833,
      "step": 642
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.797856339248179e-06,
      "loss": 0.5071,
      "step": 643
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.768462997293412e-06,
      "loss": 0.5002,
      "step": 644
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.73910079741067e-06,
      "loss": 0.507,
      "step": 645
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.709770022561197e-06,
      "loss": 0.5202,
      "step": 646
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.6804709554034075e-06,
      "loss": 0.5023,
      "step": 647
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.651203878290139e-06,
      "loss": 0.4969,
      "step": 648
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.62196907326595e-06,
      "loss": 0.465,
      "step": 649
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.592766822064396e-06,
      "loss": 0.5196,
      "step": 650
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.5635974061053085e-06,
      "loss": 0.4838,
      "step": 651
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.534461106492094e-06,
      "loss": 0.4903,
      "step": 652
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.505358204009018e-06,
      "loss": 0.4998,
      "step": 653
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.476288979118497e-06,
      "loss": 0.4731,
      "step": 654
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.4472537119584055e-06,
      "loss": 0.4717,
      "step": 655
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.418252682339362e-06,
      "loss": 0.4704,
      "step": 656
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.389286169742048e-06,
      "loss": 0.4649,
      "step": 657
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.360354453314502e-06,
      "loss": 0.4869,
      "step": 658
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.331457811869437e-06,
      "loss": 0.45,
      "step": 659
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.30259652388155e-06,
      "loss": 0.5103,
      "step": 660
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.273770867484841e-06,
      "loss": 0.5132,
      "step": 661
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.244981120469928e-06,
      "loss": 0.4875,
      "step": 662
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.216227560281378e-06,
      "loss": 0.4873,
      "step": 663
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.187510464015023e-06,
      "loss": 0.4949,
      "step": 664
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.158830108415297e-06,
      "loss": 0.4926,
      "step": 665
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.13018676987257e-06,
      "loss": 0.5051,
      "step": 666
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.101580724420478e-06,
      "loss": 0.4863,
      "step": 667
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.073012247733267e-06,
      "loss": 0.4702,
      "step": 668
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.044481615123138e-06,
      "loss": 0.4978,
      "step": 669
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.015989101537586e-06,
      "loss": 0.4919,
      "step": 670
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.987534981556762e-06,
      "loss": 0.4918,
      "step": 671
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.959119529390818e-06,
      "loss": 0.4792,
      "step": 672
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.930743018877268e-06,
      "loss": 0.4895,
      "step": 673
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.902405723478346e-06,
      "loss": 0.4959,
      "step": 674
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.874107916278375e-06,
      "loss": 0.469,
      "step": 675
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.845849869981137e-06,
      "loss": 0.5227,
      "step": 676
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.817631856907233e-06,
      "loss": 0.4817,
      "step": 677
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.789454148991477e-06,
      "loss": 0.5045,
      "step": 678
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.76131701778025e-06,
      "loss": 0.5003,
      "step": 679
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.733220734428919e-06,
      "loss": 0.4947,
      "step": 680
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.7051655696991825e-06,
      "loss": 0.5199,
      "step": 681
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.677151793956501e-06,
      "loss": 0.4872,
      "step": 682
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.649179677167456e-06,
      "loss": 0.5149,
      "step": 683
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.621249488897176e-06,
      "loss": 0.5116,
      "step": 684
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.593361498306722e-06,
      "loss": 0.4931,
      "step": 685
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.5655159741505085e-06,
      "loss": 0.4909,
      "step": 686
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.537713184773686e-06,
      "loss": 0.5087,
      "step": 687
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.5099533981095945e-06,
      "loss": 0.4962,
      "step": 688
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.482236881677141e-06,
      "loss": 0.4979,
      "step": 689
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.454563902578256e-06,
      "loss": 0.4763,
      "step": 690
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.426934727495293e-06,
      "loss": 0.4985,
      "step": 691
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.399349622688479e-06,
      "loss": 0.4933,
      "step": 692
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.37180885399333e-06,
