{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.9904,
  "eval_steps": 500,
  "global_step": 1872,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 3.5087719298245615e-06,
      "loss": 2.3172,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.017543859649123e-06,
      "loss": 2.5335,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.0526315789473684e-05,
      "loss": 2.3806,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.4035087719298246e-05,
      "loss": 2.3761,
      "step": 4
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.7543859649122806e-05,
      "loss": 2.3387,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.105263157894737e-05,
      "loss": 2.0888,
      "step": 6
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.456140350877193e-05,
      "loss": 2.4236,
      "step": 7
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.8070175438596492e-05,
      "loss": 2.3262,
      "step": 8
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.157894736842105e-05,
      "loss": 2.3452,
      "step": 9
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.508771929824561e-05,
      "loss": 2.2357,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.859649122807018e-05,
      "loss": 2.0336,
      "step": 11
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.210526315789474e-05,
      "loss": 2.1535,
      "step": 12
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.56140350877193e-05,
      "loss": 2.1177,
      "step": 13
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.912280701754386e-05,
      "loss": 2.2075,
      "step": 14
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.2631578947368424e-05,
      "loss": 2.2889,
      "step": 15
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.6140350877192984e-05,
      "loss": 2.1985,
      "step": 16
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.9649122807017544e-05,
      "loss": 2.0586,
      "step": 17
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.31578947368421e-05,
      "loss": 1.9725,
      "step": 18
    },
    {
      "epoch": 0.06,
      "learning_rate": 6.666666666666667e-05,
      "loss": 2.1229,
      "step": 19
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.017543859649122e-05,
      "loss": 2.0787,
      "step": 20
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.368421052631579e-05,
      "loss": 2.0661,
      "step": 21
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.719298245614036e-05,
      "loss": 2.1072,
      "step": 22
    },
    {
      "epoch": 0.07,
      "learning_rate": 8.070175438596491e-05,
      "loss": 1.9771,
      "step": 23
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.421052631578948e-05,
      "loss": 1.7706,
      "step": 24
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.771929824561403e-05,
      "loss": 1.9345,
      "step": 25
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.12280701754386e-05,
      "loss": 2.0117,
      "step": 26
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.473684210526316e-05,
      "loss": 2.0261,
      "step": 27
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.824561403508771e-05,
      "loss": 1.9335,
      "step": 28
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001017543859649123,
      "loss": 1.7864,
      "step": 29
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00010526315789473685,
      "loss": 1.9932,
      "step": 30
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00010877192982456141,
      "loss": 2.0425,
      "step": 31
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00011228070175438597,
      "loss": 1.8584,
      "step": 32
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00011578947368421053,
      "loss": 2.0044,
      "step": 33
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00011929824561403509,
      "loss": 1.9845,
      "step": 34
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00012280701754385965,
      "loss": 2.0119,
      "step": 35
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001263157894736842,
      "loss": 1.9309,
      "step": 36
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001298245614035088,
      "loss": 1.8312,
      "step": 37
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00013333333333333334,
      "loss": 1.7855,
      "step": 38
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001368421052631579,
      "loss": 1.8721,
      "step": 39
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00014035087719298245,
      "loss": 1.8671,
      "step": 40
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00014385964912280703,
      "loss": 1.7884,
      "step": 41
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00014736842105263158,
      "loss": 2.0239,
      "step": 42
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00015087719298245616,
      "loss": 1.96,
      "step": 43
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001543859649122807,
      "loss": 1.8249,
      "step": 44
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00015789473684210527,
      "loss": 1.818,
      "step": 45
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00016140350877192982,
      "loss": 1.8185,
      "step": 46
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001649122807017544,
      "loss": 1.6825,
      "step": 47
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00016842105263157895,
      "loss": 1.7722,
      "step": 48
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00017192982456140353,
      "loss": 1.7747,
      "step": 49
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00017543859649122806,
      "loss": 1.8761,
      "step": 50
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00017894736842105264,
      "loss": 1.7709,
      "step": 51
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001824561403508772,
      "loss": 1.8684,
      "step": 52
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018596491228070177,
      "loss": 1.7799,
      "step": 53
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00018947368421052632,
      "loss": 1.7442,
      "step": 54
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019298245614035088,
      "loss": 1.7151,
      "step": 55
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019649122807017543,
      "loss": 1.6625,
      "step": 56
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0002,
      "loss": 1.817,
      "step": 57
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019999985019842397,
      "loss": 1.7491,
      "step": 58
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019999940079414464,
      "loss": 1.9009,
      "step": 59
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019999865178850845,
      "loss": 1.7285,
      "step": 60
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001999976031837595,
      "loss": 1.7082,
      "step": 61
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019999625498303932,
      "loss": 1.6163,
      "step": 62
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019999460719038732,
      "loss": 1.8128,
      "step": 63
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001999926598107402,
      "loss": 1.5681,
      "step": 64
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019999041284993245,
      "loss": 1.7332,
      "step": 65
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019998786631469603,
      "loss": 1.9635,
      "step": 66
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001999850202126604,
      "loss": 1.7127,
      "step": 67
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001999818745523526,
      "loss": 1.5083,
      "step": 68
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001999784293431971,
      "loss": 1.7824,
      "step": 69
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001999746845955159,
      "loss": 1.6612,
      "step": 70
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019997064032052837,
      "loss": 1.8308,
      "step": 71
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019996629653035126,
      "loss": 1.6736,
      "step": 72
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001999616532379987,
      "loss": 1.6691,
      "step": 73
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001999567104573822,
      "loss": 1.9101,
      "step": 74
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001999514682033104,
      "loss": 1.7803,
      "step": 75
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00019994592649148933,
      "loss": 1.8175,
      "step": 76
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0001999400853385221,
      "loss": 1.7948,
      "step": 77
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.000199933944761909,
      "loss": 1.8029,
      "step": 78
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019992750478004738,
      "loss": 1.5079,
      "step": 79
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001999207654122316,
      "loss": 1.7893,
      "step": 80
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001999137266786531,
      "loss": 1.6126,
      "step": 81
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019990638860040006,
      "loss": 1.7386,
      "step": 82
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019989875119945768,
      "loss": 1.6334,
      "step": 83
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019989081449870778,
      "loss": 1.4597,
      "step": 84
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.000199882578521929,
      "loss": 1.5573,
      "step": 85
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019987404329379658,
      "loss": 1.7876,
      "step": 86
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019986520883988232,
      "loss": 1.795,
      "step": 87
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019985607518665456,
      "loss": 1.6929,
      "step": 88
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019984664236147794,
      "loss": 1.5556,
      "step": 89
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019983691039261357,
      "loss": 1.5542,
      "step": 90
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019982687930921873,
      "loss": 1.8002,
      "step": 91
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019981654914134686,
      "loss": 1.6953,
      "step": 92
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001998059199199474,
      "loss": 1.6642,
      "step": 93
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019979499167686591,
      "loss": 1.5977,
      "step": 94
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019978376444484375,
      "loss": 1.6627,
      "step": 95
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019977223825751801,
      "loss": 1.5319,
      "step": 96
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019976041314942155,
      "loss": 1.8042,
      "step": 97
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019974828915598277,
      "loss": 1.5658,
      "step": 98
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001997358663135255,
      "loss": 1.6145,
      "step": 99
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019972314465926903,
      "loss": 1.6792,
      "step": 100
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019971012423132775,
      "loss": 1.623,
      "step": 101
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019969680506871137,
      "loss": 1.6406,
      "step": 102
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001996831872113244,
      "loss": 1.5771,
      "step": 103
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019966927069996651,
      "loss": 1.7111,
      "step": 104
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001996550555763319,
      "loss": 1.5673,
      "step": 105
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019964054188300955,
      "loss": 1.4431,
      "step": 106
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.000199625729663483,
      "loss": 1.432,
      "step": 107
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019961061896213008,
      "loss": 1.5457,
      "step": 108
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001995952098242229,
      "loss": 1.7517,
      "step": 109
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019957950229592777,
      "loss": 1.6122,
      "step": 110
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019956349642430494,
      "loss": 1.6093,
      "step": 111
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019954719225730847,
      "loss": 1.8528,
      "step": 112
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019953058984378618,
      "loss": 1.4153,
      "step": 113
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019951368923347944,
      "loss": 1.712,
      "step": 114
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.000199496490477023,
      "loss": 1.5931,
      "step": 115
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019947899362594487,
      "loss": 1.7767,
      "step": 116
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019946119873266613,
      "loss": 1.7044,
      "step": 117
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001994431058505009,
      "loss": 1.6164,
      "step": 118
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019942471503365603,
      "loss": 1.6615,
      "step": 119
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019940602633723096,
      "loss": 1.5271,
      "step": 120
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001993870398172176,
      "loss": 1.6532,
      "step": 121
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001993677555305002,
      "loss": 1.4277,
      "step": 122
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019934817353485501,
      "loss": 1.6508,
      "step": 123
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019932829388895042,
      "loss": 1.5483,
      "step": 124
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001993081166523464,
      "loss": 1.7866,
      "step": 125
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019928764188549463,
      "loss": 1.5411,
      "step": 126
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019926686964973813,
      "loss": 1.7191,
      "step": 127
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019924580000731115,
      "loss": 1.717,
      "step": 128
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019922443302133904,
      "loss": 1.5859,
      "step": 129
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019920276875583798,
      "loss": 1.5405,
      "step": 130
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019918080727571473,
      "loss": 1.45,
      "step": 131
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019915854864676664,
      "loss": 1.5862,
      "step": 132
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0001991359929356812,
      "loss": 1.5191,
      "step": 133
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019911314021003613,
      "loss": 1.4357,
      "step": 134
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019908999053829882,
      "loss": 1.5612,
      "step": 135
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019906654398982646,
      "loss": 1.5456,
      "step": 136
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001990428006348656,
      "loss": 1.7929,
      "step": 137
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019901876054455217,
      "loss": 1.7371,
      "step": 138
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019899442379091097,
      "loss": 1.6345,
      "step": 139
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019896979044685571,
      "loss": 1.489,
      "step": 140
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019894486058618865,
      "loss": 1.6437,
      "step": 141
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019891963428360043,
      "loss": 1.4814,
      "step": 142
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019889411161466987,
      "loss": 1.673,
      "step": 143
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019886829265586368,
      "loss": 1.5479,
      "step": 144
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019884217748453623,
      "loss": 1.4067,
      "step": 145
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019881576617892952,
      "loss": 1.7148,
      "step": 146
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019878905881817252,
      "loss": 1.5538,
      "step": 147
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0001987620554822814,
      "loss": 1.465,
      "step": 148
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019873475625215898,
      "loss": 1.7048,
      "step": 149
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019870716120959462,
      "loss": 1.4952,
      "step": 150
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019867927043726392,
      "loss": 1.6577,
      "step": 151
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019865108401872857,
      "loss": 1.3972,
      "step": 152
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001986226020384359,
      "loss": 1.462,
      "step": 153
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019859382458171888,
      "loss": 1.5011,
      "step": 154
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019856475173479562,
      "loss": 1.5043,
      "step": 155
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019853538358476932,
      "loss": 1.6134,
      "step": 156
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019850572021962788,
      "loss": 1.5658,
      "step": 157
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019847576172824365,
      "loss": 1.6025,
      "step": 158
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019844550820037325,
      "loss": 1.8,
      "step": 159
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019841495972665717,
      "loss": 1.6006,
      "step": 160
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019838411639861966,
      "loss": 1.5348,
      "step": 161
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019835297830866826,
      "loss": 1.6836,
      "step": 162
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019832154555009365,
      "loss": 1.7392,
      "step": 163
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0001982898182170694,
      "loss": 1.5574,
      "step": 164
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019825779640465156,
      "loss": 1.4491,
      "step": 165
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019822548020877856,
      "loss": 1.6058,
      "step": 166
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019819286972627066,
      "loss": 1.6733,
      "step": 167
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019815996505482996,
      "loss": 1.742,
      "step": 168
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019812676629303987,
      "loss": 1.4995,
      "step": 169
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0001980932735403649,
      "loss": 1.6025,
      "step": 170
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00019805948689715041,
      "loss": 1.5137,
      "step": 171
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001980254064646223,
      "loss": 1.5378,
      "step": 172
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00019799103234488652,
      "loss": 1.6291,
      "step": 173
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00019795636464092908,
      "loss": 1.6155,
      "step": 174
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00019792140345661552,
      "loss": 1.6708,
      "step": 175
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0001978861488966906,
      "loss": 1.5125,
      "step": 176
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00019785060106677818,
      "loss": 1.5956,
      "step": 177
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00019781476007338058,
      "loss": 1.5528,
      "step": 178
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001977786260238786,
      "loss": 1.4287,
      "step": 179
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.000197742199026531,
      "loss": 1.6143,
      "step": 180
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019770547919047414,
      "loss": 1.4036,
      "step": 181
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019766846662572191,
      "loss": 1.4478,
      "step": 182
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00019763116144316505,
      "loss": 1.5474,
      "step": 183
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00019759356375457113,
      "loss": 1.4243,
      "step": 184
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0001975556736725839,
      "loss": 1.5014,
      "step": 185
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00019751749131072333,
      "loss": 1.5687,
      "step": 186
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00019747901678338496,
      "loss": 1.6393,
      "step": 187
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00019744025020583967,
      "loss": 1.5113,
      "step": 188
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00019740119169423337,
      "loss": 1.7124,
      "step": 189
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00019736184136558659,
      "loss": 1.6504,
      "step": 190
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0001973221993377941,
      "loss": 1.5775,
      "step": 191
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00019728226572962473,
      "loss": 1.6229,
      "step": 192
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00019724204066072084,
      "loss": 1.58,
      "step": 193
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00019720152425159793,
      "loss": 1.5288,
      "step": 194
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00019716071662364453,
      "loss": 1.6824,
      "step": 195
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00019711961789912148,
      "loss": 1.3646,
      "step": 196
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00019707822820116193,
      "loss": 1.5325,
      "step": 197
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0001970365476537707,
      "loss": 1.3974,
      "step": 198
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00019699457638182403,
      "loss": 1.8065,
      "step": 199
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00019695231451106912,
      "loss": 1.5716,
      "step": 200
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00019690976216812396,
      "loss": 1.6631,
      "step": 201
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00019686691948047664,
      "loss": 1.4881,
      "step": 202
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0001968237865764852,
      "loss": 1.432,
      "step": 203
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00019678036358537724,
      "loss": 1.6165,
      "step": 204
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00019673665063724938,
      "loss": 1.5807,
      "step": 205
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00019669264786306695,
      "loss": 1.5103,
      "step": 206
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001966483553946637,
      "loss": 1.6906,
      "step": 207
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019660377336474125,
      "loss": 1.8078,
      "step": 208
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019655890190686877,
      "loss": 1.4076,
      "step": 209
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019651374115548252,
      "loss": 1.8221,
      "step": 210
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0001964682912458856,
      "loss": 1.5819,
      "step": 211
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00019642255231424729,
      "loss": 1.7336,
      "step": 212
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.000196376524497603,
      "loss": 1.4339,
      "step": 213
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00019633020793385336,
      "loss": 1.4044,
      "step": 214
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019628360276176436,
      "loss": 1.4691,
      "step": 215
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019623670912096656,
      "loss": 1.4161,
      "step": 216
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019618952715195475,
      "loss": 1.4803,
      "step": 217
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00019614205699608763,
      "loss": 1.6953,
      "step": 218
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00019609429879558724,
      "loss": 1.5339,
      "step": 219
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0001960462526935387,
      "loss": 1.536,
      "step": 220
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00019599791883388962,
      "loss": 1.5036,
      "step": 221
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00019594929736144976,
      "loss": 1.5796,
      "step": 222
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0001959003884218906,
      "loss": 1.4014,
      "step": 223
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00019585119216174486,
      "loss": 1.765,
      "step": 224
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00019580170872840607,
      "loss": 1.6139,
      "step": 225
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0001957519382701282,
      "loss": 1.4663,
      "step": 226
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001957018809360251,
      "loss": 1.4978,
      "step": 227
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00019565153687607008,
      "loss": 1.6063,
      "step": 228
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00019560090624109555,
      "loss": 1.4683,
      "step": 229
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019554998918279253,
      "loss": 1.3838,
      "step": 230
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019549878585371007,
      "loss": 1.3349,
      "step": 231
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019544729640725498,
      "loss": 1.5005,
      "step": 232
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00019539552099769126,
      "loss": 1.4786,
      "step": 233
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001953434597801397,
      "loss": 1.39,
      "step": 234
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00019529111291057734,
      "loss": 1.6087,
      "step": 235
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.000195238480545837,
      "loss": 1.5579,
      "step": 236
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019518556284360696,
      "loss": 1.5945,
      "step": 237
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019513235996243032,
      "loss": 1.4493,
      "step": 238
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019507887206170454,
      "loss": 1.4569,
      "step": 239
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00019502509930168112,
      "loss": 1.5705,
      "step": 240
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00019497104184346492,
      "loss": 1.5838,
      "step": 241
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00019491669984901379,
      "loss": 1.5838,
      "step": 242
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.000194862073481138,
      "loss": 1.5343,
      "step": 243
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00019480716290349995,
      "loss": 1.3721,
      "step": 244
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00019475196828061346,
      "loss": 1.6957,
      "step": 245
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0001946964897778433,
      "loss": 1.6162,
      "step": 246
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00019464072756140486,
      "loss": 1.6806,
      "step": 247
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00019458468179836346,
      "loss": 1.5329,
      "step": 248
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00019452835265663403,
      "loss": 1.4677,
      "step": 249
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00019447174030498042,
      "loss": 1.6283,
      "step": 250
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00019441484491301503,
      "loss": 1.4402,
      "step": 251
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0001943576666511982,
      "loss": 1.3523,
      "step": 252
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00019430020569083788,
      "loss": 1.5063,
      "step": 253
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00019424246220408885,
      "loss": 1.3695,
      "step": 254
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00019418443636395248,
      "loss": 1.7036,
      "step": 255
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00019412612834427597,
      "loss": 1.5001,
      "step": 256
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00019406753831975203,
      "loss": 1.3078,
      "step": 257
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00019400866646591814,
      "loss": 1.4546,
      "step": 258
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00019394951295915628,
      "loss": 1.7531,
      "step": 259
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00019389007797669224,
      "loss": 1.5063,
      "step": 260
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019383036169659513,
      "loss": 1.539,
      "step": 261
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019377036429777672,
      "loss": 1.5303,
      "step": 262
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019371008595999112,
      "loss": 1.6219,
      "step": 263
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019364952686383417,
      "loss": 1.5351,
      "step": 264
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001935886871907428,
      "loss": 1.4384,
      "step": 265
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00019352756712299468,
      "loss": 1.3868,
      "step": 266
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001934661668437073,
      "loss": 1.5454,
      "step": 267
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.000193404486536838,
      "loss": 1.5081,
      "step": 268
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00019334252638718278,
      "loss": 1.3292,
      "step": 269
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00019328028658037626,
      "loss": 1.5167,
      "step": 270
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0001932177673028909,
      "loss": 1.6366,
      "step": 271
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0001931549687420364,
      "loss": 1.4058,
      "step": 272
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0001930918910859592,
      "loss": 1.6195,
      "step": 273
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00019302853452364196,
      "loss": 1.4995,
      "step": 274
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00019296489924490296,
      "loss": 1.4388,
      "step": 275
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00019290098544039546,
      "loss": 1.4997,
      "step": 276
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019283679330160726,
      "loss": 1.6313,
      "step": 277
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019277232302086005,
      "loss": 1.4472,
      "step": 278
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019270757479130878,
      "loss": 1.5243,
      "step": 279
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00019264254880694125,
      "loss": 1.4268,
      "step": 280
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00019257724526257727,
      "loss": 1.6627,
      "step": 281
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0001925116643538684,
      "loss": 1.5974,
      "step": 282
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00019244580627729703,
      "loss": 1.6206,
      "step": 283
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0001923796712301761,
      "loss": 1.6002,
      "step": 284
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00019231325941064832,
      "loss": 1.5034,
      "step": 285
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00019224657101768548,
      "loss": 1.6723,
      "step": 286
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001921796062510882,
      "loss": 1.556,
      "step": 287
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.000192112365311485,
      "loss": 1.5166,
      "step": 288
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001920448484003319,
      "loss": 1.8101,
      "step": 289
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00019197705571991162,
      "loss": 1.2978,
      "step": 290
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0001919089874733332,
      "loss": 1.5532,
      "step": 291
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00019184064386453128,
      "loss": 1.4579,
      "step": 292
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00019177202509826545,
      "loss": 1.4716,
      "step": 293
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00019170313138011964,
      "loss": 1.6448,
      "step": 294
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0001916339629165017,
      "loss": 1.5017,
      "step": 295
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00019156451991464245,
      "loss": 1.624,
      "step": 296
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00019149480258259533,
      "loss": 1.473,
      "step": 297
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00019142481112923568,
      "loss": 1.5146,
      "step": 298
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0001913545457642601,
      "loss": 1.389,
      "step": 299
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00019128400669818585,
      "loss": 1.484,
      "step": 300
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00019121319414235015,
      "loss": 1.512,
      "step": 301
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00019114210830890969,
      "loss": 1.5622,
      "step": 302
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00019107074941083983,
      "loss": 1.3792,
      "step": 303
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00019099911766193412,
      "loss": 1.4899,
      "step": 304
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00019092721327680355,
      "loss": 1.737,
      "step": 305
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00019085503647087585,
      "loss": 1.4976,
      "step": 306
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00019078258746039507,
      "loss": 1.6244,
      "step": 307
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00019070986646242073,
      "loss": 1.575,
      "step": 308
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0001906368736948272,
      "loss": 1.5931,
      "step": 309
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0001905636093763031,
      "loss": 1.5218,
      "step": 310
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00019049007372635068,
      "loss": 1.3608,
      "step": 311
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00019041626696528503,
      "loss": 1.3768,
      "step": 312
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00019034218931423356,
      "loss": 1.4464,
      "step": 313
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00019026784099513523,
      "loss": 1.5705,
      "step": 314
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00019019322223073995,
      "loss": 1.3149,
      "step": 315
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00019011833324460793,
      "loss": 1.4039,
      "step": 316
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0001900431742611089,
      "loss": 1.5071,
      "step": 317
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00018996774550542148,
      "loss": 1.5394,
      "step": 318
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00018989204720353274,
      "loss": 1.5935,
      "step": 319
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00018981607958223703,
      "loss": 1.6146,
      "step": 320
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00018973984286913584,
      "loss": 1.4072,
      "step": 321
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00018966333729263674,
      "loss": 1.4037,
      "step": 322
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00018958656308195283,
      "loss": 1.385,
      "step": 323
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00018950952046710207,
      "loss": 1.3117,
      "step": 324
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00018943220967890657,
      "loss": 1.5786,
      "step": 325
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00018935463094899192,
      "loss": 1.4182,
      "step": 326
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001892767845097864,
      "loss": 1.5718,
      "step": 327
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00018919867059452036,
      "loss": 1.3071,
      "step": 328
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00018912028943722563,
      "loss": 1.5695,
      "step": 329
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00018904164127273458,
      "loss": 1.3274,
      "step": 330
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00018896272633667962,
      "loss": 1.4773,
      "step": 331
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00018888354486549237,
      "loss": 1.567,
      "step": 332
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00018880409709640298,
      "loss": 1.4799,
      "step": 333
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00018872438326743955,
      "loss": 1.4917,
      "step": 334
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00018864440361742715,
      "loss": 1.3011,
      "step": 335
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00018856415838598736,
      "loss": 1.5324,
      "step": 336
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00018848364781353744,
      "loss": 1.4534,
      "step": 337
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00018840287214128956,
      "loss": 1.576,
      "step": 338
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00018832183161125024,
      "loss": 1.4228,
      "step": 339
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0001882405264662194,
      "loss": 1.5059,
      "step": 340
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00018815895694978987,
      "loss": 1.7199,
      "step": 341
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00018807712330634642,
      "loss": 1.5081,
      "step": 342
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00018799502578106534,
      "loss": 1.5493,
      "step": 343
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0001879126646199133,
      "loss": 1.3609,
      "step": 344
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00018783004006964698,
      "loss": 1.5984,
      "step": 345
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00018774715237781213,
      "loss": 1.4265,
      "step": 346
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00018766400179274286,
      "loss": 1.3539,
      "step": 347
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.000187580588563561,
      "loss": 1.5419,
      "step": 348
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00018749691294017517,
      "loss": 1.4221,
      "step": 349
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0001874129751732802,
      "loss": 1.5286,
      "step": 350
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00018732877551435627,
      "loss": 1.3725,
      "step": 351
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00018724431421566823,
      "loss": 1.4321,
      "step": 352
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00018715959153026477,
      "loss": 1.3532,
      "step": 353
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00018707460771197774,
      "loss": 1.494,
      "step": 354
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0001869893630154213,
      "loss": 1.6592,
      "step": 355
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00018690385769599133,
      "loss": 1.444,
      "step": 356
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0001868180920098644,
      "loss": 1.4647,
      "step": 357
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0001867320662139972,
      "loss": 1.4161,
      "step": 358
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00018664578056612577,
      "loss": 1.5119,
      "step": 359
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00018655923532476463,
      "loss": 1.4067,
      "step": 360
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00018647243074920602,
      "loss": 1.4104,
      "step": 361
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00018638536709951917,
      "loss": 1.4438,
      "step": 362
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00018629804463654955,
      "loss": 1.6135,
      "step": 363
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00018621046362191806,
      "loss": 1.4799,
      "step": 364
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00018612262431802007,
      "loss": 1.1761,
      "step": 365
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00018603452698802498,
      "loss": 1.5445,
      "step": 366
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00018594617189587512,
      "loss": 1.2799,
      "step": 367
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0001858575593062852,
      "loss": 1.4229,
      "step": 368
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00018576868948474127,
      "loss": 1.3834,
      "step": 369
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00018567956269750012,
      "loss": 1.4763,
      "step": 370
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00018559017921158848,
      "loss": 1.3702,
      "step": 371
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00018550053929480202,
      "loss": 1.4866,
      "step": 372
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00018541064321570477,
      "loss": 1.3384,
      "step": 373
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001853204912436282,
      "loss": 1.4708,
      "step": 374
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00018523008364867055,
      "loss": 1.5378,
      "step": 375
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001851394207016957,
      "loss": 1.3265,
      "step": 376
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001850485026743328,
      "loss": 1.6055,
      "step": 377
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00018495732983897503,
      "loss": 1.3656,
      "step": 378
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00018486590246877916,
      "loss": 1.5486,
      "step": 379
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00018477422083766444,
      "loss": 1.525,
      "step": 380
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00018468228522031195,
      "loss": 1.3502,
      "step": 381
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00018459009589216364,
      "loss": 1.4448,
      "step": 382
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001844976531294217,
      "loss": 1.4434,
      "step": 383
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00018440495720904756,
      "loss": 1.6182,
      "step": 384
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00018431200840876108,
      "loss": 1.4716,
      "step": 385
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00018421880700703982,
      "loss": 1.4804,
      "step": 386
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00018412535328311814,
      "loss": 1.4872,
      "step": 387
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00018403164751698626,
      "loss": 1.4339,
      "step": 388
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00018393768998938972,
      "loss": 1.5058,
      "step": 389
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00018384348098182815,
      "loss": 1.5038,
      "step": 390
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00018374902077655474,
      "loss": 1.4039,
      "step": 391
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00018365430965657526,
      "loss": 1.4857,
      "step": 392
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00018355934790564718,
      "loss": 1.5078,
      "step": 393
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0001834641358082789,
      "loss": 1.5121,
      "step": 394
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00018336867364972888,
      "loss": 1.6133,
      "step": 395
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00018327296171600471,
      "loss": 1.5047,
      "step": 396
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00018317700029386245,
      "loss": 1.4366,
      "step": 397
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00018308078967080546,
      "loss": 1.4306,
      "step": 398
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00018298433013508384,
      "loss": 1.4007,
      "step": 399
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00018288762197569343,
      "loss": 1.496,
      "step": 400
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00018279066548237486,
      "loss": 1.7517,
      "step": 401
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0001826934609456129,
      "loss": 1.4334,
      "step": 402
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00018259600865663534,
      "loss": 1.5764,
      "step": 403
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00018249830890741233,
      "loss": 1.4702,
      "step": 404
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00018240036199065546,
      "loss": 1.3872,
      "step": 405
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00018230216819981667,
      "loss": 1.3836,
      "step": 406
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00018220372782908777,
      "loss": 1.6156,
      "step": 407
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00018210504117339914,
      "loss": 1.5338,
      "step": 408
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00018200610852841913,
      "loss": 1.5066,
      "step": 409
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00018190693019055305,
      "loss": 1.4646,
      "step": 410
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00018180750645694236,
      "loss": 1.65,
      "step": 411
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00018170783762546365,
      "loss": 1.53,
      "step": 412
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00018160792399472792,
      "loss": 1.4526,
      "step": 413
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00018150776586407956,