      "loss": 0.5072,
      "step": 693
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.344312686818106e-06,
      "loss": 0.4793,
      "step": 694
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.316861386141241e-06,
      "loss": 0.4657,
      "step": 695
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.289455216508792e-06,
      "loss": 0.4125,
      "step": 696
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.262094442031901e-06,
      "loss": 0.2443,
      "step": 697
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.234779326384227e-06,
      "loss": 0.2325,
      "step": 698
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.207510132799436e-06,
      "loss": 0.2481,
      "step": 699
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.180287124068624e-06,
      "loss": 0.2244,
      "step": 700
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.153110562537831e-06,
      "loss": 0.2336,
      "step": 701
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.1259807101054686e-06,
      "loss": 0.2069,
      "step": 702
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.098897828219831e-06,
      "loss": 0.234,
      "step": 703
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.071862177876548e-06,
      "loss": 0.1942,
      "step": 704
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.044874019616101e-06,
      "loss": 0.2181,
      "step": 705
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.017933613521273e-06,
      "loss": 0.2164,
      "step": 706
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.9910412192146795e-06,
      "loss": 0.2041,
      "step": 707
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.964197095856237e-06,
      "loss": 0.1957,
      "step": 708
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.937401502140691e-06,
      "loss": 0.2074,
      "step": 709
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.9106546962951e-06,
      "loss": 0.2104,
      "step": 710
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.883956936076366e-06,
      "loss": 0.2039,
      "step": 711
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.857308478768733e-06,
      "loss": 0.2143,
      "step": 712
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.830709581181329e-06,
      "loss": 0.1926,
      "step": 713
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.804160499645667e-06,
      "loss": 0.2139,
      "step": 714
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.777661490013189e-06,
      "loss": 0.2083,
      "step": 715
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.7512128076528065e-06,
      "loss": 0.2053,
      "step": 716
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.724814707448418e-06,
      "loss": 0.2042,
      "step": 717
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.698467443796479e-06,
      "loss": 0.2028,
      "step": 718
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.672171270603524e-06,
      "loss": 0.1978,
      "step": 719
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.645926441283745e-06,
      "loss": 0.2083,
      "step": 720
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.619733208756526e-06,
      "loss": 0.1911,
      "step": 721
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.593591825444028e-06,
      "loss": 0.2079,
      "step": 722
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.567502543268728e-06,
      "loss": 0.1977,
      "step": 723
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.541465613651034e-06,
      "loss": 0.205,
      "step": 724
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.515481287506811e-06,
      "loss": 0.1975,
      "step": 725
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.489549815245008e-06,
      "loss": 0.1866,
      "step": 726
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.463671446765206e-06,
      "loss": 0.19,
      "step": 727
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.437846431455249e-06,
      "loss": 0.2133,
      "step": 728
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.412075018188805e-06,
      "loss": 0.2095,
      "step": 729
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.3863574553229936e-06,
      "loss": 0.199,
      "step": 730
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.360693990695973e-06,
      "loss": 0.2122,
      "step": 731
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.335084871624572e-06,
      "loss": 0.2057,
      "step": 732
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.309530344901886e-06,
      "loss": 0.2102,
      "step": 733
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.284030656794908e-06,
      "loss": 0.2184,
      "step": 734
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.258586053042166e-06,
      "loss": 0.2099,
      "step": 735
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.2331967788513295e-06,
      "loss": 0.2085,
      "step": 736
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.2078630788968775e-06,
      "loss": 0.2026,
      "step": 737
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.182585197317709e-06,
      "loss": 0.2119,
      "step": 738
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.157363377714819e-06,
      "loss": 0.1838,
      "step": 739
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.132197863148925e-06,
      "loss": 0.2047,
      "step": 740
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.10708889613815e-06,
      "loss": 0.1981,
      "step": 741
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.082036718655657e-06,
      "loss": 0.1987,
      "step": 742
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.057041572127348e-06,
      "loss": 0.1942,
      "step": 743
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.032103697429516e-06,