      "loss": 1.5442,
      "step": 414
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00018140736353359547,
      "loss": 1.3152,
      "step": 415
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0001813067173040842,
      "loss": 1.3214,
      "step": 416
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00018120582747708502,
      "loss": 1.2655,
      "step": 417
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00018110469435486703,
      "loss": 1.4215,
      "step": 418
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00018100331824042826,
      "loss": 1.44,
      "step": 419
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00018090169943749476,
      "loss": 1.5909,
      "step": 420
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00018079983825051966,
      "loss": 1.5193,
      "step": 421
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00018069773498468223,
      "loss": 1.2997,
      "step": 422
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00018059538994588716,
      "loss": 1.2226,
      "step": 423
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00018049280344076336,
      "loss": 1.5868,
      "step": 424
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00018038997577666323,
      "loss": 1.3425,
      "step": 425
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00018028690726166173,
      "loss": 1.4769,
      "step": 426
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00018018359820455536,
      "loss": 1.4619,
      "step": 427
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0001800800489148613,
      "loss": 1.4564,
      "step": 428
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0001799762597028165,
      "loss": 1.4828,
      "step": 429
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0001798722308793767,
      "loss": 1.4694,
      "step": 430
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00017976796275621555,
      "loss": 1.4664,
      "step": 431
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0001796634556457236,
      "loss": 1.4943,
      "step": 432
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0001795587098610075,
      "loss": 1.4162,
      "step": 433
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0001794537257158889,
      "loss": 1.4718,
      "step": 434
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00017934850352490357,
      "loss": 1.3284,
      "step": 435
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00017924304360330057,
      "loss": 1.5258,
      "step": 436
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0001791373462670411,
      "loss": 1.4521,
      "step": 437
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00017903141183279778,
      "loss": 1.4795,
      "step": 438
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0001789252406179534,
      "loss": 1.5658,
      "step": 439
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00017881883294060038,
      "loss": 1.4198,
      "step": 440
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0001787121891195394,
      "loss": 1.4466,
      "step": 441
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00017860530947427875,
      "loss": 1.2728,
      "step": 442
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00017849819432503322,
      "loss": 1.3773,
      "step": 443
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00017839084399272315,
      "loss": 1.4465,
      "step": 444
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00017828325879897352,
      "loss": 1.4003,
      "step": 445
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00017817543906611296,
      "loss": 1.5592,
      "step": 446
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0001780673851171728,
      "loss": 1.5148,
      "step": 447
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0001779590972758861,
      "loss": 1.4684,
      "step": 448
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0001778505758666866,
      "loss": 1.526,
      "step": 449
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0001777418212147079,
      "loss": 1.3826,
      "step": 450
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00017763283364578234,
      "loss": 1.7297,
      "step": 451
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0001775236134864401,
      "loss": 1.4602,
      "step": 452
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00017741416106390826,
      "loss": 1.3728,
      "step": 453
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0001773044767061097,
      "loss": 1.5668,
      "step": 454
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0001771945607416622,
      "loss": 1.3894,
      "step": 455
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00017708441349987753,
      "loss": 1.4742,
      "step": 456
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0001769740353107602,
      "loss": 1.584,
      "step": 457
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00017686342650500683,
      "loss": 1.5898,
      "step": 458
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001767525874140048,
      "loss": 1.5895,
      "step": 459
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00017664151836983157,
      "loss": 1.3694,
      "step": 460
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00017653021970525355,
      "loss": 1.5367,
      "step": 461
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00017641869175372493,
      "loss": 1.3614,
      "step": 462
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00017630693484938705,
      "loss": 1.3826,
      "step": 463
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0001761949493270671,
      "loss": 1.4948,
      "step": 464
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0001760827355222772,
      "loss": 1.3946,
      "step": 465
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00017597029377121356,
      "loss": 1.4301,
      "step": 466
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00017585762441075503,
      "loss": 1.3465,
      "step": 467
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00017574472777846274,
      "loss": 1.5121,
      "step": 468
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00017563160421257845,
      "loss": 1.4873,
      "step": 469
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.000175518254052024,
      "loss": 1.5072,
      "step": 470
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00017540467763639994,
      "loss": 1.4668,
      "step": 471
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0001752908753059849,
      "loss": 1.419,
      "step": 472
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00017517684740173418,
      "loss": 1.2955,
      "step": 473
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00017506259426527902,
      "loss": 1.4077,
      "step": 474
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0001749481162389254,
      "loss": 1.5678,
      "step": 475
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0001748334136656531,
      "loss": 1.4427,
      "step": 476
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00017471848688911464,
      "loss": 1.3247,
      "step": 477
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00017460333625363427,
      "loss": 1.3435,
      "step": 478
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0001744879621042069,
      "loss": 1.558,
      "step": 479
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00017437236478649716,
      "loss": 1.3896,
      "step": 480
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00017425654464683823,
      "loss": 1.4133,
      "step": 481
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0001741405020322309,
      "loss": 1.5063,
      "step": 482
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0001740242372903425,
      "loss": 1.3532,
      "step": 483
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0001739077507695059,
      "loss": 1.5716,
      "step": 484
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0001737910428187183,
      "loss": 1.379,
      "step": 485
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0001736741137876405,
      "loss": 1.4828,
      "step": 486
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00017355696402659548,
      "loss": 1.6053,
      "step": 487
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0001734395938865677,
      "loss": 1.4128,
      "step": 488
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00017332200371920174,
      "loss": 1.7034,
      "step": 489
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00017320419387680143,
      "loss": 1.5378,
      "step": 490
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00017308616471232885,
      "loss": 1.3741,
      "step": 491
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.000172967916579403,
      "loss": 1.5632,
      "step": 492
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00017284944983229908,
      "loss": 1.4579,
      "step": 493
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0001727307648259472,
      "loss": 1.2098,
      "step": 494
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00017261186191593135,
      "loss": 1.3943,
      "step": 495
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0001724927414584884,
      "loss": 1.5527,
      "step": 496
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00017237340381050703,
      "loss": 1.3841,
      "step": 497
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00017225384932952656,
      "loss": 1.3339,
      "step": 498
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00017213407837373597,
      "loss": 1.5513,
      "step": 499
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00017201409130197286,
      "loss": 1.2548,
      "step": 500
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00017189388847372225,
      "loss": 1.2755,
      "step": 501
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00017177347024911562,
      "loss": 1.485,
      "step": 502
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00017165283698892975,
      "loss": 1.3735,
      "step": 503
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00017153198905458573,
      "loss": 1.567,
      "step": 504
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0001714109268081477,
      "loss": 1.5942,
      "step": 505
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00017128965061232205,
      "loss": 1.3594,
      "step": 506
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00017116816083045602,
      "loss": 1.5214,
      "step": 507
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0001710464578265369,
      "loss": 1.3841,
      "step": 508
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00017092454196519064,
      "loss": 1.3861,
      "step": 509
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00017080241361168107,
      "loss": 1.3898,
      "step": 510
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00017068007313190858,
      "loss": 1.5725,
      "step": 511
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00017055752089240907,
      "loss": 1.3655,
      "step": 512
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00017043475726035288,
      "loss": 1.3229,
      "step": 513
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00017031178260354385,
      "loss": 1.5203,
      "step": 514
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.00017018859729041783,
      "loss": 1.4181,
      "step": 515
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.00017006520169004187,
      "loss": 1.3891,
      "step": 516
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.00016994159617211317,
      "loss": 1.3101,
      "step": 517
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00016981778110695774,
      "loss": 1.4156,
      "step": 518
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00016969375686552937,
      "loss": 1.4226,
      "step": 519
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00016956952381940861,
      "loss": 1.3968,
      "step": 520
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00016944508234080158,
      "loss": 1.3244,
      "step": 521
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0001693204328025389,
      "loss": 1.5186,
      "step": 522
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00016919557557807454,
      "loss": 1.3186,
      "step": 523
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00016907051104148458,
      "loss": 1.441,
      "step": 524
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00016894523956746639,
      "loss": 1.5353,
      "step": 525
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00016881976153133724,
      "loss": 1.2521,
      "step": 526
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0001686940773090333,
      "loss": 1.4658,
      "step": 527
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00016856818727710847,
      "loss": 1.3583,
      "step": 528
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00016844209181273318,
      "loss": 1.2828,
      "step": 529
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00016831579129369346,
      "loss": 1.459,
      "step": 530
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00016818928609838967,
      "loss": 1.5672,
      "step": 531
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00016806257660583534,
      "loss": 1.4211,
      "step": 532
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.00016793566319565608,
      "loss": 1.4859,
      "step": 533
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0001678085462480885,
      "loss": 1.4854,
      "step": 534
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.000167681226143979,
      "loss": 1.4347,
      "step": 535
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00016755370326478256,
      "loss": 1.3787,
      "step": 536
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00016742597799256182,
      "loss": 1.3458,
      "step": 537
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00016729805070998569,
      "loss": 1.4266,
      "step": 538
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00016716992180032831,
      "loss": 1.3014,
      "step": 539
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00016704159164746796,
      "loss": 1.4763,
      "step": 540
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00016691306063588583,
      "loss": 1.4189,
      "step": 541
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00016678432915066488,
      "loss": 1.5143,
      "step": 542
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.00016665539757748868,
      "loss": 1.5072,
      "step": 543
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.00016652626630264032,
      "loss": 1.5244,
      "step": 544
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.00016639693571300114,
      "loss": 1.3944,
      "step": 545
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00016626740619604967,
      "loss": 1.3387,
      "step": 546
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00016613767813986044,
      "loss": 1.4616,
      "step": 547
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00016600775193310275,
      "loss": 1.3723,
      "step": 548
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.00016587762796503968,
      "loss": 1.3278,
      "step": 549
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0001657473066255267,
      "loss": 1.4045,
      "step": 550
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.00016561678830501065,
      "loss": 1.3904,
      "step": 551
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.00016548607339452853,
      "loss": 1.3531,
      "step": 552
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0001653551622857063,
      "loss": 1.5287,
      "step": 553
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0001652240553707579,
      "loss": 1.4645,
      "step": 554
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00016509275304248363,
      "loss": 1.4232,
      "step": 555
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00016496125569426945,
      "loss": 1.4697,
      "step": 556
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0001648295637200856,
      "loss": 1.427,
      "step": 557
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00016469767751448538,
      "loss": 1.626,
      "step": 558
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00016456559747260395,
      "loss": 1.4028,
      "step": 559
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00016443332399015734,
      "loss": 1.3729,
      "step": 560
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00016430085746344108,
      "loss": 1.4234,
      "step": 561
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.000164168198289329,
      "loss": 1.5274,
      "step": 562
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00016403534686527225,
      "loss": 1.5213,
      "step": 563
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0001639023035892978,
      "loss": 1.4918,
      "step": 564
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0001637690688600076,
      "loss": 1.3838,
      "step": 565
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00016363564307657696,
      "loss": 1.3532,
      "step": 566
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00016350202663875386,
      "loss": 1.4937,
      "step": 567
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00016336821994685728,
      "loss": 1.4443,
      "step": 568
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00016323422340177636,
      "loss": 1.4337,
      "step": 569
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00016310003740496886,
      "loss": 1.3915,
      "step": 570
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0001629656623584604,
      "loss": 1.4657,
      "step": 571
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0001628310986648427,
      "loss": 1.4089,
      "step": 572
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00016269634672727294,
      "loss": 1.4674,
      "step": 573
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00016256140694947217,
      "loss": 1.476,
      "step": 574
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00016242627973572416,
      "loss": 1.5458,
      "step": 575
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00016229096549087434,
      "loss": 1.3093,
      "step": 576
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0001621554646203284,
      "loss": 1.2719,
      "step": 577
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0001620197775300513,
      "loss": 1.5138,
      "step": 578
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0001618839046265658,
      "loss": 1.4086,
      "step": 579
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0001617478463169514,
      "loss": 1.4949,
      "step": 580
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0001616116030088431,
      "loss": 1.45,
      "step": 581
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0001614751751104301,
      "loss": 1.6165,
      "step": 582
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00016133856303045474,
      "loss": 1.5151,
      "step": 583
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00016120176717821107,
      "loss": 1.4335,
      "step": 584
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00016106478796354382,
      "loss": 1.386,
      "step": 585
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.000160927625796847,
      "loss": 1.4458,
      "step": 586
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00016079028108906282,
      "loss": 1.3679,
      "step": 587
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00016065275425168032,
      "loss": 1.4089,
      "step": 588
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00016051504569673426,
      "loss": 1.5585,
      "step": 589
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.00016037715583680376,
      "loss": 1.487,
      "step": 590
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.00016023908508501128,
      "loss": 1.4683,
      "step": 591
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0001601008338550211,
      "loss": 1.3903,
      "step": 592
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00015996240256103823,
      "loss": 1.483,
      "step": 593
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00015982379161780724,
      "loss": 1.3927,
      "step": 594
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00015968500144061085,
      "loss": 1.493,
      "step": 595
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0001595460324452688,
      "loss": 1.4701,
      "step": 596
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00015940688504813662,
      "loss": 1.3127,
      "step": 597
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00015926755966610428,
      "loss": 1.3694,
      "step": 598
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00015912805671659501,
      "loss": 1.4446,
      "step": 599
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00015898837661756406,
      "loss": 1.6506,
      "step": 600
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00015884851978749737,
      "loss": 1.3577,
      "step": 601
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.00015870848664541044,
      "loss": 1.3611,
      "step": 602
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.00015856827761084698,
      "loss": 1.4103,
      "step": 603
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.00015842789310387768,
      "loss": 1.3667,
      "step": 604
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00015828733354509894,
      "loss": 1.3321,
      "step": 605
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00015814659935563163,
      "loss": 1.2261,
      "step": 606
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00015800569095711982,
      "loss": 1.5144,
      "step": 607
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00015786460877172952,
      "loss": 1.2642,
      "step": 608
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00015772335322214738,
      "loss": 1.3354,
      "step": 609
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00015758192473157949,
      "loss": 1.394,
      "step": 610
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00015744032372375007,
      "loss": 1.4857,
      "step": 611
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00015729855062290022,
      "loss": 1.4895,
      "step": 612
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00015715660585378658,
      "loss": 1.4756,
      "step": 613
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00015701448984168018,
      "loss": 1.5323,
      "step": 614
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0001568722030123651,
      "loss": 1.5812,
      "step": 615
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00015672974579213707,
      "loss": 1.386,
      "step": 616
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0001565871186078025,
      "loss": 1.4958,
      "step": 617
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.00015644432188667695,
      "loss": 1.4814,
      "step": 618
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.00015630135605658381,
      "loss": 1.5019,
      "step": 619
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.00015615822154585324,
      "loss": 1.2698,
      "step": 620
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00015601491878332077,
      "loss": 1.3101,
      "step": 621
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.000155871448198326,
      "loss": 1.5265,
      "step": 622
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00015572781022071128,
      "loss": 1.6532,
      "step": 623
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00015558400528082057,
      "loss": 1.2782,
      "step": 624
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.000155440033809498,
      "loss": 1.2672,
      "step": 625
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00015529589623808658,
      "loss": 1.3212,
      "step": 626
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.00015515159299842707,
      "loss": 1.386,
      "step": 627
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.00015500712452285652,
      "loss": 1.1956,
      "step": 628
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.000154862491244207,
      "loss": 1.3593,
      "step": 629
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0001547176935958044,
      "loss": 1.2638,
      "step": 630
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.00015457273201146708,
      "loss": 1.4192,
      "step": 631
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.00015442760692550443,
      "loss": 1.3326,
      "step": 632
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.00015428231877271582,
      "loss": 1.3679,
      "step": 633
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.00015413686798838912,
      "loss": 1.5289,
      "step": 634
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.00015399125500829955,
      "loss": 1.4695,
      "step": 635
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.00015384548026870805,
      "loss": 1.4593,
      "step": 636
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.00015369954420636048,
      "loss": 1.3698,
      "step": 637
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.00015355344725848577,
      "loss": 1.2768,
      "step": 638
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.00015340718986279502,
      "loss": 1.4136,
      "step": 639
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00015326077245747999,
      "loss": 1.5304,
      "step": 640
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00015311419548121188,
      "loss": 1.4138,
      "step": 641
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00015296745937313987,
      "loss": 1.2821,
      "step": 642
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00015282056457289,
      "loss": 1.3221,
      "step": 643
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00015267351152056373,
      "loss": 1.3617,
      "step": 644
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00015252630065673662,
      "loss": 1.3652,
      "step": 645
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00015237893242245706,
      "loss": 1.3539,
      "step": 646
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00015223140725924495,
      "loss": 1.3032,
      "step": 647
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0001520837256090903,
      "loss": 1.2945,
      "step": 648
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.000151935887914452,
      "loss": 1.1101,
      "step": 649
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.00015178789461825644,
      "loss": 1.306,
      "step": 650
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0001516397461638962,
      "loss": 1.165,
      "step": 651
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.00015149144299522873,
      "loss": 1.5056,
      "step": 652
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.000151342985556575,
      "loss": 1.4979,
      "step": 653
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.00015119437429271813,
      "loss": 1.3254,
      "step": 654
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00015104560964890223,
      "loss": 1.3995,
      "step": 655
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00015089669207083074,
      "loss": 1.3916,
      "step": 656
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00015074762200466556,
      "loss": 1.293,
      "step": 657
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00015059839989702524,
      "loss": 1.26,
      "step": 658
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00015044902619498394,
      "loss": 1.5445,
      "step": 659
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00015029950134606992,
      "loss": 1.2491,
      "step": 660
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.0001501498257982644,
      "loss": 1.3863,
      "step": 661
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00015000000000000001,
      "loss": 1.4273,
      "step": 662
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00014985002440015958,
      "loss": 1.5397,
      "step": 663
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00014969989944807471,
      "loss": 1.4518,
      "step": 664
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00014954962559352456,
      "loss": 1.2563,
      "step": 665
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00014939920328673422,
      "loss": 1.4494,
      "step": 666
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00014924863297837378,
      "loss": 1.3395,
      "step": 667
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00014909791511955655,
      "loss": 1.2413,
      "step": 668
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00014894705016183803,
      "loss": 1.4431,
      "step": 669
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00014879603855721434,
      "loss": 1.2565,
      "step": 670
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.00014864488075812105,
      "loss": 1.3512,
      "step": 671
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.00014849357721743168,
      "loss": 1.288,
      "step": 672
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.00014834212838845637,
      "loss": 1.3445,
      "step": 673
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00014819053472494068,
      "loss": 1.4874,
      "step": 674
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00014803879668106394,
      "loss": 1.3243,
      "step": 675
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00014788691471143806,
      "loss": 1.2465,
      "step": 676
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.00014773488927110633,
      "loss": 1.338,
      "step": 677
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.00014758272081554167,
      "loss": 1.2973,
      "step": 678
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.0001474304098006456,
      "loss": 1.4739,
      "step": 679
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.0001472779566827467,
      "loss": 1.4921,
      "step": 680
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.00014712536191859932,
      "loss": 1.4539,
      "step": 681
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.00014697262596538227,
      "loss": 1.4138,
      "step": 682
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.00014681974928069713,
      "loss": 1.3871,
      "step": 683
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.00014666673232256738,
      "loss": 1.6175,
      "step": 684
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0001465135755494366,
      "loss": 1.3067,
      "step": 685
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00014636027942016734,
      "loss": 1.4646,
      "step": 686
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00014620684439403962,
      "loss": 1.4539,
      "step": 687
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00014605327093074962,
      "loss": 1.3881,
      "step": 688
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00014589955949040822,
      "loss": 1.3616,
      "step": 689
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.00014574571053353988,
      "loss": 1.3358,
      "step": 690
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.00014559172452108075,
      "loss": 1.2683,
      "step": 691
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0001454376019143779,
      "loss": 1.4341,
      "step": 692
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00014528334317518747,
      "loss": 1.3556,
      "step": 693
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00014512894876567354,
      "loss": 1.4937,
      "step": 694
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0001449744191484066,
      "loss": 1.2967,
      "step": 695
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0001448197547863622,
      "loss": 1.3828,
      "step": 696
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.00014466495614291977,
      "loss": 1.3779,
      "step": 697
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.00014451002368186084,
      "loss": 1.3222,
      "step": 698
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00014435495786736794,
      "loss": 1.4921,
      "step": 699
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00014419975916402317,
      "loss": 1.4153,
      "step": 700
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00014404442803680673,
      "loss": 1.1315,
      "step": 701
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.0001438889649510956,
      "loss": 1.4833,
      "step": 702
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00014373337037266198,
      "loss": 1.3764,
      "step": 703
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.0001435776447676723,
      "loss": 1.4428,
      "step": 704
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.00014342178860268524,
      "loss": 1.3092,
      "step": 705
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.00014326580234465085,
      "loss": 1.56,
      "step": 706
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.00014310968646090883,
      "loss": 1.445,
      "step": 707
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.00014295344141918733,
      "loss": 1.393,
      "step": 708
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0001427970676876014,
      "loss": 1.2979,
      "step": 709
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0001426405657346517,
      "loss": 1.475,
      "step": 710
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.000142483936029223,
      "loss": 1.3098,
      "step": 711
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0001423271790405828,
      "loss": 1.243,
      "step": 712
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.00014217029523837997,
      "loss": 1.335,
      "step": 713
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0001420132850926434,
      "loss": 1.3024,
      "step": 714
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0001418561490737804,
      "loss": 1.4524,
      "step": 715
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0001416988876525754,
      "loss": 1.2081,
      "step": 716
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.00014154150130018866,
      "loss": 1.2101,
      "step": 717
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.00014138399048815457,
      "loss": 1.4265,
      "step": 718
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.00014122635568838052,
      "loss": 1.3994,
      "step": 719
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0001410685973731453,
      "loss": 1.4248,
      "step": 720
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.00014091071601509788,
      "loss": 1.4572,
      "step": 721
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0001407527120872557,
      "loss": 1.4662,
      "step": 722
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.00014059458606300356,
      "loss": 1.4821,
      "step": 723
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.000140436338416092,
      "loss": 1.3322,
      "step": 724
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.00014027796962063596,
      "loss": 1.3303,
      "step": 725
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.00014011948015111333,
      "loss": 1.3796,
      "step": 726
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.00013996087048236358,
      "loss": 1.3012,
      "step": 727
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.00013980214108958624,
      "loss": 1.4473,
      "step": 728
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0001396432924483396,
      "loss": 1.3776,
      "step": 729
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.00013948432503453918,
      "loss": 1.2369,
      "step": 730
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.0001393252393244564,
      "loss": 1.4466,
      "step": 731
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.00013916603579471705,
      "loss": 1.3428,
      "step": 732
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00013900671492229986,
      "loss": 1.5522,
      "step": 733
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00013884727718453527,
      "loss": 1.3796,
      "step": 734
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00013868772305910377,
      "loss": 1.4047,
      "step": 735
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00013852805302403445,
      "loss": 1.4663,
      "step": 736
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00013836826755770384,
      "loss": 1.2887,
      "step": 737
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00013820836713883422,
      "loss": 1.3056,
      "step": 738
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0001380483522464923,
      "loss": 1.358,
      "step": 739
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.00013788822336008774,
      "loss": 1.2406,
      "step": 740
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0001377279809593717,
      "loss": 1.4731,
      "step": 741
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.00013756762552443553,
      "loss": 1.2764,
      "step": 742
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.0001374071575357091,
      "loss": 1.3851,
      "step": 743
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00013724657747395957,
      "loss": 1.4557,
      "step": 744
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00013708588582028993,
      "loss": 1.3181,
      "step": 745
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0001369250830561373,
      "loss": 1.3265,
      "step": 746
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.000136764169663272,
      "loss": 1.3695,
      "step": 747
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.00013660314612379555,
      "loss": 1.3988,
      "step": 748
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00013644201292013953,
      "loss": 1.222,
      "step": 749
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.0001362807705350641,
      "loss": 1.3944,
      "step": 750
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00013611941945165657,
      "loss": 1.4731,
      "step": 751
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.00013595796015332984,
      "loss": 1.4761,
      "step": 752
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.00013579639312382105,
      "loss": 1.3754,
      "step": 753
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.00013563471884719012,
      "loss": 1.3759,
      "step": 754
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.0001354729378078183,
      "loss": 1.5125,
      "step": 755
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00013531105049040666,
      "loss": 1.2853,
      "step": 756
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00013514905737997473,
      "loss": 1.4425,
      "step": 757
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.00013498695896185895,
      "loss": 1.2394,
      "step": 758
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0001348247557217113,
      "loss": 1.2201,
      "step": 759
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0001346624481454978,
      "loss": 1.2557,
      "step": 760
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.00013450003671949706,
      "loss": 1.413,
      "step": 761
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.00013433752193029886,
      "loss": 1.5174,
      "step": 762
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.00013417490426480258,
      "loss": 1.2966,
      "step": 763
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0001340121842102159,
      "loss": 1.4125,
      "step": 764
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00013384936225405326,
      "loss": 1.533,
      "step": 765
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.0001336864388841343,
      "loss": 1.3615,
      "step": 766
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00013352341458858265,
      "loss": 1.2005,
      "step": 767
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0001333602898558242,
      "loss": 1.5058,
      "step": 768
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0001331970651745858,
      "loss": 1.3939,
      "step": 769
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0001330337410338937,
      "loss": 1.3842,
      "step": 770
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.00013287031792307225,
      "loss": 1.2204,
      "step": 771
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.00013270679633174218,
      "loss": 1.3189,
      "step": 772
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.00013254317674981936,
      "loss": 1.3576,
      "step": 773
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0001323794596675132,
      "loss": 1.3569,
      "step": 774
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.00013221564557532527,
      "loss": 1.4423,
      "step": 775
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.00013205173496404768,
      "loss": 1.2972,
      "step": 776
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00013188772832476188,
      "loss": 1.3156,
      "step": 777
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00013172362614883687,
      "loss": 1.3911,
      "step": 778
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00013155942892792796,
      "loss": 1.474,
      "step": 779
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.00013139513715397521,
      "loss": 1.4454,
      "step": 780
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.00013123075131920195,
      "loss": 1.2392,
      "step": 781
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.00013106627191611332,
      "loss": 1.1867,
      "step": 782
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00013090169943749476,
      "loss": 1.5269,
      "step": 783
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00013073703437641065,
      "loss": 1.3705,
      "step": 784
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0001305722772262027,
      "loss": 1.3898,
      "step": 785
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0001304074284804885,
      "loss": 1.3475,
      "step": 786
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.00013024248863316012,
      "loss": 1.4471,
      "step": 787
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0001300774581783825,
      "loss": 1.1908,
      "step": 788
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.00012991233761059214,
      "loss": 1.4066,
      "step": 789
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.00012974712742449542,
      "loss": 1.4146,
      "step": 790
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.00012958182811506727,
      "loss": 1.3315,
      "step": 791
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.00012941644017754964,
      "loss": 1.2937,
      "step": 792
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00012925096410745,
      "loss": 1.3984,
      "step": 793
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.0001290854004005399,
      "loss": 1.4314,
      "step": 794
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.0001289197495528534,
      "loss": 1.3854,
      "step": 795
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.00012875401206068567,
      "loss": 1.4179,
      "step": 796
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.00012858818842059145,
      "loss": 1.4297,
      "step": 797
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.00012842227912938359,
      "loss": 1.3907,
      "step": 798
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.00012825628468413156,
      "loss": 1.1615,
      "step": 799
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.0001280902055821599,
      "loss": 1.5687,
      "step": 800
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.00012792404232104697,
      "loss": 1.476,
      "step": 801
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00012775779539862304,
      "loss": 1.5659,
      "step": 802
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.0001275914653129691,
      "loss": 1.2925,
      "step": 803
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00012742505256241543,
      "loss": 1.4044,
      "step": 804
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.0001272585576455398,
      "loss": 1.4869,
      "step": 805
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00012709198106116624,
      "loss": 1.4272,
      "step": 806
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00012692532330836346,
      "loss": 1.336,
      "step": 807
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.00012675858488644335,
      "loss": 1.4513,
      "step": 808
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.00012659176629495946,
      "loss": 1.423,
      "step": 809
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.00012642486803370552,
      "loss": 1.5103,
      "step": 810
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00012625789060271403,
      "loss": 1.2999,
      "step": 811
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.0001260908345022547,
      "loss": 1.2637,
      "step": 812
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.0001259237002328327,
      "loss": 1.2512,
      "step": 813
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00012575648829518768,
      "loss": 1.4709,
      "step": 814
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.00012558919919029187,
      "loss": 1.3393,
      "step": 815
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.00012542183341934872,
      "loss": 1.1899,
      "step": 816
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.00012525439148379128,
      "loss": 1.344,
      "step": 817
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.00012508687388528093,
      "loss": 1.5634,
      "step": 818
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.00012491928112570567,
      "loss": 1.2246,
      "step": 819
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.00012475161370717868,
      "loss": 1.3249,
      "step": 820
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0001245838721320368,
      "loss": 1.2384,
      "step": 821
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.00012441605690283915,
      "loss": 1.3659,
      "step": 822
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0001242481685223654,
      "loss": 1.4046,
      "step": 823