      "loss": 0.2057,
      "step": 744
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.007223334886531e-06,
      "loss": 0.2144,
      "step": 745
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.982400724268516e-06,
      "loss": 0.1993,
      "step": 746
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.957636104789056e-06,
      "loss": 0.2095,
      "step": 747
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.932929715102863e-06,
      "loss": 0.1995,
      "step": 748
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.908281793303513e-06,
      "loss": 0.2107,
      "step": 749
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.883692576921111e-06,
      "loss": 0.2102,
      "step": 750
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.859162302920042e-06,
      "loss": 0.1994,
      "step": 751
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.834691207696649e-06,
      "loss": 0.1906,
      "step": 752
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.810279527076992e-06,
      "loss": 0.2158,
      "step": 753
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.785927496314543e-06,
      "loss": 0.1981,
      "step": 754
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.7616353500879354e-06,
      "loss": 0.2093,
      "step": 755
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.7374033224987084e-06,
      "loss": 0.1909,
      "step": 756
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.713231647069031e-06,
      "loss": 0.2107,
      "step": 757
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.689120556739475e-06,
      "loss": 0.1996,
      "step": 758
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.665070283866746e-06,
      "loss": 0.2056,
      "step": 759
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.6410810602214687e-06,
      "loss": 0.2148,
      "step": 760
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.6171531169859273e-06,
      "loss": 0.2019,
      "step": 761
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.5932866847518654e-06,
      "loss": 0.2224,
      "step": 762
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.5694819935182403e-06,
      "loss": 0.2059,
      "step": 763
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.545739272689024e-06,
      "loss": 0.2013,
      "step": 764
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.5220587510709734e-06,
      "loss": 0.1987,
      "step": 765
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.498440656871449e-06,
      "loss": 0.2119,
      "step": 766
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.4748852176961912e-06,
      "loss": 0.1992,
      "step": 767
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.4513926605471504e-06,
      "loss": 0.2097,
      "step": 768
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.4279632118202744e-06,
      "loss": 0.2071,
      "step": 769
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.404597097303356e-06,
      "loss": 0.2004,
      "step": 770
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.3812945421738243e-06,
      "loss": 0.203,
      "step": 771
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.3580557709966066e-06,
      "loss": 0.2057,
      "step": 772
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.334881007721941e-06,
      "loss": 0.2026,
      "step": 773
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.3117704756832226e-06,
      "loss": 0.1955,
      "step": 774
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.288724397594867e-06,
      "loss": 0.209,
      "step": 775
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.2657429955501396e-06,
      "loss": 0.2065,
      "step": 776
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.2428264910190398e-06,
      "loss": 0.2289,
      "step": 777
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.2199751048461414e-06,
      "loss": 0.1911,
      "step": 778
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.1971890572484907e-06,
      "loss": 0.1984,
      "step": 779
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.174468567813461e-06,
      "loss": 0.2048,
      "step": 780
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.151813855496655e-06,
      "loss": 0.2068,
      "step": 781
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.1292251386197727e-06,
      "loss": 0.2018,
      "step": 782
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.1067026348685436e-06,
      "loss": 0.1992,
      "step": 783
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.084246561290584e-06,
      "loss": 0.2133,
      "step": 784
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0618571342933436e-06,
      "loss": 0.2064,
      "step": 785
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.039534569641992e-06,
      "loss": 0.21,
      "step": 786
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0172790824573627e-06,
      "loss": 0.2054,
      "step": 787
    },
    {
      "epoch": 2.26,
      "learning_rate": 2.9950908872138585e-06,
      "loss": 0.2362,
      "step": 788
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9729701977374035e-06,
      "loss": 0.2095,
      "step": 789
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9509172272033683e-06,
      "loss": 0.1924,
      "step": 790
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9289321881345257e-06,
      "loss": 0.2121,
      "step": 791
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.9070152923989947e-06,
      "loss": 0.202,
      "step": 792
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.8851667512082004e-06,
      "loss": 0.1999,
      "step": 793
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.863386775114848e-06,
      "loss": 0.1984,
      "step": 794
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.841675574010878e-06,
      "loss": 0.1996,
      "step": 795
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.8200333571254603e-06,
      "loss": 0.209,
      "step": 796
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.798460333022959e-06,
      "loss": 0.2072,
      "step": 797
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.776956709600941e-06,
      "loss": 0.1918,
      "step": 798
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.7555226940881584e-06,
      "loss": 0.2159,
      "step": 799
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.7341584930425656e-06,
      "loss": 0.2148,
      "step": 800
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.712864312349305e-06,
      "loss": 0.1932,
      "step": 801
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.691640357218759e-06,
      "loss": 0.2151,
      "step": 802
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6704868321845344e-06,
      "loss": 0.2123,
      "step": 803
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6494039411015192e-06,
      "loss": 0.2246,
      "step": 804
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6283918871438974e-06,
      "loss": 0.2481,
      "step": 805
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.6074508728032135e-06,
      "loss": 0.2314,
      "step": 806
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.586581099886396e-06,
      "loss": 0.2062,
      "step": 807
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.565782769513837e-06,
      "loss": 0.22,
      "step": 808
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.545056082117433e-06,
      "loss": 0.2049,
      "step": 809
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5244012374386695e-06,
      "loss": 0.2054,
      "step": 810
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.503818434526685e-06,
      "loss": 0.2229,
      "step": 811
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.4833078717363544e-06,
      "loss": 0.2139,
      "step": 812
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4628697467263916e-06,
      "loss": 0.2138,
      "step": 813
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4425042564574186e-06,
      "loss": 0.2216,
      "step": 814
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.422211597190095e-06,
      "loss": 0.21,
      "step": 815
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4019919644832023e-06,
      "loss": 0.2088,
      "step": 816
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.381845553191783e-06,
      "loss": 0.2141,
      "step": 817
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.3617725574652352e-06,
      "loss": 0.211,
      "step": 818
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.3417731707454737e-06,
      "loss": 0.2087,
      "step": 819
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3218475857650347e-06,
      "loss": 0.2057,
      "step": 820
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.301995994545241e-06,
      "loss": 0.2133,
      "step": 821
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.2822185883943404e-06,
      "loss": 0.2218,
      "step": 822
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.262515557905669e-06,
      "loss": 0.2146,
      "step": 823
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2428870929558012e-06,
      "loss": 0.2135,
      "step": 824
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.22333338270274e-06,
      "loss": 0.2073,
      "step": 825
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2038546155840733e-06,
      "loss": 0.219,
      "step": 826
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.184450979315177e-06,
      "loss": 0.2216,
      "step": 827
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.1651226608873876e-06,
      "loss": 0.2111,
      "step": 828
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.145869846566219e-06,
      "loss": 0.2244,
      "step": 829
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1266927218895507e-06,
      "loss": 0.2269,
      "step": 830
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.107591471665845e-06,
      "loss": 0.2156,
      "step": 831
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0885662799723807e-06,
      "loss": 0.2073,
      "step": 832
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.0696173301534494e-06,
      "loss": 0.2214,
      "step": 833
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0507448048186205e-06,
      "loss": 0.2292,
      "step": 834
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0319488858409552e-06,
      "loss": 0.2029,
      "step": 835
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.013229754355276e-06,
      "loss": 0.2171,
      "step": 836
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.994587590756397e-06,
      "loss": 0.2163,
      "step": 837
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9760225746974136e-06,
      "loss": 0.2237,
      "step": 838
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9575348850879437e-06,
      "loss": 0.227,
      "step": 839
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9391247000924227e-06,
      "loss": 0.2118,
      "step": 840
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9207921971283796e-06,
      "loss": 0.2069,
      "step": 841
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9025375528647294e-06,
      "loss": 0.206,
      "step": 842
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.8843609432200604e-06,
      "loss": 0.2034,
      "step": 843
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8662625433609582e-06,
      "loss": 0.1972,
      "step": 844
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.848242527700298e-06,
      "loss": 0.2144,
      "step": 845