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0001240802074936144,
      "loss": 1.5047,
      "step": 824
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.00012391217431980274,
      "loss": 1.2689,
      "step": 825
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.00012374406950436308,
      "loss": 1.4129,
      "step": 826
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00012357589355094275,
      "loss": 1.3343,
      "step": 827
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.0001234076469634022,
      "loss": 1.3085,
      "step": 828
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00012323933024581353,
      "loss": 1.3645,
      "step": 829
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0001230709439024589,
      "loss": 1.3138,
      "step": 830
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00012290248843782915,
      "loss": 1.3685,
      "step": 831
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00012273396435662212,
      "loss": 1.2769,
      "step": 832
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00012256537216374127,
      "loss": 1.1786,
      "step": 833
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00012239671236429414,
      "loss": 1.4945,
      "step": 834
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0001222279854635908,
      "loss": 1.3742,
      "step": 835
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00012205919196714237,
      "loss": 1.2341,
      "step": 836
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0001218903323806595,
      "loss": 1.518,
      "step": 837
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00012172140721005079,
      "loss": 1.3742,
      "step": 838
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00012155241696142141,
      "loss": 1.4558,
      "step": 839
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00012138336214107147,
      "loss": 1.3801,
      "step": 840
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00012121424325549449,
      "loss": 1.3469,
      "step": 841
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00012104506081137608,
      "loss": 1.3006,
      "step": 842
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.00012087581531559207,
      "loss": 1.3248,
      "step": 843
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.00012070650727520739,
      "loss": 1.3545,
      "step": 844
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.00012053713719747421,
      "loss": 1.3684,
      "step": 845
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.00012036770558983066,
      "loss": 1.4559,
      "step": 846
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.00012019821295989912,
      "loss": 1.5158,
      "step": 847
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0001200286598154849,
      "loss": 1.549,
      "step": 848
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.00011985904666457455,
      "loss": 1.2669,
      "step": 849
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.00011968937401533442,
      "loss": 1.5132,
      "step": 850
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.00011951964237610911,
      "loss": 1.3185,
      "step": 851
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00011934985225541998,
      "loss": 1.4872,
      "step": 852
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00011918000416196353,
      "loss": 1.2543,
      "step": 853
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00011901009860461007,
      "loss": 1.255,
      "step": 854
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00011884013609240199,
      "loss": 1.3011,
      "step": 855
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00011867011713455229,
      "loss": 1.2608,
      "step": 856
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00011850004224044315,
      "loss": 1.5666,
      "step": 857
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00011832991191962435,
      "loss": 1.2798,
      "step": 858
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00011815972668181165,
      "loss": 1.5307,
      "step": 859
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00011798948703688539,
      "loss": 1.3173,
      "step": 860
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00011781919349488895,
      "loss": 1.3589,
      "step": 861
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.0001176488465660271,
      "loss": 1.3508,
      "step": 862
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00011747844676066461,
      "loss": 1.4931,
      "step": 863
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00011730799458932474,
      "loss": 1.5375,
      "step": 864
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.00011713749056268748,
      "loss": 1.4085,
      "step": 865
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.00011696693519158832,
      "loss": 1.5244,
      "step": 866
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.00011679632898701649,
      "loss": 1.3602,
      "step": 867
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.0001166256724601136,
      "loss": 1.3171,
      "step": 868
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00011645496612217198,
      "loss": 1.38,
      "step": 869
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00011628421048463314,
      "loss": 1.4396,
      "step": 870
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00011611340605908642,
      "loss": 1.2921,
      "step": 871
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00011594255335726724,
      "loss": 1.2498,
      "step": 872
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00011577165289105565,
      "loss": 1.4407,
      "step": 873
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0001156007051724749,
      "loss": 1.351,
      "step": 874
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00011542971071368969,
      "loss": 1.2621,
      "step": 875
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00011525867002700484,
      "loss": 1.2043,
      "step": 876
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00011508758362486358,
      "loss": 1.3231,
      "step": 877
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00011491645201984624,
      "loss": 1.4388,
      "step": 878
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00011474527572466847,
      "loss": 1.3816,
      "step": 879
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00011457405525217982,
      "loss": 1.5261,
      "step": 880
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00011440279111536227,
      "loss": 1.1705,
      "step": 881
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00011423148382732853,
      "loss": 1.458,
      "step": 882
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.00011406013390132067,
      "loss": 1.2028,
      "step": 883
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.00011388874185070843,
      "loss": 1.2344,
      "step": 884
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.00011371730818898785,
      "loss": 1.3039,
      "step": 885
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00011354583342977953,
      "loss": 1.2682,
      "step": 886
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.0001133743180868273,
      "loss": 1.2636,
      "step": 887
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.0001132027626739965,
      "loss": 1.2851,
      "step": 888
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00011303116770527256,
      "loss": 1.3523,
      "step": 889
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00011285953369475942,
      "loss": 1.2867,
      "step": 890
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00011268786115667798,
      "loss": 1.333,
      "step": 891
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.0001125161506053646,
      "loss": 1.3355,
      "step": 892
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.00011234440255526948,
      "loss": 1.2983,
      "step": 893
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.00011217261752095518,
      "loss": 1.4745,
      "step": 894
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.00011200079601709512,
      "loss": 1.3189,
      "step": 895
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.00011182893855847192,
      "loss": 1.5048,
      "step": 896
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.00011165704565997593,
      "loss": 1.4876,
      "step": 897
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0001114851178366037,
      "loss": 1.4408,
      "step": 898
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00011131315560345641,
      "loss": 1.4398,
      "step": 899
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00011114115947573833,
      "loss": 1.4294,
      "step": 900
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00011096912996875528,
      "loss": 1.3758,
      "step": 901
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00011079706759791311,
      "loss": 1.3502,
      "step": 902
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00011062497287871605,
      "loss": 1.5768,
      "step": 903
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00011045284632676536,
      "loss": 1.3264,
      "step": 904
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.00011028068845775758,
      "loss": 1.2573,
      "step": 905
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.00011010849978748314,
      "loss": 1.2733,
      "step": 906
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.00010993628083182467,
      "loss": 1.3333,
      "step": 907
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00010976403210675565,
      "loss": 1.3108,
      "step": 908
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00010959175412833869,
      "loss": 1.3264,
      "step": 909
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00010941944741272398,
      "loss": 1.2375,
      "step": 910
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.00010924711247614794,
      "loss": 1.4198,
      "step": 911
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.00010907474983493144,
      "loss": 1.43,
      "step": 912
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.00010890236000547837,
      "loss": 1.3624,
      "step": 913
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.00010872994350427415,
      "loss": 1.416,
      "step": 914
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.00010855750084788398,
      "loss": 1.3358,
      "step": 915
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.00010838503255295156,
      "loss": 1.4332,
      "step": 916
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.00010821253913619726,
      "loss": 1.3955,
      "step": 917
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.00010804002111441689,
      "loss": 1.1979,
      "step": 918
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.00010786747900447981,
      "loss": 1.4212,
      "step": 919
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.00010769491332332764,
      "loss": 1.2965,
      "step": 920
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00010752232458797262,
      "loss": 1.4038,
      "step": 921
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00010734971331549603,
      "loss": 1.539,
      "step": 922
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00010717708002304665,
      "loss": 1.3764,
      "step": 923
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.00010700442522783932,
      "loss": 1.3815,
      "step": 924
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0001068317494471532,
      "loss": 1.2545,
      "step": 925
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.00010665905319833041,
      "loss": 1.3997,
      "step": 926
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0001064863369987743,
      "loss": 1.3864,
      "step": 927
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.00010631360136594812,
      "loss": 1.5698,
      "step": 928
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.00010614084681737325,
      "loss": 1.5721,
      "step": 929
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0001059680738706277,
      "loss": 1.2551,
      "step": 930
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.00010579528304334476,
      "loss": 1.2956,
      "step": 931
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.00010562247485321115,
      "loss": 1.4605,
      "step": 932
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.00010544964981796563,
      "loss": 1.446,
      "step": 933
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.00010527680845539749,
      "loss": 1.5389,
      "step": 934
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.00010510395128334489,
      "loss": 1.4377,
      "step": 935
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.00010493107881969336,
      "loss": 1.3477,
      "step": 936
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.00010475819158237425,
      "loss": 1.3055,
      "step": 937
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.00010458529008936318,
      "loss": 1.3796,
      "step": 938
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.00010441237485867846,
      "loss": 1.3942,
      "step": 939
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.0001042394464083796,
      "loss": 1.4863,
      "step": 940
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.00010406650525656565,
      "loss": 1.3718,
      "step": 941
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.00010389355192137377,
      "loss": 1.3426,
      "step": 942
    },
    {
      "epoch": 3.02,
      "learning_rate": 0.00010372058692097764,
      "loss": 1.3124,
      "step": 943
    },
    {
      "epoch": 3.02,
      "learning_rate": 0.00010354761077358576,
      "loss": 1.3903,
      "step": 944
    },
    {
      "epoch": 3.02,
      "learning_rate": 0.00010337462399744024,
      "loss": 1.104,
      "step": 945
    },
    {
      "epoch": 3.03,
      "learning_rate": 0.00010320162711081482,
      "loss": 1.2279,
      "step": 946
    },
    {
      "epoch": 3.03,
      "learning_rate": 0.00010302862063201367,
      "loss": 1.3015,
      "step": 947
    },
    {
      "epoch": 3.03,
      "learning_rate": 0.00010285560507936961,
      "loss": 1.3083,
      "step": 948
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.00010268258097124277,
      "loss": 1.4147,
      "step": 949
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.00010250954882601875,
      "loss": 1.2396,
      "step": 950
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.00010233650916210735,
      "loss": 1.1795,
      "step": 951
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.00010216346249794087,
      "loss": 1.2717,
      "step": 952
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.00010199040935197253,
      "loss": 1.3819,
      "step": 953
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.00010181735024267505,
      "loss": 1.3903,
      "step": 954
    },
    {
      "epoch": 3.06,
      "learning_rate": 0.00010164428568853894,
      "loss": 1.3343,
      "step": 955
    },
    {
      "epoch": 3.06,
      "learning_rate": 0.00010147121620807109,
      "loss": 1.2576,
      "step": 956
    },
    {
      "epoch": 3.06,
      "learning_rate": 0.0001012981423197931,
      "loss": 1.3104,
      "step": 957
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0001011250645422398,
      "loss": 1.2107,
      "step": 958
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.00010095198339395769,
      "loss": 1.3125,
      "step": 959
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0001007788993935033,
      "loss": 1.3291,
      "step": 960
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.00010060581305944177,
      "loss": 1.2793,
      "step": 961
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.00010043272491034523,
      "loss": 1.3669,
      "step": 962
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.0001002596354647912,
      "loss": 1.3872,
      "step": 963
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.00010008654524136117,
      "loss": 1.1399,
      "step": 964
    },
    {
      "epoch": 3.09,
      "learning_rate": 9.991345475863886e-05,
      "loss": 1.2777,
      "step": 965
    },
    {
      "epoch": 3.09,
      "learning_rate": 9.974036453520881e-05,
      "loss": 1.2572,
      "step": 966
    },
    {
      "epoch": 3.09,
      "learning_rate": 9.956727508965481e-05,
      "loss": 1.3605,
      "step": 967
    },
    {
      "epoch": 3.1,
      "learning_rate": 9.939418694055823e-05,
      "loss": 1.4463,
      "step": 968
    },
    {
      "epoch": 3.1,
      "learning_rate": 9.922110060649672e-05,
      "loss": 1.348,
      "step": 969
    },
    {
      "epoch": 3.1,
      "learning_rate": 9.904801660604234e-05,
      "loss": 1.2193,
      "step": 970
    },
    {
      "epoch": 3.11,
      "learning_rate": 9.887493545776022e-05,
      "loss": 1.2353,
      "step": 971
    },
    {
      "epoch": 3.11,
      "learning_rate": 9.870185768020693e-05,
      "loss": 1.2958,
      "step": 972
    },
    {
      "epoch": 3.11,
      "learning_rate": 9.852878379192895e-05,
      "loss": 1.331,
      "step": 973
    },
    {
      "epoch": 3.12,
      "learning_rate": 9.83557143114611e-05,
      "loss": 1.179,
      "step": 974
    },
    {
      "epoch": 3.12,
      "learning_rate": 9.818264975732496e-05,
      "loss": 1.3302,
      "step": 975
    },
    {
      "epoch": 3.12,
      "learning_rate": 9.800959064802747e-05,
      "loss": 1.3682,
      "step": 976
    },
    {
      "epoch": 3.13,
      "learning_rate": 9.783653750205915e-05,
      "loss": 1.3259,
      "step": 977
    },
    {
      "epoch": 3.13,
      "learning_rate": 9.766349083789266e-05,
      "loss": 1.2246,
      "step": 978
    },
    {
      "epoch": 3.13,
      "learning_rate": 9.749045117398128e-05,
      "loss": 1.2108,
      "step": 979
    },
    {
      "epoch": 3.14,
      "learning_rate": 9.731741902875727e-05,
      "loss": 1.1323,
      "step": 980
    },
    {
      "epoch": 3.14,
      "learning_rate": 9.71443949206304e-05,
      "loss": 1.1987,
      "step": 981
    },
    {
      "epoch": 3.14,
      "learning_rate": 9.697137936798634e-05,
      "loss": 1.4362,
      "step": 982
    },
    {
      "epoch": 3.15,
      "learning_rate": 9.679837288918519e-05,
      "loss": 1.4815,
      "step": 983
    },
    {
      "epoch": 3.15,
      "learning_rate": 9.662537600255978e-05,
      "loss": 1.2198,
      "step": 984
    },
    {
      "epoch": 3.15,
      "learning_rate": 9.645238922641425e-05,
      "loss": 1.4181,
      "step": 985
    },
    {
      "epoch": 3.16,
      "learning_rate": 9.627941307902238e-05,
      "loss": 1.3045,
      "step": 986
    },
    {
      "epoch": 3.16,
      "learning_rate": 9.610644807862625e-05,
      "loss": 1.3761,
      "step": 987
    },
    {
      "epoch": 3.16,
      "learning_rate": 9.593349474343436e-05,
      "loss": 1.3264,
      "step": 988
    },
    {
      "epoch": 3.16,
      "learning_rate": 9.576055359162043e-05,
      "loss": 1.1476,
      "step": 989
    },
    {
      "epoch": 3.17,
      "learning_rate": 9.558762514132157e-05,
      "loss": 1.2759,
      "step": 990
    },
    {
      "epoch": 3.17,
      "learning_rate": 9.541470991063683e-05,
      "loss": 1.299,
      "step": 991
    },
    {
      "epoch": 3.17,
      "learning_rate": 9.524180841762577e-05,
      "loss": 1.1853,
      "step": 992
    },
    {
      "epoch": 3.18,
      "learning_rate": 9.506892118030668e-05,
      "loss": 1.1818,
      "step": 993
    },
    {
      "epoch": 3.18,
      "learning_rate": 9.489604871665515e-05,
      "loss": 1.0827,
      "step": 994
    },
    {
      "epoch": 3.18,
      "learning_rate": 9.472319154460253e-05,
      "loss": 1.34,
      "step": 995
    },
    {
      "epoch": 3.19,
      "learning_rate": 9.455035018203438e-05,
      "loss": 1.4402,
      "step": 996
    },
    {
      "epoch": 3.19,
      "learning_rate": 9.437752514678887e-05,
      "loss": 1.3385,
      "step": 997
    },
    {
      "epoch": 3.19,
      "learning_rate": 9.420471695665525e-05,
      "loss": 1.411,
      "step": 998
    },
    {
      "epoch": 3.2,
      "learning_rate": 9.403192612937231e-05,
      "loss": 1.2933,
      "step": 999
    },
    {
      "epoch": 3.2,
      "learning_rate": 9.385915318262679e-05,
      "loss": 1.1911,
      "step": 1000
    },
    {
      "epoch": 3.2,
      "learning_rate": 9.368639863405191e-05,
      "loss": 1.3399,
      "step": 1001
    },
    {
      "epoch": 3.21,
      "learning_rate": 9.35136630012257e-05,
      "loss": 1.1753,
      "step": 1002
    },
    {
      "epoch": 3.21,
      "learning_rate": 9.334094680166962e-05,
      "loss": 1.3408,
      "step": 1003
    },
    {
      "epoch": 3.21,
      "learning_rate": 9.316825055284682e-05,
      "loss": 1.2199,
      "step": 1004
    },
    {
      "epoch": 3.22,
      "learning_rate": 9.299557477216072e-05,
      "loss": 1.4121,
      "step": 1005
    },
    {
      "epoch": 3.22,
      "learning_rate": 9.282291997695338e-05,
      "loss": 1.2965,
      "step": 1006
    },
    {
      "epoch": 3.22,
      "learning_rate": 9.265028668450402e-05,
      "loss": 1.1521,
      "step": 1007
    },
    {
      "epoch": 3.23,
      "learning_rate": 9.247767541202738e-05,
      "loss": 1.2923,
      "step": 1008
    },
    {
      "epoch": 3.23,
      "learning_rate": 9.230508667667234e-05,
      "loss": 1.3632,
      "step": 1009
    },
    {
      "epoch": 3.23,
      "learning_rate": 9.21325209955202e-05,
      "loss": 1.3307,
      "step": 1010
    },
    {
      "epoch": 3.24,
      "learning_rate": 9.195997888558312e-05,
      "loss": 1.4302,
      "step": 1011
    },
    {
      "epoch": 3.24,
      "learning_rate": 9.178746086380275e-05,
      "loss": 1.2271,
      "step": 1012
    },
    {
      "epoch": 3.24,
      "learning_rate": 9.161496744704848e-05,
      "loss": 1.3719,
      "step": 1013
    },
    {
      "epoch": 3.24,
      "learning_rate": 9.144249915211605e-05,
      "loss": 1.2219,
      "step": 1014
    },
    {
      "epoch": 3.25,
      "learning_rate": 9.127005649572586e-05,
      "loss": 1.242,
      "step": 1015
    },
    {
      "epoch": 3.25,
      "learning_rate": 9.109763999452164e-05,
      "loss": 1.3562,
      "step": 1016
    },
    {
      "epoch": 3.25,
      "learning_rate": 9.092525016506858e-05,
      "loss": 1.3506,
      "step": 1017
    },
    {
      "epoch": 3.26,
      "learning_rate": 9.075288752385209e-05,
      "loss": 1.3419,
      "step": 1018
    },
    {
      "epoch": 3.26,
      "learning_rate": 9.058055258727604e-05,
      "loss": 1.3926,
      "step": 1019
    },
    {
      "epoch": 3.26,
      "learning_rate": 9.040824587166136e-05,
      "loss": 1.2554,
      "step": 1020
    },
    {
      "epoch": 3.27,
      "learning_rate": 9.023596789324437e-05,
      "loss": 1.3751,
      "step": 1021
    },
    {
      "epoch": 3.27,
      "learning_rate": 9.006371916817534e-05,
      "loss": 1.3235,
      "step": 1022
    },
    {
      "epoch": 3.27,
      "learning_rate": 8.989150021251689e-05,
      "loss": 1.2754,
      "step": 1023
    },
    {
      "epoch": 3.28,
      "learning_rate": 8.971931154224244e-05,
      "loss": 1.2093,
      "step": 1024
    },
    {
      "epoch": 3.28,
      "learning_rate": 8.954715367323468e-05,