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8303010698955803e-06,
      "loss": 0.2043,
      "step": 846
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8124383428472402e-06,
      "loss": 0.2075,
      "step": 847
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.7946545186970022e-06,
      "loss": 0.2197,
      "step": 848
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.7769497688261972e-06,
      "loss": 0.2163,
      "step": 849
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.7593242638541352e-06,
      "loss": 0.223,
      "step": 850
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7417781736364414e-06,
      "loss": 0.2125,
      "step": 851
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7243116672634264e-06,
      "loss": 0.2238,
      "step": 852
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7069249130584664e-06,
      "loss": 0.2214,
      "step": 853
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.6896180785763594e-06,
      "loss": 0.2046,
      "step": 854
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.672391330601737e-06,
      "loss": 0.2084,
      "step": 855
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.6552448351474305e-06,
      "loss": 0.2106,
      "step": 856
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.638178757452894e-06,
      "loss": 0.2126,
      "step": 857
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6211932619825932e-06,
      "loss": 0.2215,
      "step": 858
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.604288512424439e-06,
      "loss": 0.2153,
      "step": 859
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.587464671688187e-06,
      "loss": 0.2193,
      "step": 860
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.570721901903899e-06,
      "loss": 0.2265,
      "step": 861
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.5540603644203423e-06,
      "loss": 0.2063,
      "step": 862
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.5374802198034711e-06,
      "loss": 0.2167,
      "step": 863
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.520981627834851e-06,
      "loss": 0.2141,
      "step": 864
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5045647475101399e-06,
      "loss": 0.2169,
      "step": 865
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.4882297370375388e-06,
      "loss": 0.2203,
      "step": 866
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.471976753836285e-06,
      "loss": 0.2117,
      "step": 867
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.4558059545351144e-06,
      "loss": 0.2086,
      "step": 868
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4397174949707727e-06,
      "loss": 0.2202,
      "step": 869
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.423711530186498e-06,
      "loss": 0.2065,
      "step": 870
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4077882144305321e-06,
      "loss": 0.2225,
      "step": 871
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.3919477011546423e-06,
      "loss": 0.24,
      "step": 872
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.3761901430126223e-06,
      "loss": 0.218,
      "step": 873
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.360515691858847e-06,
      "loss": 0.2253,
      "step": 874
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.3449244987467836e-06,
      "loss": 0.2204,
      "step": 875
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3294167139275594e-06,
      "loss": 0.2053,
      "step": 876
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.31399248684849e-06,
      "loss": 0.2212,
      "step": 877
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.298651966151665e-06,
      "loss": 0.2187,
      "step": 878
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2833952996724864e-06,
      "loss": 0.2501,
      "step": 879
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2682226344382798e-06,
      "loss": 0.2517,
      "step": 880
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2531341166668364e-06,
      "loss": 0.2125,
      "step": 881
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2381298917650408e-06,
      "loss": 0.2199,
      "step": 882
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2232101043274437e-06,
      "loss": 0.226,
      "step": 883
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2083748981348832e-06,
      "loss": 0.2081,
      "step": 884
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.1936244161530918e-06,
      "loss": 0.2127,
      "step": 885
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1789588005313257e-06,
      "loss": 0.2091,
      "step": 886
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1643781926009846e-06,
      "loss": 0.2389,
      "step": 887
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1498827328742623e-06,
      "loss": 0.2209,
      "step": 888
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1354725610427807e-06,
      "loss": 0.2032,
      "step": 889
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.121147815976248e-06,
      "loss": 0.2353,
      "step": 890
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.106908635721129e-06,
      "loss": 0.2077,
      "step": 891
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.0927551574992967e-06,
      "loss": 0.225,
      "step": 892
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0786875177067313e-06,
      "loss": 0.2086,
      "step": 893
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0647058519121822e-06,
      "loss": 0.2377,
      "step": 894
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0508102948558862e-06,
      "loss": 0.23,
      "step": 895
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0370009804482483e-06,
      "loss": 0.2251,
      "step": 896