      "loss": 1.2374,
      "step": 1025
    },
    {
      "epoch": 3.28,
      "learning_rate": 8.937502712128398e-05,
      "loss": 1.1616,
      "step": 1026
    },
    {
      "epoch": 3.29,
      "learning_rate": 8.920293240208694e-05,
      "loss": 1.2439,
      "step": 1027
    },
    {
      "epoch": 3.29,
      "learning_rate": 8.903087003124471e-05,
      "loss": 1.3077,
      "step": 1028
    },
    {
      "epoch": 3.29,
      "learning_rate": 8.885884052426168e-05,
      "loss": 1.1907,
      "step": 1029
    },
    {
      "epoch": 3.3,
      "learning_rate": 8.86868443965436e-05,
      "loss": 1.4337,
      "step": 1030
    },
    {
      "epoch": 3.3,
      "learning_rate": 8.851488216339631e-05,
      "loss": 1.2902,
      "step": 1031
    },
    {
      "epoch": 3.3,
      "learning_rate": 8.83429543400241e-05,
      "loss": 1.1554,
      "step": 1032
    },
    {
      "epoch": 3.31,
      "learning_rate": 8.817106144152811e-05,
      "loss": 1.2265,
      "step": 1033
    },
    {
      "epoch": 3.31,
      "learning_rate": 8.799920398290489e-05,
      "loss": 1.2509,
      "step": 1034
    },
    {
      "epoch": 3.31,
      "learning_rate": 8.782738247904481e-05,
      "loss": 1.3738,
      "step": 1035
    },
    {
      "epoch": 3.32,
      "learning_rate": 8.765559744473053e-05,
      "loss": 1.2889,
      "step": 1036
    },
    {
      "epoch": 3.32,
      "learning_rate": 8.748384939463543e-05,
      "loss": 1.3373,
      "step": 1037
    },
    {
      "epoch": 3.32,
      "learning_rate": 8.731213884332205e-05,
      "loss": 1.124,
      "step": 1038
    },
    {
      "epoch": 3.32,
      "learning_rate": 8.714046630524059e-05,
      "loss": 1.1227,
      "step": 1039
    },
    {
      "epoch": 3.33,
      "learning_rate": 8.696883229472748e-05,
      "loss": 1.2135,
      "step": 1040
    },
    {
      "epoch": 3.33,
      "learning_rate": 8.679723732600354e-05,
      "loss": 1.2333,
      "step": 1041
    },
    {
      "epoch": 3.33,
      "learning_rate": 8.662568191317273e-05,
      "loss": 1.1977,
      "step": 1042
    },
    {
      "epoch": 3.34,
      "learning_rate": 8.645416657022049e-05,
      "loss": 1.1225,
      "step": 1043
    },
    {
      "epoch": 3.34,
      "learning_rate": 8.628269181101216e-05,
      "loss": 1.3574,
      "step": 1044
    },
    {
      "epoch": 3.34,
      "learning_rate": 8.611125814929158e-05,
      "loss": 1.5737,
      "step": 1045
    },
    {
      "epoch": 3.35,
      "learning_rate": 8.593986609867937e-05,
      "loss": 1.3247,
      "step": 1046
    },
    {
      "epoch": 3.35,
      "learning_rate": 8.57685161726715e-05,
      "loss": 1.2416,
      "step": 1047
    },
    {
      "epoch": 3.35,
      "learning_rate": 8.559720888463776e-05,
      "loss": 1.2362,
      "step": 1048
    },
    {
      "epoch": 3.36,
      "learning_rate": 8.542594474782018e-05,
      "loss": 1.2614,
      "step": 1049
    },
    {
      "epoch": 3.36,
      "learning_rate": 8.525472427533156e-05,
      "loss": 1.3379,
      "step": 1050
    },
    {
      "epoch": 3.36,
      "learning_rate": 8.508354798015378e-05,
      "loss": 1.3216,
      "step": 1051
    },
    {
      "epoch": 3.37,
      "learning_rate": 8.491241637513644e-05,
      "loss": 1.1923,
      "step": 1052
    },
    {
      "epoch": 3.37,
      "learning_rate": 8.474132997299521e-05,
      "loss": 1.3196,
      "step": 1053
    },
    {
      "epoch": 3.37,
      "learning_rate": 8.457028928631035e-05,
      "loss": 1.1431,
      "step": 1054
    },
    {
      "epoch": 3.38,
      "learning_rate": 8.43992948275251e-05,
      "loss": 1.321,
      "step": 1055
    },
    {
      "epoch": 3.38,
      "learning_rate": 8.422834710894434e-05,
      "loss": 1.2481,
      "step": 1056
    },
    {
      "epoch": 3.38,
      "learning_rate": 8.405744664273278e-05,
      "loss": 1.1889,
      "step": 1057
    },
    {
      "epoch": 3.39,
      "learning_rate": 8.38865939409136e-05,
      "loss": 1.5162,
      "step": 1058
    },
    {
      "epoch": 3.39,
      "learning_rate": 8.371578951536688e-05,
      "loss": 1.4067,
      "step": 1059
    },
    {
      "epoch": 3.39,
      "learning_rate": 8.354503387782807e-05,
      "loss": 1.2849,
      "step": 1060
    },
    {
      "epoch": 3.4,
      "learning_rate": 8.337432753988644e-05,
      "loss": 1.3631,
      "step": 1061
    },
    {
      "epoch": 3.4,
      "learning_rate": 8.320367101298351e-05,
      "loss": 1.3636,
      "step": 1062
    },
    {
      "epoch": 3.4,
      "learning_rate": 8.303306480841169e-05,
      "loss": 1.3534,
      "step": 1063
    },
    {
      "epoch": 3.4,
      "learning_rate": 8.286250943731255e-05,
      "loss": 1.2118,
      "step": 1064
    },
    {
      "epoch": 3.41,
      "learning_rate": 8.26920054106753e-05,
      "loss": 1.2789,
      "step": 1065
    },
    {
      "epoch": 3.41,
      "learning_rate": 8.25215532393354e-05,
      "loss": 1.2696,
      "step": 1066
    },
    {
      "epoch": 3.41,
      "learning_rate": 8.235115343397295e-05,
      "loss": 1.1039,
      "step": 1067
    },
    {
      "epoch": 3.42,
      "learning_rate": 8.218080650511106e-05,
      "loss": 1.2347,
      "step": 1068
    },
    {
      "epoch": 3.42,
      "learning_rate": 8.201051296311462e-05,
      "loss": 1.2621,
      "step": 1069
    },
    {
      "epoch": 3.42,
      "learning_rate": 8.184027331818837e-05,
      "loss": 1.3454,
      "step": 1070
    },
    {
      "epoch": 3.43,
      "learning_rate": 8.167008808037567e-05,
      "loss": 1.3923,
      "step": 1071
    },
    {
      "epoch": 3.43,
      "learning_rate": 8.149995775955686e-05,
      "loss": 1.2362,
      "step": 1072
    },
    {
      "epoch": 3.43,
      "learning_rate": 8.132988286544773e-05,
      "loss": 1.2425,
      "step": 1073
    },
    {
      "epoch": 3.44,
      "learning_rate": 8.115986390759806e-05,
      "loss": 1.4329,
      "step": 1074
    },
    {
      "epoch": 3.44,
      "learning_rate": 8.098990139538992e-05,
      "loss": 1.2758,
      "step": 1075
    },
    {
      "epoch": 3.44,
      "learning_rate": 8.081999583803645e-05,
      "loss": 1.2817,
      "step": 1076
    },
    {
      "epoch": 3.45,
      "learning_rate": 8.065014774458003e-05,
      "loss": 1.1962,
      "step": 1077
    },
    {
      "epoch": 3.45,
      "learning_rate": 8.048035762389091e-05,
      "loss": 1.2684,
      "step": 1078
    },
    {
      "epoch": 3.45,
      "learning_rate": 8.031062598466561e-05,
      "loss": 1.5399,
      "step": 1079
    },
    {
      "epoch": 3.46,
      "learning_rate": 8.014095333542548e-05,
      "loss": 1.2474,
      "step": 1080
    },
    {
      "epoch": 3.46,
      "learning_rate": 7.997134018451512e-05,
      "loss": 1.4074,
      "step": 1081
    },
    {
      "epoch": 3.46,
      "learning_rate": 7.980178704010089e-05,
      "loss": 1.1161,
      "step": 1082
    },
    {
      "epoch": 3.47,
      "learning_rate": 7.963229441016937e-05,
      "loss": 1.408,
      "step": 1083
    },
    {
      "epoch": 3.47,
      "learning_rate": 7.94628628025258e-05,
      "loss": 1.3084,
      "step": 1084
    },
    {
      "epoch": 3.47,
      "learning_rate": 7.929349272479262e-05,
      "loss": 1.4394,
      "step": 1085
    },
    {
      "epoch": 3.48,
      "learning_rate": 7.912418468440794e-05,
      "loss": 1.2045,
      "step": 1086
    },
    {
      "epoch": 3.48,
      "learning_rate": 7.895493918862396e-05,
      "loss": 1.2776,
      "step": 1087
    },
    {
      "epoch": 3.48,
      "learning_rate": 7.87857567445055e-05,
      "loss": 1.1677,
      "step": 1088
    },
    {
      "epoch": 3.48,
      "learning_rate": 7.861663785892857e-05,
      "loss": 1.3639,
      "step": 1089
    },
    {
      "epoch": 3.49,
      "learning_rate": 7.844758303857861e-05,
      "loss": 1.4624,
      "step": 1090
    },
    {
      "epoch": 3.49,
      "learning_rate": 7.827859278994925e-05,
      "loss": 1.3362,
      "step": 1091
    },
    {
      "epoch": 3.49,
      "learning_rate": 7.810966761934053e-05,
      "loss": 1.3749,
      "step": 1092
    },
    {
      "epoch": 3.5,
      "learning_rate": 7.794080803285765e-05,
      "loss": 1.3206,
      "step": 1093
    },
    {
      "epoch": 3.5,
      "learning_rate": 7.777201453640923e-05,
      "loss": 1.4122,
      "step": 1094
    },
    {
      "epoch": 3.5,
      "learning_rate": 7.760328763570588e-05,
      "loss": 1.3728,
      "step": 1095
    },
    {
      "epoch": 3.51,
      "learning_rate": 7.743462783625876e-05,
      "loss": 1.2326,
      "step": 1096
    },
    {
      "epoch": 3.51,
      "learning_rate": 7.726603564337791e-05,
      "loss": 1.3717,
      "step": 1097
    },
    {
      "epoch": 3.51,
      "learning_rate": 7.709751156217089e-05,
      "loss": 1.3719,
      "step": 1098
    },
    {
      "epoch": 3.52,
      "learning_rate": 7.692905609754113e-05,
      "loss": 1.1622,
      "step": 1099
    },
    {
      "epoch": 3.52,
      "learning_rate": 7.676066975418652e-05,
      "loss": 1.4444,
      "step": 1100
    },
    {
      "epoch": 3.52,
      "learning_rate": 7.659235303659784e-05,
      "loss": 1.3996,
      "step": 1101
    },
    {
      "epoch": 3.53,
      "learning_rate": 7.642410644905726e-05,
      "loss": 1.3139,
      "step": 1102
    },
    {
      "epoch": 3.53,
      "learning_rate": 7.625593049563694e-05,
      "loss": 1.2598,
      "step": 1103
    },
    {
      "epoch": 3.53,
      "learning_rate": 7.608782568019729e-05,
      "loss": 1.1569,
      "step": 1104
    },
    {
      "epoch": 3.54,
      "learning_rate": 7.591979250638562e-05,
      "loss": 1.1641,
      "step": 1105
    },
    {
      "epoch": 3.54,
      "learning_rate": 7.575183147763466e-05,
      "loss": 1.1082,
      "step": 1106
    },
    {
      "epoch": 3.54,
      "learning_rate": 7.558394309716088e-05,
      "loss": 1.3512,
      "step": 1107
    },
    {
      "epoch": 3.55,
      "learning_rate": 7.541612786796318e-05,
      "loss": 1.194,
      "step": 1108
    },
    {
      "epoch": 3.55,
      "learning_rate": 7.524838629282132e-05,
      "loss": 1.3879,
      "step": 1109
    },
    {
      "epoch": 3.55,
      "learning_rate": 7.508071887429433e-05,
      "loss": 1.1615,
      "step": 1110
    },
    {
      "epoch": 3.56,
      "learning_rate": 7.491312611471908e-05,
      "loss": 1.2892,
      "step": 1111
    },
    {
      "epoch": 3.56,
      "learning_rate": 7.474560851620873e-05,
      "loss": 1.1959,
      "step": 1112
    },
    {
      "epoch": 3.56,
      "learning_rate": 7.457816658065134e-05,
      "loss": 1.3179,
      "step": 1113
    },
    {
      "epoch": 3.56,
      "learning_rate": 7.441080080970817e-05,
      "loss": 1.2095,
      "step": 1114
    },
    {
      "epoch": 3.57,
      "learning_rate": 7.424351170481231e-05,
      "loss": 1.125,
      "step": 1115
    },
    {
      "epoch": 3.57,
      "learning_rate": 7.407629976716732e-05,
      "loss": 1.4471,
      "step": 1116
    },
    {
      "epoch": 3.57,
      "learning_rate": 7.390916549774536e-05,
      "loss": 1.3451,
      "step": 1117
    },
    {
      "epoch": 3.58,
      "learning_rate": 7.374210939728596e-05,
      "loss": 1.4308,
      "step": 1118
    },
    {
      "epoch": 3.58,
      "learning_rate": 7.35751319662945e-05,
      "loss": 1.4483,
      "step": 1119
    },
    {
      "epoch": 3.58,
      "learning_rate": 7.34082337050406e-05,
      "loss": 1.3593,
      "step": 1120
    },
    {
      "epoch": 3.59,
      "learning_rate": 7.324141511355671e-05,
      "loss": 1.4362,
      "step": 1121
    },
    {
      "epoch": 3.59,
      "learning_rate": 7.307467669163655e-05,
      "loss": 1.3069,
      "step": 1122
    },
    {
      "epoch": 3.59,
      "learning_rate": 7.290801893883376e-05,
      "loss": 1.3042,
      "step": 1123
    },
    {
      "epoch": 3.6,
      "learning_rate": 7.274144235446023e-05,
      "loss": 1.3799,
      "step": 1124
    },
    {
      "epoch": 3.6,
      "learning_rate": 7.257494743758459e-05,
      "loss": 1.3643,
      "step": 1125
    },
    {
      "epoch": 3.6,
      "learning_rate": 7.24085346870309e-05,
      "loss": 1.3339,
      "step": 1126
    },
    {
      "epoch": 3.61,
      "learning_rate": 7.224220460137701e-05,
      "loss": 1.2917,
      "step": 1127
    },
    {
      "epoch": 3.61,
      "learning_rate": 7.207595767895302e-05,
      "loss": 1.2448,
      "step": 1128
    },
    {
      "epoch": 3.61,
      "learning_rate": 7.190979441784007e-05,
      "loss": 1.5699,
      "step": 1129
    },
    {
      "epoch": 3.62,
      "learning_rate": 7.174371531586848e-05,
      "loss": 1.3727,
      "step": 1130
    },
    {
      "epoch": 3.62,
      "learning_rate": 7.157772087061645e-05,
      "loss": 1.455,
      "step": 1131
    },
    {
      "epoch": 3.62,
      "learning_rate": 7.141181157940859e-05,
      "loss": 1.4807,
      "step": 1132
    },
    {
      "epoch": 3.63,
      "learning_rate": 7.124598793931437e-05,
      "loss": 1.3418,
      "step": 1133
    },
    {
      "epoch": 3.63,
      "learning_rate": 7.108025044714661e-05,
      "loss": 1.376,
      "step": 1134
    },
    {
      "epoch": 3.63,
      "learning_rate": 7.09145995994601e-05,
      "loss": 1.4902,
      "step": 1135
    },
    {
      "epoch": 3.64,
      "learning_rate": 7.074903589255e-05,
      "loss": 0.9871,
      "step": 1136
    },
    {
      "epoch": 3.64,
      "learning_rate": 7.058355982245037e-05,
      "loss": 1.2642,
      "step": 1137
    },
    {
      "epoch": 3.64,
      "learning_rate": 7.041817188493276e-05,
      "loss": 1.4263,
      "step": 1138
    },
    {
      "epoch": 3.64,
      "learning_rate": 7.025287257550461e-05,
      "loss": 1.4165,
      "step": 1139
    },
    {
      "epoch": 3.65,
      "learning_rate": 7.00876623894079e-05,
      "loss": 1.4433,
      "step": 1140
    },
    {
      "epoch": 3.65,
      "learning_rate": 6.992254182161753e-05,
      "loss": 1.1097,
      "step": 1141
    },
    {
      "epoch": 3.65,
      "learning_rate": 6.97575113668399e-05,
      "loss": 1.331,
      "step": 1142
    },
    {
      "epoch": 3.66,
      "learning_rate": 6.959257151951153e-05,
      "loss": 1.2074,
      "step": 1143
    },
    {
      "epoch": 3.66,
      "learning_rate": 6.942772277379733e-05,
      "loss": 1.4632,
      "step": 1144
    },
    {
      "epoch": 3.66,
      "learning_rate": 6.926296562358936e-05,
      "loss": 1.1376,
      "step": 1145
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.909830056250527e-05,
      "loss": 1.3313,
      "step": 1146
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.893372808388675e-05,
      "loss": 1.1283,
      "step": 1147
    },
    {
      "epoch": 3.67,
      "learning_rate": 6.87692486807981e-05,
      "loss": 1.2681,
      "step": 1148
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.860486284602478e-05,
      "loss": 1.0567,
      "step": 1149
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.844057107207205e-05,
      "loss": 1.2563,
      "step": 1150
    },
    {
      "epoch": 3.68,
      "learning_rate": 6.827637385116314e-05,
      "loss": 1.1704,
      "step": 1151
    },
    {
      "epoch": 3.69,
      "learning_rate": 6.811227167523815e-05,
      "loss": 1.2412,
      "step": 1152
    },
    {
      "epoch": 3.69,
      "learning_rate": 6.794826503595235e-05,
      "loss": 1.3055,
      "step": 1153
    },
    {
      "epoch": 3.69,
      "learning_rate": 6.778435442467479e-05,
      "loss": 1.3366,
      "step": 1154
    },
    {
      "epoch": 3.7,
      "learning_rate": 6.762054033248681e-05,
      "loss": 1.2709,
      "step": 1155
    },
    {
      "epoch": 3.7,
      "learning_rate": 6.745682325018066e-05,
      "loss": 1.4311,
      "step": 1156
    },
    {
      "epoch": 3.7,
      "learning_rate": 6.729320366825784e-05,
      "loss": 1.3485,
      "step": 1157
    },
    {
      "epoch": 3.71,
      "learning_rate": 6.712968207692778e-05,
      "loss": 1.1492,
      "step": 1158
    },
    {
      "epoch": 3.71,
      "learning_rate": 6.696625896610631e-05,
      "loss": 1.3594,
      "step": 1159
    },
    {
      "epoch": 3.71,
      "learning_rate": 6.680293482541424e-05,
      "loss": 1.065,
      "step": 1160
    },
    {
      "epoch": 3.72,
      "learning_rate": 6.663971014417586e-05,
      "loss": 1.4161,
      "step": 1161
    },
    {
      "epoch": 3.72,
      "learning_rate": 6.647658541141735e-05,
      "loss": 1.3668,
      "step": 1162
    },
    {
      "epoch": 3.72,
      "learning_rate": 6.63135611158657e-05,
      "loss": 1.2785,
      "step": 1163
    },
    {
      "epoch": 3.72,
      "learning_rate": 6.615063774594676e-05,
      "loss": 1.1588,
      "step": 1164
    },
    {
      "epoch": 3.73,
      "learning_rate": 6.59878157897841e-05,
      "loss": 1.2491,
      "step": 1165
    },
    {
      "epoch": 3.73,
      "learning_rate": 6.582509573519744e-05,
      "loss": 1.2931,
      "step": 1166
    },
    {
      "epoch": 3.73,
      "learning_rate": 6.566247806970119e-05,
      "loss": 1.2204,
      "step": 1167
    },
    {
      "epoch": 3.74,
      "learning_rate": 6.549996328050296e-05,
      "loss": 1.3976,
      "step": 1168
    },
    {
      "epoch": 3.74,
      "learning_rate": 6.533755185450219e-05,
      "loss": 1.3315,
      "step": 1169
    },
    {
      "epoch": 3.74,
      "learning_rate": 6.51752442782887e-05,
      "loss": 1.2285,
      "step": 1170
    },
    {
      "epoch": 3.75,
      "learning_rate": 6.501304103814105e-05,
      "loss": 1.3739,
      "step": 1171
    },
    {
      "epoch": 3.75,
      "learning_rate": 6.485094262002529e-05,
      "loss": 1.2786,
      "step": 1172
    },
    {
      "epoch": 3.75,
      "learning_rate": 6.468894950959336e-05,
      "loss": 1.249,
      "step": 1173
    },
    {
      "epoch": 3.76,
      "learning_rate": 6.452706219218173e-05,
      "loss": 1.1481,
      "step": 1174
    },
    {
      "epoch": 3.76,
      "learning_rate": 6.436528115280988e-05,
      "loss": 1.2411,
      "step": 1175
    },
    {
      "epoch": 3.76,
      "learning_rate": 6.420360687617897e-05,
      "loss": 1.2995,
      "step": 1176
    },
    {
      "epoch": 3.77,
      "learning_rate": 6.404203984667019e-05,
      "loss": 1.3201,
      "step": 1177
    },
    {
      "epoch": 3.77,
      "learning_rate": 6.388058054834345e-05,
      "loss": 1.3395,
      "step": 1178
    },
    {
      "epoch": 3.77,
      "learning_rate": 6.371922946493591e-05,
      "loss": 1.3594,
      "step": 1179
    },
    {
      "epoch": 3.78,
      "learning_rate": 6.355798707986052e-05,
      "loss": 1.3485,
      "step": 1180
    },
    {
      "epoch": 3.78,
      "learning_rate": 6.339685387620449e-05,
      "loss": 1.2761,
      "step": 1181
    },
    {
      "epoch": 3.78,
      "learning_rate": 6.323583033672799e-05,
      "loss": 1.1875,
      "step": 1182
    },
    {
      "epoch": 3.79,
      "learning_rate": 6.307491694386268e-05,
      "loss": 1.3771,
      "step": 1183
    },
    {
      "epoch": 3.79,
      "learning_rate": 6.291411417971012e-05,
      "loss": 1.3409,
      "step": 1184
    },
    {
      "epoch": 3.79,
      "learning_rate": 6.275342252604044e-05,
      "loss": 1.4196,
      "step": 1185
    },
    {
      "epoch": 3.8,
      "learning_rate": 6.259284246429093e-05,
      "loss": 1.1734,
      "step": 1186
    },
    {
      "epoch": 3.8,
      "learning_rate": 6.243237447556449e-05,
      "loss": 1.2801,
      "step": 1187
    },
    {
      "epoch": 3.8,
      "learning_rate": 6.227201904062829e-05,
      "loss": 1.341,
      "step": 1188
    },
    {
      "epoch": 3.8,
      "learning_rate": 6.211177663991225e-05,
      "loss": 1.0995,
      "step": 1189
    },
    {
      "epoch": 3.81,
      "learning_rate": 6.19516477535077e-05,
      "loss": 1.1759,
      "step": 1190
    },
    {
      "epoch": 3.81,
      "learning_rate": 6.179163286116581e-05,
      "loss": 1.1452,
      "step": 1191
    },
    {
      "epoch": 3.81,
      "learning_rate": 6.163173244229619e-05,
      "loss": 1.287,
      "step": 1192
    },
    {
      "epoch": 3.82,
      "learning_rate": 6.147194697596559e-05,
      "loss": 1.2441,
      "step": 1193
    },
    {
      "epoch": 3.82,
      "learning_rate": 6.13122769408963e-05,
      "loss": 1.3802,
      "step": 1194
    },
    {
      "epoch": 3.82,
      "learning_rate": 6.115272281546474e-05,
      "loss": 1.277,
      "step": 1195
    },
    {
      "epoch": 3.83,
      "learning_rate": 6.0993285077700145e-05,
      "loss": 1.2389,
      "step": 1196
    },
    {
      "epoch": 3.83,
      "learning_rate": 6.083396420528298e-05,
      "loss": 1.3652,
      "step": 1197
    },
    {
      "epoch": 3.83,
      "learning_rate": 6.0674760675543607e-05,
      "loss": 1.3338,
      "step": 1198
    },
    {
      "epoch": 3.84,
      "learning_rate": 6.051567496546082e-05,
      "loss": 1.2031,
      "step": 1199
    },
    {
      "epoch": 3.84,
      "learning_rate": 6.035670755166043e-05,
      "loss": 1.3884,
      "step": 1200
    },
    {
      "epoch": 3.84,
      "learning_rate": 6.019785891041381e-05,
      "loss": 1.3153,
      "step": 1201
    },
    {
      "epoch": 3.85,
      "learning_rate": 6.0039129517636435e-05,
      "loss": 1.2513,
      "step": 1202
    },
    {
      "epoch": 3.85,
      "learning_rate": 5.988051984888667e-05,
      "loss": 1.2049,
      "step": 1203
    },
    {
      "epoch": 3.85,
      "learning_rate": 5.972203037936406e-05,
      "loss": 1.0361,
      "step": 1204
    },
    {
      "epoch": 3.86,
      "learning_rate": 5.956366158390802e-05,
      "loss": 1.4749,
      "step": 1205
    },
    {
      "epoch": 3.86,
      "learning_rate": 5.940541393699646e-05,
      "loss": 1.2684,
      "step": 1206
    },
    {
      "epoch": 3.86,
      "learning_rate": 5.924728791274432e-05,
      "loss": 1.3203,
      "step": 1207
    },
    {
      "epoch": 3.87,
      "learning_rate": 5.9089283984902166e-05,
      "loss": 1.2304,
      "step": 1208
    },
    {
      "epoch": 3.87,
      "learning_rate": 5.893140262685469e-05,
      "loss": 1.2197,
      "step": 1209
    },
    {
      "epoch": 3.87,
      "learning_rate": 5.8773644311619514e-05,
      "loss": 1.3555,
      "step": 1210
    },
    {
      "epoch": 3.88,
      "learning_rate": 5.861600951184545e-05,
      "loss": 1.2924,
      "step": 1211
    },
    {
      "epoch": 3.88,
      "learning_rate": 5.845849869981137e-05,
      "loss": 1.21,
      "step": 1212
    },
    {
      "epoch": 3.88,
      "learning_rate": 5.830111234742459e-05,
      "loss": 1.2152,
      "step": 1213
    },
    {
      "epoch": 3.88,
      "learning_rate": 5.814385092621961e-05,
      "loss": 1.2634,
      "step": 1214
    },
    {
      "epoch": 3.89,
      "learning_rate": 5.798671490735661e-05,
      "loss": 1.3623,
      "step": 1215
    },
    {
      "epoch": 3.89,
      "learning_rate": 5.782970476162004e-05,
      "loss": 1.3035,
      "step": 1216
    },
    {
      "epoch": 3.89,
      "learning_rate": 5.7672820959417254e-05,
      "loss": 1.3839,
      "step": 1217
    },
    {
      "epoch": 3.9,
      "learning_rate": 5.751606397077702e-05,
      "loss": 1.2685,
      "step": 1218
    },
    {
      "epoch": 3.9,
      "learning_rate": 5.73594342653483e-05,
      "loss": 1.2318,
      "step": 1219
    },
    {
      "epoch": 3.9,
      "learning_rate": 5.7202932312398594e-05,