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.023278041768565e-06,
      "loss": 0.2487,
      "step": 897
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.009641611063732e-06,
      "loss": 0.2193,
      "step": 898
    },
    {
      "epoch": 2.58,
      "learning_rate": 9.960918197469771e-07,
      "loss": 0.2124,
      "step": 899
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.82628798396592e-07,
      "loss": 0.2287,
      "step": 900
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.692526767546727e-07,
      "loss": 0.2336,
      "step": 901
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.559635837258653e-07,
      "loss": 0.2357,
      "step": 902
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.427616473761347e-07,
      "loss": 0.2309,
      "step": 903
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.296469949315157e-07,
      "loss": 0.2221,
      "step": 904
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.166197527769016e-07,
      "loss": 0.2385,
      "step": 905
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.036800464548157e-07,
      "loss": 0.2267,
      "step": 906
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.90828000664209e-07,
      "loss": 0.2208,
      "step": 907
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.780637392592494e-07,
      "loss": 0.2129,
      "step": 908
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.653873852481365e-07,
      "loss": 0.2292,
      "step": 909
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.527990607919145e-07,
      "loss": 0.2297,
      "step": 910
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.402988872032902e-07,
      "loss": 0.2156,
      "step": 911
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.278869849454718e-07,
      "loss": 0.2256,
      "step": 912
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.15563473630997e-07,
      "loss": 0.1999,
      "step": 913
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.033284720205947e-07,
      "loss": 0.2302,
      "step": 914
    },
    {
      "epoch": 2.63,
      "learning_rate": 7.911820980220253e-07,
      "loss": 0.2083,
      "step": 915
    },
    {
      "epoch": 2.63,
      "learning_rate": 7.791244686889587e-07,
      "loss": 0.2378,
      "step": 916
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.671557002198316e-07,
      "loss": 0.2154,
      "step": 917
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.552759079567418e-07,
      "loss": 0.2194,
      "step": 918
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.434852063843278e-07,
      "loss": 0.2212,
      "step": 919
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.317837091286706e-07,
      "loss": 0.2287,
      "step": 920
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.201715289561883e-07,
      "loss": 0.2212,
      "step": 921
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.086487777725682e-07,
      "loss": 0.2151,
      "step": 922
    },
    {
      "epoch": 2.65,
      "learning_rate": 6.972155666216684e-07,
      "loss": 0.2114,
      "step": 923
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.858720056844614e-07,
      "loss": 0.2154,
      "step": 924
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.746182042779636e-07,
      "loss": 0.217,
      "step": 925
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.634542708541936e-07,
      "loss": 0.2316,
      "step": 926
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.52380312999108e-07,
      "loss": 0.2383,
      "step": 927
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.413964374315851e-07,
      "loss": 0.2381,
      "step": 928
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.305027500023841e-07,
      "loss": 0.2249,
      "step": 929
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.196993556931242e-07,
      "loss": 0.2086,
      "step": 930
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.089863586152856e-07,
      "loss": 0.2224,
      "step": 931
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.983638620091858e-07,
      "loss": 0.2155,
      "step": 932
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.878319682430089e-07,
      "loss": 0.2198,
      "step": 933
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.77390778811796e-07,
      "loss": 0.2197,
      "step": 934
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.670403943364832e-07,
      "loss": 0.2413,
      "step": 935
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.567809145629244e-07,
      "loss": 0.2414,
      "step": 936
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.466124383609317e-07,
      "loss": 0.2155,
      "step": 937
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.365350637233236e-07,
      "loss": 0.2312,
      "step": 938
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.265488877649816e-07,
      "loss": 0.2352,
      "step": 939
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.166540067219072e-07,
      "loss": 0.2103,
      "step": 940
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.068505159503079e-07,
      "loss": 0.2251,
      "step": 941
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.971385099256609e-07,
      "loss": 0.2244,
      "step": 942
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.875180822418191e-07,
      "loss": 0.2342,
      "step": 943
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.779893256100987e-07,
      "loss": 0.2148,
      "step": 944
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.6855233185839175e-07,
      "loss": 0.2106,
      "step": 945
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.592071919302743e-07,
      "loss": 0.2388,
      "step": 946
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.4995399588413925e-07,
      "loss": 0.2159,
      "step": 947
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.407928328923194e-07,
      "loss": 0.2423,
      "step": 948
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.3172379124023166e-07,
      "loss": 0.2154,
      "step": 949
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.227469583255328e-07,
      "loss": 0.2225,
      "step": 950
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.138624206572628e-07,
      "loss": 0.2248,
      "step": 951
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.0507026385502747e-07,
      "loss": 0.2226,
      "step": 952
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.9637057264815814e-07,
      "loss": 0.2229,
      "step": 953
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.877634308749079e-07,
      "loss": 0.2269,
      "step": 954
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.7924892148163417e-07,
      "loss": 0.217,
      "step": 955
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.708271265220087e-07,
      "loss": 0.224,
      "step": 956
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.6249812715621114e-07,
      "loss": 0.2455,
      "step": 957
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.5426200365017207e-07,
      "loss": 0.2323,
      "step": 958
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.461188353747702e-07,
      "loss": 0.2229,
      "step": 959
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.380687008050909e-07,
      "loss": 0.211,
      "step": 960
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.3011167751965776e-07,
      "loss": 0.2297,
      "step": 961
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.222478421996877e-07,
      "loss": 0.2338,
      "step": 962
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.1447727062835275e-07,
      "loss": 0.2365,
      "step": 963
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.0680003769005153e-07,
      "loss": 0.2212,
      "step": 964
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.992162173696822e-07,
      "loss": 0.2182,
      "step": 965
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.9172588275193536e-07,
      "loss": 0.2244,
      "step": 966
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.843291060205855e-07,
      "loss": 0.2279,
      "step": 967
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.770259584577972e-07,
      "loss": 0.2241,
      "step": 968
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.698165104434403e-07,
      "loss": 0.2408,
      "step": 969
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.627008314544044e-07,
      "loss": 0.2259,
      "step": 970
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.5567899006393893e-07,
      "loss": 0.2103,
      "step": 971
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.4875105394098654e-07,
      "loss": 0.2185,
      "step": 972
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.4191708984953044e-07,
      "loss": 0.2267,
      "step": 973
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.3517716364795384e-07,
      "loss": 0.2342,
      "step": 974
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.2853134028840594e-07,
      "loss": 0.22,
      "step": 975
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.2197968381616807e-07,
      "loss": 0.2236,
      "step": 976
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.1552225736905074e-07,
      "loss": 0.222,
      "step": 977
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.091591231767709e-07,
      "loss": 0.2399,
      "step": 978
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.0289034256036122e-07,
      "loss": 0.2133,
      "step": 979
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.967159759315751e-07,
      "loss": 0.2133,
      "step": 980
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.9063608279230593e-07,
      "loss": 0.2135,
      "step": 981
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.846507217340121e-07,
      "loss": 0.2203,
      "step": 982
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.7875995043715733e-07,
      "loss": 0.2207,
      "step": 983
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.7296382567064674e-07,
      "loss": 0.2307,
      "step": 984
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.6726240329128728e-07,
      "loss": 0.2152,
      "step": 985
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.6165573824324486e-07,
      "loss": 0.2336,
      "step": 986
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.5614388455751695e-07,
      "loss": 0.2196,
      "step": 987
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.5072689535141072e-07,
      "loss": 0.2119,
      "step": 988
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.4540482282803136e-07,
      "loss": 0.2244,
      "step": 989
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.401777182757813e-07,
      "loss": 0.2285,
      "step": 990
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.350456320678617e-07,
      "loss": 0.2219,
      "step": 991
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.3000861366179064e-07,
      "loss": 0.2168,
      "step": 992
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.2506671159892347e-07,
      "loss": 0.2291,
      "step": 993
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.2021997350399107e-07,
      "loss": 0.2153,
      "step": 994
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.1546844608462893e-07,
      "loss": 0.2334,
      "step": 995
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.1081217513094211e-07,
      "loss": 0.2358,
      "step": 996
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0625120551505219e-07,
      "loss": 0.2234,
      "step": 997
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0178558119067316e-07,