      "loss": 1.2179,
      "step": 1220
    },
    {
      "epoch": 3.91,
      "learning_rate": 5.704655858081268e-05,
      "loss": 1.3159,
      "step": 1221
    },
    {
      "epoch": 3.91,
      "learning_rate": 5.68903135390912e-05,
      "loss": 1.2683,
      "step": 1222
    },
    {
      "epoch": 3.91,
      "learning_rate": 5.6734197655349156e-05,
      "loss": 1.401,
      "step": 1223
    },
    {
      "epoch": 3.92,
      "learning_rate": 5.657821139731476e-05,
      "loss": 1.064,
      "step": 1224
    },
    {
      "epoch": 3.92,
      "learning_rate": 5.6422355232327714e-05,
      "loss": 1.2446,
      "step": 1225
    },
    {
      "epoch": 3.92,
      "learning_rate": 5.6266629627338e-05,
      "loss": 1.255,
      "step": 1226
    },
    {
      "epoch": 3.93,
      "learning_rate": 5.611103504890444e-05,
      "loss": 1.293,
      "step": 1227
    },
    {
      "epoch": 3.93,
      "learning_rate": 5.595557196319329e-05,
      "loss": 1.2781,
      "step": 1228
    },
    {
      "epoch": 3.93,
      "learning_rate": 5.5800240835976813e-05,
      "loss": 1.3756,
      "step": 1229
    },
    {
      "epoch": 3.94,
      "learning_rate": 5.564504213263205e-05,
      "loss": 1.2948,
      "step": 1230
    },
    {
      "epoch": 3.94,
      "learning_rate": 5.548997631813917e-05,
      "loss": 1.4605,
      "step": 1231
    },
    {
      "epoch": 3.94,
      "learning_rate": 5.533504385708024e-05,
      "loss": 1.3298,
      "step": 1232
    },
    {
      "epoch": 3.95,
      "learning_rate": 5.518024521363778e-05,
      "loss": 1.4434,
      "step": 1233
    },
    {
      "epoch": 3.95,
      "learning_rate": 5.5025580851593436e-05,
      "loss": 1.4257,
      "step": 1234
    },
    {
      "epoch": 3.95,
      "learning_rate": 5.487105123432646e-05,
      "loss": 1.2131,
      "step": 1235
    },
    {
      "epoch": 3.96,
      "learning_rate": 5.4716656824812505e-05,
      "loss": 1.2569,
      "step": 1236
    },
    {
      "epoch": 3.96,
      "learning_rate": 5.456239808562209e-05,
      "loss": 1.3278,
      "step": 1237
    },
    {
      "epoch": 3.96,
      "learning_rate": 5.440827547891925e-05,
      "loss": 1.2411,
      "step": 1238
    },
    {
      "epoch": 3.96,
      "learning_rate": 5.425428946646016e-05,
      "loss": 1.1827,
      "step": 1239
    },
    {
      "epoch": 3.97,
      "learning_rate": 5.4100440509591776e-05,
      "loss": 1.4103,
      "step": 1240
    },
    {
      "epoch": 3.97,
      "learning_rate": 5.394672906925042e-05,
      "loss": 1.2475,
      "step": 1241
    },
    {
      "epoch": 3.97,
      "learning_rate": 5.379315560596038e-05,
      "loss": 1.2197,
      "step": 1242
    },
    {
      "epoch": 3.98,
      "learning_rate": 5.363972057983265e-05,
      "loss": 1.1452,
      "step": 1243
    },
    {
      "epoch": 3.98,
      "learning_rate": 5.34864244505634e-05,
      "loss": 1.1936,
      "step": 1244
    },
    {
      "epoch": 3.98,
      "learning_rate": 5.333326767743263e-05,
      "loss": 1.2406,
      "step": 1245
    },
    {
      "epoch": 3.99,
      "learning_rate": 5.318025071930288e-05,
      "loss": 1.3903,
      "step": 1246
    },
    {
      "epoch": 3.99,
      "learning_rate": 5.3027374034617785e-05,
      "loss": 1.2061,
      "step": 1247
    },
    {
      "epoch": 3.99,
      "learning_rate": 5.2874638081400694e-05,
      "loss": 1.3037,
      "step": 1248
    },
    {
      "epoch": 4.0,
      "learning_rate": 5.272204331725331e-05,
      "loss": 1.5323,
      "step": 1249
    },
    {
      "epoch": 4.0,
      "learning_rate": 5.2569590199354434e-05,
      "loss": 1.1467,
      "step": 1250
    },
    {
      "epoch": 4.0,
      "learning_rate": 5.241727918445836e-05,
      "loss": 1.2066,
      "step": 1251
    },
    {
      "epoch": 4.01,
      "learning_rate": 5.226511072889371e-05,
      "loss": 1.3306,
      "step": 1252
    },
    {
      "epoch": 4.01,
      "learning_rate": 5.211308528856196e-05,
      "loss": 1.3291,
      "step": 1253
    },
    {
      "epoch": 4.01,
      "learning_rate": 5.1961203318936116e-05,
      "loss": 1.2561,
      "step": 1254
    },
    {
      "epoch": 4.02,
      "learning_rate": 5.180946527505932e-05,
      "loss": 1.1574,
      "step": 1255
    },
    {
      "epoch": 4.02,
      "learning_rate": 5.1657871611543605e-05,
      "loss": 1.1799,
      "step": 1256
    },
    {
      "epoch": 4.02,
      "learning_rate": 5.1506422782568345e-05,
      "loss": 1.0446,
      "step": 1257
    },
    {
      "epoch": 4.03,
      "learning_rate": 5.135511924187897e-05,
      "loss": 1.2101,
      "step": 1258
    },
    {
      "epoch": 4.03,
      "learning_rate": 5.120396144278569e-05,
      "loss": 1.1848,
      "step": 1259
    },
    {
      "epoch": 4.03,
      "learning_rate": 5.105294983816202e-05,
      "loss": 1.257,
      "step": 1260
    },
    {
      "epoch": 4.04,
      "learning_rate": 5.090208488044348e-05,
      "loss": 1.1187,
      "step": 1261
    },
    {
      "epoch": 4.04,
      "learning_rate": 5.0751367021626215e-05,
      "loss": 1.21,
      "step": 1262
    },
    {
      "epoch": 4.04,
      "learning_rate": 5.0600796713265765e-05,
      "loss": 1.2219,
      "step": 1263
    },
    {
      "epoch": 4.04,
      "learning_rate": 5.045037440647547e-05,
      "loss": 1.1662,
      "step": 1264
    },
    {
      "epoch": 4.05,
      "learning_rate": 5.030010055192529e-05,
      "loss": 1.115,
      "step": 1265
    },
    {
      "epoch": 4.05,
      "learning_rate": 5.014997559984045e-05,
      "loss": 1.1381,
      "step": 1266
    },
    {
      "epoch": 4.05,
      "learning_rate": 5.000000000000002e-05,
      "loss": 1.0469,
      "step": 1267
    },
    {
      "epoch": 4.06,
      "learning_rate": 4.985017420173564e-05,
      "loss": 1.2503,
      "step": 1268
    },
    {
      "epoch": 4.06,
      "learning_rate": 4.970049865393008e-05,
      "loss": 1.4018,
      "step": 1269
    },
    {
      "epoch": 4.06,
      "learning_rate": 4.955097380501609e-05,
      "loss": 1.2093,
      "step": 1270
    },
    {
      "epoch": 4.07,
      "learning_rate": 4.940160010297476e-05,
      "loss": 1.128,
      "step": 1271
    },
    {
      "epoch": 4.07,
      "learning_rate": 4.9252377995334444e-05,
      "loss": 1.1447,
      "step": 1272
    },
    {
      "epoch": 4.07,
      "learning_rate": 4.9103307929169265e-05,
      "loss": 1.2808,
      "step": 1273
    },
    {
      "epoch": 4.08,
      "learning_rate": 4.895439035109783e-05,
      "loss": 1.2258,
      "step": 1274
    },
    {
      "epoch": 4.08,
      "learning_rate": 4.8805625707281877e-05,
      "loss": 1.0974,
      "step": 1275
    },
    {
      "epoch": 4.08,
      "learning_rate": 4.865701444342503e-05,
      "loss": 1.1391,
      "step": 1276
    },
    {
      "epoch": 4.09,
      "learning_rate": 4.85085570047713e-05,
      "loss": 1.17,
      "step": 1277
    },
    {
      "epoch": 4.09,
      "learning_rate": 4.836025383610382e-05,
      "loss": 1.1401,
      "step": 1278
    },
    {
      "epoch": 4.09,
      "learning_rate": 4.8212105381743586e-05,
      "loss": 1.0493,
      "step": 1279
    },
    {
      "epoch": 4.1,
      "learning_rate": 4.806411208554804e-05,
      "loss": 1.2585,
      "step": 1280
    },
    {
      "epoch": 4.1,
      "learning_rate": 4.791627439090975e-05,
      "loss": 1.0471,
      "step": 1281
    },
    {
      "epoch": 4.1,
      "learning_rate": 4.776859274075506e-05,
      "loss": 1.1916,
      "step": 1282
    },
    {
      "epoch": 4.11,
      "learning_rate": 4.7621067577542944e-05,
      "loss": 1.1773,
      "step": 1283
    },
    {
      "epoch": 4.11,
      "learning_rate": 4.74736993432634e-05,
      "loss": 1.185,
      "step": 1284
    },
    {
      "epoch": 4.11,
      "learning_rate": 4.732648847943629e-05,
      "loss": 1.1491,
      "step": 1285
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.717943542711003e-05,
      "loss": 1.1123,
      "step": 1286
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.703254062686017e-05,
      "loss": 1.2279,
      "step": 1287
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.688580451878818e-05,
      "loss": 1.2954,
      "step": 1288
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.673922754252002e-05,
      "loss": 1.2513,
      "step": 1289
    },
    {
      "epoch": 4.13,
      "learning_rate": 4.6592810137205e-05,
      "loss": 1.0544,
      "step": 1290
    },
    {
      "epoch": 4.13,
      "learning_rate": 4.644655274151426e-05,
      "loss": 1.2183,
      "step": 1291
    },
    {
      "epoch": 4.13,
      "learning_rate": 4.630045579363957e-05,
      "loss": 1.409,
      "step": 1292
    },
    {
      "epoch": 4.14,
      "learning_rate": 4.615451973129196e-05,
      "loss": 1.2237,
      "step": 1293
    },
    {
      "epoch": 4.14,
      "learning_rate": 4.60087449917005e-05,
      "loss": 1.0769,
      "step": 1294
    },
    {
      "epoch": 4.14,
      "learning_rate": 4.586313201161086e-05,
      "loss": 1.1479,
      "step": 1295
    },
    {
      "epoch": 4.15,
      "learning_rate": 4.57176812272842e-05,
      "loss": 1.1276,
      "step": 1296
    },
    {
      "epoch": 4.15,
      "learning_rate": 4.557239307449561e-05,
      "loss": 1.2646,
      "step": 1297
    },
    {
      "epoch": 4.15,
      "learning_rate": 4.542726798853297e-05,
      "loss": 1.269,
      "step": 1298
    },
    {
      "epoch": 4.16,
      "learning_rate": 4.528230640419562e-05,
      "loss": 1.1258,
      "step": 1299
    },
    {
      "epoch": 4.16,
      "learning_rate": 4.513750875579303e-05,
      "loss": 1.2835,
      "step": 1300
    },
    {
      "epoch": 4.16,
      "learning_rate": 4.499287547714354e-05,
      "loss": 1.1923,
      "step": 1301
    },
    {
      "epoch": 4.17,
      "learning_rate": 4.484840700157295e-05,
      "loss": 1.1872,
      "step": 1302
    },
    {
      "epoch": 4.17,
      "learning_rate": 4.470410376191344e-05,
      "loss": 1.3357,
      "step": 1303
    },
    {
      "epoch": 4.17,
      "learning_rate": 4.455996619050205e-05,
      "loss": 1.1122,
      "step": 1304
    },
    {
      "epoch": 4.18,
      "learning_rate": 4.4415994719179456e-05,
      "loss": 1.2327,
      "step": 1305
    },
    {
      "epoch": 4.18,
      "learning_rate": 4.427218977928875e-05,
      "loss": 1.2382,
      "step": 1306
    },
    {
      "epoch": 4.18,
      "learning_rate": 4.412855180167406e-05,
      "loss": 1.2533,
      "step": 1307
    },
    {
      "epoch": 4.19,
      "learning_rate": 4.398508121667925e-05,
      "loss": 1.1103,
      "step": 1308
    },
    {
      "epoch": 4.19,
      "learning_rate": 4.384177845414679e-05,
      "loss": 1.2481,
      "step": 1309
    },
    {
      "epoch": 4.19,
      "learning_rate": 4.3698643943416236e-05,
      "loss": 1.1265,
      "step": 1310
    },
    {
      "epoch": 4.2,
      "learning_rate": 4.355567811332311e-05,
      "loss": 1.219,
      "step": 1311
    },
    {
      "epoch": 4.2,
      "learning_rate": 4.3412881392197526e-05,
      "loss": 1.1521,
      "step": 1312
    },
    {
      "epoch": 4.2,
      "learning_rate": 4.327025420786293e-05,
      "loss": 1.2066,
      "step": 1313
    },
    {
      "epoch": 4.2,
      "learning_rate": 4.312779698763493e-05,
      "loss": 1.201,
      "step": 1314
    },
    {
      "epoch": 4.21,
      "learning_rate": 4.2985510158319834e-05,
      "loss": 1.2988,
      "step": 1315
    },
    {
      "epoch": 4.21,
      "learning_rate": 4.284339414621343e-05,
      "loss": 1.0763,
      "step": 1316
    },
    {
      "epoch": 4.21,
      "learning_rate": 4.270144937709981e-05,
      "loss": 1.08,
      "step": 1317
    },
    {
      "epoch": 4.22,
      "learning_rate": 4.255967627624993e-05,
      "loss": 1.0768,
      "step": 1318
    },
    {
      "epoch": 4.22,
      "learning_rate": 4.241807526842051e-05,
      "loss": 1.1131,
      "step": 1319
    },
    {
      "epoch": 4.22,
      "learning_rate": 4.2276646777852636e-05,
      "loss": 1.1794,
      "step": 1320
    },
    {
      "epoch": 4.23,
      "learning_rate": 4.2135391228270506e-05,
      "loss": 1.2526,
      "step": 1321
    },
    {
      "epoch": 4.23,
      "learning_rate": 4.19943090428802e-05,
      "loss": 1.361,
      "step": 1322
    },
    {
      "epoch": 4.23,
      "learning_rate": 4.185340064436839e-05,
      "loss": 1.1011,
      "step": 1323
    },
    {
      "epoch": 4.24,
      "learning_rate": 4.171266645490105e-05,
      "loss": 1.3457,
      "step": 1324
    },
    {
      "epoch": 4.24,
      "learning_rate": 4.157210689612231e-05,
      "loss": 1.1207,
      "step": 1325
    },
    {
      "epoch": 4.24,
      "learning_rate": 4.1431722389153016e-05,
      "loss": 1.3307,
      "step": 1326
    },
    {
      "epoch": 4.25,
      "learning_rate": 4.129151335458957e-05,
      "loss": 1.2303,
      "step": 1327
    },
    {
      "epoch": 4.25,
      "learning_rate": 4.1151480212502656e-05,
      "loss": 1.2741,
      "step": 1328
    },
    {
      "epoch": 4.25,
      "learning_rate": 4.101162338243595e-05,
      "loss": 1.1483,
      "step": 1329
    },
    {
      "epoch": 4.26,
      "learning_rate": 4.087194328340499e-05,
      "loss": 1.0521,
      "step": 1330
    },
    {
      "epoch": 4.26,
      "learning_rate": 4.073244033389572e-05,
      "loss": 1.0913,
      "step": 1331
    },
    {
      "epoch": 4.26,
      "learning_rate": 4.059311495186338e-05,
      "loss": 1.3718,
      "step": 1332
    },
    {
      "epoch": 4.27,
      "learning_rate": 4.045396755473121e-05,
      "loss": 1.2559,
      "step": 1333
    },
    {
      "epoch": 4.27,
      "learning_rate": 4.031499855938918e-05,
      "loss": 1.2552,
      "step": 1334
    },
    {
      "epoch": 4.27,
      "learning_rate": 4.017620838219276e-05,
      "loss": 1.0405,
      "step": 1335
    },
    {
      "epoch": 4.28,
      "learning_rate": 4.003759743896176e-05,
      "loss": 1.194,
      "step": 1336
    },
    {
      "epoch": 4.28,
      "learning_rate": 3.9899166144978904e-05,
      "loss": 1.2027,
      "step": 1337
    },
    {
      "epoch": 4.28,
      "learning_rate": 3.976091491498871e-05,
      "loss": 1.1522,
      "step": 1338
    },
    {
      "epoch": 4.28,
      "learning_rate": 3.962284416319623e-05,
      "loss": 1.1695,
      "step": 1339
    },
    {
      "epoch": 4.29,
      "learning_rate": 3.948495430326577e-05,
      "loss": 1.3513,
      "step": 1340
    },
    {
      "epoch": 4.29,
      "learning_rate": 3.9347245748319705e-05,
      "loss": 1.0502,
      "step": 1341
    },
    {
      "epoch": 4.29,
      "learning_rate": 3.920971891093718e-05,
      "loss": 1.313,
      "step": 1342
    },
    {
      "epoch": 4.3,
      "learning_rate": 3.907237420315298e-05,
      "loss": 1.1907,
      "step": 1343
    },
    {
      "epoch": 4.3,
      "learning_rate": 3.893521203645618e-05,
      "loss": 1.083,
      "step": 1344
    },
    {
      "epoch": 4.3,
      "learning_rate": 3.879823282178894e-05,
      "loss": 1.2055,
      "step": 1345
    },
    {
      "epoch": 4.31,
      "learning_rate": 3.8661436969545286e-05,
      "loss": 1.3066,
      "step": 1346
    },
    {
      "epoch": 4.31,
      "learning_rate": 3.852482488956992e-05,
      "loss": 1.2298,
      "step": 1347
    },
    {
      "epoch": 4.31,
      "learning_rate": 3.838839699115695e-05,
      "loss": 1.141,
      "step": 1348
    },
    {
      "epoch": 4.32,
      "learning_rate": 3.82521536830486e-05,
      "loss": 1.1427,
      "step": 1349
    },
    {
      "epoch": 4.32,
      "learning_rate": 3.81160953734342e-05,
      "loss": 1.127,
      "step": 1350
    },
    {
      "epoch": 4.32,
      "learning_rate": 3.79802224699487e-05,
      "loss": 1.1916,
      "step": 1351
    },
    {
      "epoch": 4.33,
      "learning_rate": 3.784453537967161e-05,
      "loss": 1.2902,
      "step": 1352
    },
    {
      "epoch": 4.33,
      "learning_rate": 3.770903450912571e-05,
      "loss": 1.1547,
      "step": 1353
    },
    {
      "epoch": 4.33,
      "learning_rate": 3.757372026427588e-05,
      "loss": 1.271,
      "step": 1354
    },
    {
      "epoch": 4.34,
      "learning_rate": 3.7438593050527845e-05,
      "loss": 1.1449,
      "step": 1355
    },
    {
      "epoch": 4.34,
      "learning_rate": 3.7303653272727057e-05,
      "loss": 1.1238,
      "step": 1356
    },
    {
      "epoch": 4.34,
      "learning_rate": 3.7168901335157315e-05,
      "loss": 1.0949,
      "step": 1357
    },
    {
      "epoch": 4.35,
      "learning_rate": 3.703433764153965e-05,
      "loss": 1.2965,
      "step": 1358
    },
    {
      "epoch": 4.35,
      "learning_rate": 3.689996259503116e-05,
      "loss": 1.3075,
      "step": 1359
    },
    {
      "epoch": 4.35,
      "learning_rate": 3.67657765982237e-05,
      "loss": 1.2963,
      "step": 1360
    },
    {
      "epoch": 4.36,
      "learning_rate": 3.663178005314275e-05,
      "loss": 1.2864,
      "step": 1361
    },
    {
      "epoch": 4.36,
      "learning_rate": 3.649797336124615e-05,
      "loss": 1.3258,
      "step": 1362
    },
    {
      "epoch": 4.36,
      "learning_rate": 3.636435692342305e-05,
      "loss": 1.1906,
      "step": 1363
    },
    {
      "epoch": 4.36,
      "learning_rate": 3.623093113999244e-05,
      "loss": 1.1597,
      "step": 1364
    },
    {
      "epoch": 4.37,
      "learning_rate": 3.609769641070221e-05,
      "loss": 1.2214,
      "step": 1365
    },
    {
      "epoch": 4.37,
      "learning_rate": 3.5964653134727776e-05,
      "loss": 1.2597,
      "step": 1366
    },
    {
      "epoch": 4.37,
      "learning_rate": 3.583180171067101e-05,
      "loss": 1.078,
      "step": 1367
    },
    {
      "epoch": 4.38,
      "learning_rate": 3.569914253655896e-05,
      "loss": 1.1157,
      "step": 1368
    },
    {
      "epoch": 4.38,
      "learning_rate": 3.556667600984266e-05,
      "loss": 1.2919,
      "step": 1369
    },
    {
      "epoch": 4.38,
      "learning_rate": 3.543440252739606e-05,
      "loss": 1.2627,
      "step": 1370
    },
    {
      "epoch": 4.39,
      "learning_rate": 3.530232248551466e-05,
      "loss": 1.2114,
      "step": 1371
    },
    {
      "epoch": 4.39,
      "learning_rate": 3.517043627991441e-05,
      "loss": 1.3068,
      "step": 1372
    },
    {
      "epoch": 4.39,
      "learning_rate": 3.5038744305730566e-05,
      "loss": 1.0742,
      "step": 1373
    },
    {
      "epoch": 4.4,
      "learning_rate": 3.490724695751642e-05,
      "loss": 1.1632,
      "step": 1374
    },
    {
      "epoch": 4.4,
      "learning_rate": 3.477594462924212e-05,
      "loss": 1.1117,
      "step": 1375
    },
    {
      "epoch": 4.4,
      "learning_rate": 3.4644837714293674e-05,
      "loss": 0.9966,
      "step": 1376
    },
    {
      "epoch": 4.41,
      "learning_rate": 3.45139266054715e-05,
      "loss": 1.1261,
      "step": 1377
    },
    {
      "epoch": 4.41,
      "learning_rate": 3.438321169498939e-05,
      "loss": 1.2275,
      "step": 1378
    },
    {
      "epoch": 4.41,
      "learning_rate": 3.4252693374473346e-05,
      "loss": 1.3216,
      "step": 1379
    },
    {
      "epoch": 4.42,
      "learning_rate": 3.412237203496036e-05,
      "loss": 1.2235,
      "step": 1380
    },
    {
      "epoch": 4.42,
      "learning_rate": 3.399224806689728e-05,
      "loss": 1.1613,
      "step": 1381
    },
    {
      "epoch": 4.42,
      "learning_rate": 3.3862321860139576e-05,
      "loss": 1.0705,
      "step": 1382
    },
    {
      "epoch": 4.43,
      "learning_rate": 3.3732593803950355e-05,
      "loss": 1.3148,
      "step": 1383
    },
    {
      "epoch": 4.43,
      "learning_rate": 3.3603064286998884e-05,
      "loss": 1.1271,
      "step": 1384
    },
    {
      "epoch": 4.43,
      "learning_rate": 3.347373369735971e-05,
      "loss": 1.2326,
      "step": 1385
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.334460242251134e-05,
      "loss": 1.3092,
      "step": 1386
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.3215670849335155e-05,
      "loss": 1.1891,
      "step": 1387
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.308693936411421e-05,
      "loss": 1.3438,
      "step": 1388
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.295840835253206e-05,
      "loss": 1.2838,
      "step": 1389
    },
    {
      "epoch": 4.45,
      "learning_rate": 3.283007819967172e-05,
      "loss": 1.1708,
      "step": 1390
    },
    {
      "epoch": 4.45,
      "learning_rate": 3.270194929001436e-05,
      "loss": 1.3464,
      "step": 1391
    },
    {
      "epoch": 4.45,
      "learning_rate": 3.257402200743821e-05,
      "loss": 1.2723,
      "step": 1392
    },
    {
      "epoch": 4.46,
      "learning_rate": 3.244629673521746e-05,
      "loss": 1.1378,
      "step": 1393
    },
    {
      "epoch": 4.46,
      "learning_rate": 3.231877385602106e-05,
      "loss": 1.1926,
      "step": 1394
    },
    {
      "epoch": 4.46,
      "learning_rate": 3.21914537519115e-05,
      "loss": 1.3834,
      "step": 1395
    },
    {
      "epoch": 4.47,
      "learning_rate": 3.206433680434394e-05,
      "loss": 1.2257,
      "step": 1396
    },
    {
      "epoch": 4.47,
      "learning_rate": 3.19374233941647e-05,
      "loss": 1.2369,
      "step": 1397
    },
    {
      "epoch": 4.47,
      "learning_rate": 3.181071390161037e-05,
      "loss": 1.3084,
      "step": 1398
    },
    {
      "epoch": 4.48,
      "learning_rate": 3.1684208706306574e-05,
      "loss": 1.2545,
      "step": 1399
    },
    {
      "epoch": 4.48,
      "learning_rate": 3.155790818726687e-05,
      "loss": 1.4355,
      "step": 1400
    },
    {
      "epoch": 4.48,
      "learning_rate": 3.1431812722891594e-05,
      "loss": 1.2244,
      "step": 1401
    },
    {
      "epoch": 4.49,
      "learning_rate": 3.130592269096671e-05,
      "loss": 1.4236,
      "step": 1402
    },
    {
      "epoch": 4.49,
      "learning_rate": 3.118023846866276e-05,
      "loss": 1.2158,
      "step": 1403
    },
    {
      "epoch": 4.49,
      "learning_rate": 3.1054760432533624e-05,
      "loss": 1.2444,
      "step": 1404
    },
    {
      "epoch": 4.5,
      "learning_rate": 3.092948895851545e-05,
      "loss": 1.1612,
      "step": 1405
    },
    {
      "epoch": 4.5,
      "learning_rate": 3.0804424421925514e-05,
      "loss": 1.2585,
      "step": 1406
    },
    {
      "epoch": 4.5,
      "learning_rate": 3.0679567197461134e-05,
      "loss": 1.2995,
      "step": 1407
    },
    {
      "epoch": 4.51,
      "learning_rate": 3.055491765919846e-05,
      "loss": 1.2044,
      "step": 1408
    },
    {
      "epoch": 4.51,
      "learning_rate": 3.0430476180591416e-05,
      "loss": 1.2242,
      "step": 1409
    },
    {
      "epoch": 4.51,
      "learning_rate": 3.030624313447067e-05,
      "loss": 1.1604,
      "step": 1410
    },
    {
      "epoch": 4.52,
      "learning_rate": 3.0182218893042303e-05,
      "loss": 1.2403,
      "step": 1411
    },
    {
      "epoch": 4.52,
      "learning_rate": 3.005840382788685e-05,
      "loss": 1.3903,
      "step": 1412
    },
    {
      "epoch": 4.52,
      "learning_rate": 2.9934798309958146e-05,
      "loss": 1.282,
      "step": 1413
    },
    {
      "epoch": 4.52,
      "learning_rate": 2.9811402709582193e-05,
      "loss": 1.0673,
      "step": 1414
    },
    {
      "epoch": 4.53,
      "learning_rate": 2.9688217396456164e-05,