      "loss": 0.2335,
      "step": 998
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.741534519267737e-08,
      "loss": 0.2187,
      "step": 999
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.314053963669245e-08,
      "loss": 0.226,
      "step": 1000
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.896120571868839e-08,
      "loss": 0.2355,
      "step": 1001
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.487738371458443e-08,
      "loss": 0.2359,
      "step": 1002
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.088911297985502e-08,
      "loss": 0.2511,
      "step": 1003
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.699643194915784e-08,
      "loss": 0.2337,
      "step": 1004
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.319937813595857e-08,
      "loss": 0.2361,
      "step": 1005
    },
    {
      "epoch": 2.89,
      "learning_rate": 6.949798813217002e-08,
      "loss": 0.2417,
      "step": 1006
    },
    {
      "epoch": 2.89,
      "learning_rate": 6.589229760780358e-08,
      "loss": 0.2219,
      "step": 1007
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.238234131061616e-08,
      "loss": 0.2182,
      "step": 1008
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.8968153065788184e-08,
      "loss": 0.2158,
      "step": 1009
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.5649765775586115e-08,
      "loss": 0.2356,
      "step": 1010
    },
    {
      "epoch": 2.91,
      "learning_rate": 5.2427211419051605e-08,
      "loss": 0.2074,
      "step": 1011
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.930052105169281e-08,
      "loss": 0.2128,
      "step": 1012
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.626972480518133e-08,
      "loss": 0.2325,
      "step": 1013
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.333485188706577e-08,
      "loss": 0.2221,
      "step": 1014
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.049593058048973e-08,
      "loss": 0.2346,
      "step": 1015
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.7752988243919817e-08,
      "loss": 0.2442,
      "step": 1016
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.5106051310876963e-08,
      "loss": 0.2315,
      "step": 1017
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.255514528968884e-08,
      "loss": 0.2338,
      "step": 1018
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.010029476323895e-08,
      "loss": 0.2222,
      "step": 1019
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.774152338873126e-08,
      "loss": 0.2399,
      "step": 1020
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.547885389746485e-08,
      "loss": 0.2213,
      "step": 1021
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.3312308094607382e-08,
      "loss": 0.2173,
      "step": 1022
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.124190685899641e-08,
      "loss": 0.2249,
      "step": 1023
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.9267670142926187e-08,
      "loss": 0.2138,
      "step": 1024
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.7389616971964506e-08,
      "loss": 0.2235,
      "step": 1025
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.5607765444762834e-08,
      "loss": 0.2164,
      "step": 1026
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.3922132732888671e-08,
      "loss": 0.2431,
      "step": 1027
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.2332735080651248e-08,
      "loss": 0.2424,
      "step": 1028
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.0839587804954976e-08,
      "loss": 0.2249,
      "step": 1029
    },
    {
      "epoch": 2.96,
      "learning_rate": 9.442705295145127e-09,
      "loss": 0.2324,
      "step": 1030
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.142101012867943e-09,
      "loss": 0.2223,
      "step": 1031
    },
    {
      "epoch": 2.97,
      "learning_rate": 6.937787491947401e-09,
      "loss": 0.2437,
      "step": 1032
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.8297763382597625e-09,
      "loss": 0.2381,
      "step": 1033
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.818078229622547e-09,
      "loss": 0.2198,
      "step": 1034
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.902702915693501e-09,
      "loss": 0.2328,
      "step": 1035
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.083659217871793e-09,
      "loss": 0.2381,
      "step": 1036
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.3609550292180704e-09,
      "loss": 0.2351,
      "step": 1037
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.7345973143789717e-09,
      "loss": 0.2441,
      "step": 1038
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.2045921095127366e-09,
      "loss": 0.219,
      "step": 1039
    },
    {
      "epoch": 2.99,
      "learning_rate": 7.709445222403578e-10,
      "loss": 0.2414,
      "step": 1040
    },
    {
      "epoch": 2.99,
      "learning_rate": 4.3365873158895953e-10,
      "loss": 0.2268,
      "step": 1041
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.9273798795627074e-10,
      "loss": 0.2366,
      "step": 1042
    },
    {
      "epoch": 3.0,
      "learning_rate": 4.818461307731781e-11,
      "loss": 0.225,
      "step": 1043
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 0.1941,
      "step": 1044
    },
    {
      "epoch": 3.0,
      "step": 1044,
      "total_flos": 2.4979286595039396e+18,
      "train_loss": 0.11779194978918608,
      "train_runtime": 8266.3031,
      "train_samples_per_second": 16.147,
      "train_steps_per_second": 0.126
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1044,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 2.4979286595039396e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}