      "loss": 1.1409,
      "step": 1415
    },
    {
      "epoch": 4.53,
      "learning_rate": 2.9565242739647114e-05,
      "loss": 1.1593,
      "step": 1416
    },
    {
      "epoch": 4.53,
      "learning_rate": 2.944247910759097e-05,
      "loss": 1.1304,
      "step": 1417
    },
    {
      "epoch": 4.54,
      "learning_rate": 2.931992686809146e-05,
      "loss": 1.3066,
      "step": 1418
    },
    {
      "epoch": 4.54,
      "learning_rate": 2.9197586388318932e-05,
      "loss": 1.3068,
      "step": 1419
    },
    {
      "epoch": 4.54,
      "learning_rate": 2.9075458034809366e-05,
      "loss": 1.122,
      "step": 1420
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.8953542173463133e-05,
      "loss": 1.1337,
      "step": 1421
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.8831839169543996e-05,
      "loss": 1.1717,
      "step": 1422
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.871034938767798e-05,
      "loss": 1.2355,
      "step": 1423
    },
    {
      "epoch": 4.56,
      "learning_rate": 2.8589073191852302e-05,
      "loss": 1.2026,
      "step": 1424
    },
    {
      "epoch": 4.56,
      "learning_rate": 2.8468010945414303e-05,
      "loss": 1.1116,
      "step": 1425
    },
    {
      "epoch": 4.56,
      "learning_rate": 2.8347163011070254e-05,
      "loss": 1.182,
      "step": 1426
    },
    {
      "epoch": 4.57,
      "learning_rate": 2.8226529750884402e-05,
      "loss": 1.0533,
      "step": 1427
    },
    {
      "epoch": 4.57,
      "learning_rate": 2.8106111526277767e-05,
      "loss": 1.0424,
      "step": 1428
    },
    {
      "epoch": 4.57,
      "learning_rate": 2.798590869802714e-05,
      "loss": 1.2169,
      "step": 1429
    },
    {
      "epoch": 4.58,
      "learning_rate": 2.7865921626264026e-05,
      "loss": 1.2382,
      "step": 1430
    },
    {
      "epoch": 4.58,
      "learning_rate": 2.7746150670473458e-05,
      "loss": 1.4082,
      "step": 1431
    },
    {
      "epoch": 4.58,
      "learning_rate": 2.7626596189492983e-05,
      "loss": 1.1795,
      "step": 1432
    },
    {
      "epoch": 4.59,
      "learning_rate": 2.7507258541511604e-05,
      "loss": 1.1854,
      "step": 1433
    },
    {
      "epoch": 4.59,
      "learning_rate": 2.738813808406866e-05,
      "loss": 1.0491,
      "step": 1434
    },
    {
      "epoch": 4.59,
      "learning_rate": 2.7269235174052788e-05,
      "loss": 1.1658,
      "step": 1435
    },
    {
      "epoch": 4.6,
      "learning_rate": 2.7150550167700906e-05,
      "loss": 1.3358,
      "step": 1436
    },
    {
      "epoch": 4.6,
      "learning_rate": 2.7032083420597e-05,
      "loss": 1.0827,
      "step": 1437
    },
    {
      "epoch": 4.6,
      "learning_rate": 2.6913835287671175e-05,
      "loss": 1.2645,
      "step": 1438
    },
    {
      "epoch": 4.6,
      "learning_rate": 2.6795806123198575e-05,
      "loss": 1.2385,
      "step": 1439
    },
    {
      "epoch": 4.61,
      "learning_rate": 2.667799628079829e-05,
      "loss": 1.2225,
      "step": 1440
    },
    {
      "epoch": 4.61,
      "learning_rate": 2.6560406113432325e-05,
      "loss": 1.2008,
      "step": 1441
    },
    {
      "epoch": 4.61,
      "learning_rate": 2.6443035973404496e-05,
      "loss": 1.2234,
      "step": 1442
    },
    {
      "epoch": 4.62,
      "learning_rate": 2.6325886212359498e-05,
      "loss": 1.0005,
      "step": 1443
    },
    {
      "epoch": 4.62,
      "learning_rate": 2.620895718128169e-05,
      "loss": 1.3141,
      "step": 1444
    },
    {
      "epoch": 4.62,
      "learning_rate": 2.6092249230494126e-05,
      "loss": 1.1985,
      "step": 1445
    },
    {
      "epoch": 4.63,
      "learning_rate": 2.5975762709657504e-05,
      "loss": 1.2794,
      "step": 1446
    },
    {
      "epoch": 4.63,
      "learning_rate": 2.585949796776912e-05,
      "loss": 1.2525,
      "step": 1447
    },
    {
      "epoch": 4.63,
      "learning_rate": 2.5743455353161794e-05,
      "loss": 1.4426,
      "step": 1448
    },
    {
      "epoch": 4.64,
      "learning_rate": 2.562763521350283e-05,
      "loss": 1.1397,
      "step": 1449
    },
    {
      "epoch": 4.64,
      "learning_rate": 2.5512037895793084e-05,
      "loss": 1.1846,
      "step": 1450
    },
    {
      "epoch": 4.64,
      "learning_rate": 2.5396663746365733e-05,
      "loss": 1.1873,
      "step": 1451
    },
    {
      "epoch": 4.65,
      "learning_rate": 2.528151311088537e-05,
      "loss": 1.1061,
      "step": 1452
    },
    {
      "epoch": 4.65,
      "learning_rate": 2.516658633434691e-05,
      "loss": 1.2572,
      "step": 1453
    },
    {
      "epoch": 4.65,
      "learning_rate": 2.5051883761074614e-05,
      "loss": 1.1367,
      "step": 1454
    },
    {
      "epoch": 4.66,
      "learning_rate": 2.4937405734720966e-05,
      "loss": 1.1734,
      "step": 1455
    },
    {
      "epoch": 4.66,
      "learning_rate": 2.482315259826581e-05,
      "loss": 1.2402,
      "step": 1456
    },
    {
      "epoch": 4.66,
      "learning_rate": 2.4709124694015116e-05,
      "loss": 1.1024,
      "step": 1457
    },
    {
      "epoch": 4.67,
      "learning_rate": 2.4595322363600072e-05,
      "loss": 1.2532,
      "step": 1458
    },
    {
      "epoch": 4.67,
      "learning_rate": 2.4481745947976055e-05,
      "loss": 1.1677,
      "step": 1459
    },
    {
      "epoch": 4.67,
      "learning_rate": 2.4368395787421572e-05,
      "loss": 1.2269,
      "step": 1460
    },
    {
      "epoch": 4.68,
      "learning_rate": 2.4255272221537295e-05,
      "loss": 1.1235,
      "step": 1461
    },
    {
      "epoch": 4.68,
      "learning_rate": 2.4142375589244957e-05,
      "loss": 1.2849,
      "step": 1462
    },
    {
      "epoch": 4.68,
      "learning_rate": 2.4029706228786473e-05,
      "loss": 1.119,
      "step": 1463
    },
    {
      "epoch": 4.68,
      "learning_rate": 2.3917264477722788e-05,
      "loss": 1.2696,
      "step": 1464
    },
    {
      "epoch": 4.69,
      "learning_rate": 2.3805050672932928e-05,
      "loss": 1.2948,
      "step": 1465
    },
    {
      "epoch": 4.69,
      "learning_rate": 2.369306515061298e-05,
      "loss": 1.0595,
      "step": 1466
    },
    {
      "epoch": 4.69,
      "learning_rate": 2.3581308246275103e-05,
      "loss": 1.2036,
      "step": 1467
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.3469780294746502e-05,
      "loss": 1.1153,
      "step": 1468
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.3358481630168418e-05,
      "loss": 1.0462,
      "step": 1469
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.324741258599521e-05,
      "loss": 1.0569,
      "step": 1470
    },
    {
      "epoch": 4.71,
      "learning_rate": 2.3136573494993207e-05,
      "loss": 1.2894,
      "step": 1471
    },
    {
      "epoch": 4.71,
      "learning_rate": 2.302596468923981e-05,
      "loss": 1.1393,
      "step": 1472
    },
    {
      "epoch": 4.71,
      "learning_rate": 2.29155865001225e-05,
      "loss": 1.1649,
      "step": 1473
    },
    {
      "epoch": 4.72,
      "learning_rate": 2.2805439258337802e-05,
      "loss": 1.4263,
      "step": 1474
    },
    {
      "epoch": 4.72,
      "learning_rate": 2.2695523293890308e-05,
      "loss": 1.1159,
      "step": 1475
    },
    {
      "epoch": 4.72,
      "learning_rate": 2.2585838936091754e-05,
      "loss": 1.1235,
      "step": 1476
    },
    {
      "epoch": 4.73,
      "learning_rate": 2.247638651355991e-05,
      "loss": 1.209,
      "step": 1477
    },
    {
      "epoch": 4.73,
      "learning_rate": 2.2367166354217695e-05,
      "loss": 1.1779,
      "step": 1478
    },
    {
      "epoch": 4.73,
      "learning_rate": 2.225817878529214e-05,
      "loss": 1.158,
      "step": 1479
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.2149424133313422e-05,
      "loss": 1.1273,
      "step": 1480
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.204090272411393e-05,
      "loss": 1.2305,
      "step": 1481
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.1932614882827197e-05,
      "loss": 1.1411,
      "step": 1482
    },
    {
      "epoch": 4.75,
      "learning_rate": 2.182456093388705e-05,
      "loss": 1.1479,
      "step": 1483
    },
    {
      "epoch": 4.75,
      "learning_rate": 2.1716741201026503e-05,
      "loss": 1.1379,
      "step": 1484
    },
    {
      "epoch": 4.75,
      "learning_rate": 2.1609156007276876e-05,
      "loss": 1.202,
      "step": 1485
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.150180567496681e-05,
      "loss": 1.2302,
      "step": 1486
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.139469052572127e-05,
      "loss": 1.2228,
      "step": 1487
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.1287810880460635e-05,
      "loss": 1.1287,
      "step": 1488
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.1181167059399642e-05,
      "loss": 1.3416,
      "step": 1489
    },
    {
      "epoch": 4.77,
      "learning_rate": 2.1074759382046616e-05,
      "loss": 0.9246,
      "step": 1490
    },
    {
      "epoch": 4.77,
      "learning_rate": 2.0968588167202262e-05,
      "loss": 1.2516,
      "step": 1491
    },
    {
      "epoch": 4.77,
      "learning_rate": 2.0862653732958915e-05,
      "loss": 1.0697,
      "step": 1492
    },
    {
      "epoch": 4.78,
      "learning_rate": 2.0756956396699456e-05,
      "loss": 1.3123,
      "step": 1493
    },
    {
      "epoch": 4.78,
      "learning_rate": 2.0651496475096453e-05,
      "loss": 0.9967,
      "step": 1494
    },
    {
      "epoch": 4.78,
      "learning_rate": 2.0546274284111123e-05,
      "loss": 1.3664,
      "step": 1495
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.044129013899251e-05,
      "loss": 1.2267,
      "step": 1496
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.03365443542764e-05,
      "loss": 1.0337,
      "step": 1497
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.0232037243784475e-05,
      "loss": 1.2871,
      "step": 1498
    },
    {
      "epoch": 4.8,
      "learning_rate": 2.012776912062332e-05,
      "loss": 1.1402,
      "step": 1499
    },
    {
      "epoch": 4.8,
      "learning_rate": 2.0023740297183534e-05,
      "loss": 1.1569,
      "step": 1500
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.9919951085138732e-05,
      "loss": 1.1455,
      "step": 1501
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.981640179544466e-05,
      "loss": 1.2812,
      "step": 1502
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.971309273833828e-05,
      "loss": 1.2593,
      "step": 1503
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.9610024223336787e-05,
      "loss": 1.115,
      "step": 1504
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.9507196559236672e-05,
      "loss": 1.1534,
      "step": 1505
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.940461005411288e-05,
      "loss": 1.2684,
      "step": 1506
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.93022650153178e-05,
      "loss": 1.0638,
      "step": 1507
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.9200161749480406e-05,
      "loss": 1.2762,
      "step": 1508
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 1.3793,
      "step": 1509
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.8996681759571756e-05,
      "loss": 1.2928,
      "step": 1510
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.8895305645133e-05,
      "loss": 1.2253,
      "step": 1511
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.879417252291502e-05,
      "loss": 1.4696,
      "step": 1512
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.869328269591585e-05,
      "loss": 1.327,
      "step": 1513
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.8592636466404544e-05,
      "loss": 1.3288,
      "step": 1514
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.8492234135920462e-05,
      "loss": 1.0862,
      "step": 1515
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.8392076005272096e-05,
      "loss": 1.2503,
      "step": 1516
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.829216237453637e-05,
      "loss": 1.2772,
      "step": 1517
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.8192493543057674e-05,
      "loss": 1.1799,
      "step": 1518
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.809306980944696e-05,
      "loss": 1.2278,
      "step": 1519
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.7993891471580893e-05,
      "loss": 1.3143,
      "step": 1520
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.7894958826600882e-05,
      "loss": 1.2151,
      "step": 1521
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.7796272170912253e-05,
      "loss": 1.1679,
      "step": 1522
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.769783180018334e-05,
      "loss": 1.2284,
      "step": 1523
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.7599638009344566e-05,
      "loss": 1.1237,
      "step": 1524
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.7501691092587668e-05,
      "loss": 1.1196,
      "step": 1525
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.740399134336469e-05,
      "loss": 0.9734,
      "step": 1526
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.730653905438714e-05,
      "loss": 1.1438,
      "step": 1527
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.7209334517625152e-05,
      "loss": 1.2978,
      "step": 1528
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.7112378024306565e-05,
      "loss": 0.9801,
      "step": 1529
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.701566986491614e-05,
      "loss": 1.1764,
      "step": 1530
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.6919210329194533e-05,
      "loss": 1.185,
      "step": 1531
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.6822999706137567e-05,
      "loss": 1.2751,
      "step": 1532
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.672703828399529e-05,
      "loss": 1.189,
      "step": 1533
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.663132635027116e-05,
      "loss": 1.0255,
      "step": 1534
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.653586419172114e-05,
      "loss": 1.0851,
      "step": 1535
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.644065209435284e-05,
      "loss": 1.2234,
      "step": 1536
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.634569034342476e-05,
      "loss": 1.2177,
      "step": 1537
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.6250979223445262e-05,
      "loss": 1.2742,
      "step": 1538
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.6156519018171857e-05,
      "loss": 1.2485,
      "step": 1539
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.606231001061029e-05,
      "loss": 1.258,
      "step": 1540
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.5968352483013738e-05,
      "loss": 1.1914,
      "step": 1541
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.587464671688187e-05,
      "loss": 1.3168,
      "step": 1542
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.578119299296017e-05,
      "loss": 1.1822,
      "step": 1543
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.5687991591238916e-05,
      "loss": 1.2512,
      "step": 1544
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.559504279095244e-05,
      "loss": 1.1833,
      "step": 1545
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.55023468705783e-05,
      "loss": 1.1073,
      "step": 1546
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.5409904107836358e-05,
      "loss": 1.1512,
      "step": 1547
    },
    {
      "epoch": 4.95,
      "learning_rate": 1.5317714779688074e-05,
      "loss": 1.1547,
      "step": 1548
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.5225779162335552e-05,
      "loss": 1.1388,
      "step": 1549
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.5134097531220836e-05,
      "loss": 1.1778,
      "step": 1550
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.5042670161024974e-05,
      "loss": 1.1152,
      "step": 1551
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.495149732566723e-05,
      "loss": 1.2545,
      "step": 1552
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.4860579298304312e-05,
      "loss": 0.9586,
      "step": 1553
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.4769916351329493e-05,
      "loss": 1.3217,
      "step": 1554
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4679508756371807e-05,
      "loss": 1.1042,
      "step": 1555
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4589356784295249e-05,
      "loss": 1.0651,
      "step": 1556
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.4499460705197998e-05,
      "loss": 1.1695,
      "step": 1557
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.4409820788411544e-05,
      "loss": 1.0911,
      "step": 1558
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.432043730249988e-05,
      "loss": 1.3344,
      "step": 1559
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.4231310515258744e-05,
      "loss": 1.2518,
      "step": 1560
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.414244069371482e-05,
      "loss": 1.0974,
      "step": 1561
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.4053828104124867e-05,
      "loss": 1.0685,
      "step": 1562
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.3965473011975038e-05,
      "loss": 1.1913,
      "step": 1563
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.3877375681979943e-05,
      "loss": 1.124,
      "step": 1564
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.378953637808198e-05,
      "loss": 1.0339,
      "step": 1565
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.3701955363450447e-05,
      "loss": 1.1407,
      "step": 1566
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.361463290048085e-05,
      "loss": 1.1691,
      "step": 1567
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.3527569250794014e-05,
      "loss": 1.0583,
      "step": 1568
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.3440764675235384e-05,
      "loss": 1.0202,
      "step": 1569
    },
    {
      "epoch": 5.02,
      "learning_rate": 1.3354219433874227e-05,
      "loss": 1.0051,
      "step": 1570
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.3267933786002796e-05,
      "loss": 1.1353,
      "step": 1571
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.3181907990135622e-05,
      "loss": 1.1874,
      "step": 1572
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.3096142304008696e-05,
      "loss": 1.0563,
      "step": 1573
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.3010636984578706e-05,
      "loss": 1.1368,
      "step": 1574
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.2925392288022298e-05,
      "loss": 1.0399,
      "step": 1575
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.2840408469735243e-05,
      "loss": 1.0432,
      "step": 1576
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.2755685784331783e-05,
      "loss": 1.0724,
      "step": 1577
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.267122448564374e-05,
      "loss": 1.2116,
      "step": 1578
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.2587024826719817e-05,
      "loss": 1.0607,
      "step": 1579
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.2503087059824848e-05,
      "loss": 1.307,
      "step": 1580
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.2419411436439022e-05,
      "loss": 1.2161,
      "step": 1581
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.2335998207257137e-05,
      "loss": 1.1286,
      "step": 1582
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.2252847622187902e-05,
      "loss": 1.1494,
      "step": 1583
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.2169959930353047e-05,
      "loss": 1.0668,
      "step": 1584
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.2087335380086729e-05,
      "loss": 1.0798,
      "step": 1585
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.2004974218934695e-05,
      "loss": 1.0851,
      "step": 1586
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.1922876693653585e-05,
      "loss": 1.1828,
      "step": 1587
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.1841043050210177e-05,
      "loss": 1.1193,
      "step": 1588
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.1759473533780618e-05,
      "loss": 1.2097,
      "step": 1589
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.1678168388749788e-05,
      "loss": 0.924,
      "step": 1590
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.1597127858710455e-05,
      "loss": 0.9915,
      "step": 1591
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.1516352186462586e-05,
      "loss": 1.0905,
      "step": 1592
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.1435841614012666e-05,
      "loss": 1.1279,
      "step": 1593
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.1355596382572886e-05,
      "loss": 1.0474,
      "step": 1594
    },
    {
      "epoch": 5.1,
      "learning_rate": 1.127561673256049e-05,
      "loss": 1.0961,
      "step": 1595
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.1195902903597023e-05,
      "loss": 1.0213,
      "step": 1596
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.1116455134507664e-05,
      "loss": 1.1695,
      "step": 1597
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.1037273663320403e-05,
      "loss": 1.1003,
      "step": 1598
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.0958358727265438e-05,
      "loss": 1.308,
      "step": 1599
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.0879710562774403e-05,
      "loss": 1.2282,
      "step": 1600
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.0801329405479665e-05,
      "loss": 0.9867,
      "step": 1601
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.0723215490213634e-05,
      "loss": 1.1957,
      "step": 1602
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.0645369051008102e-05,
      "loss": 1.1209,
      "step": 1603
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.056779032109343e-05,
      "loss": 1.1652,
      "step": 1604
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.0490479532897946e-05,
      "loss": 1.0627,
      "step": 1605
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.0413436918047193e-05,
      "loss": 1.2177,
      "step": 1606
    },
    {
      "epoch": 5.14,
      "learning_rate": 1.0336662707363287e-05,
      "loss": 1.091,
      "step": 1607
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.026015713086418e-05,
      "loss": 1.0286,
      "step": 1608
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.018392041776297e-05,
      "loss": 1.0867,
      "step": 1609
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.010795279646729e-05,
      "loss": 1.0439,
      "step": 1610
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.003225449457852e-05,
      "loss": 1.1293,
      "step": 1611
    },
    {
      "epoch": 5.16,
      "learning_rate": 9.95682573889114e-06,
      "loss": 1.1013,
      "step": 1612
    },
    {
      "epoch": 5.16,
      "learning_rate": 9.881666755392083e-06,
      "loss": 1.1734,
      "step": 1613
    },
    {
      "epoch": 5.16,
      "learning_rate": 9.806777769260033e-06,
      "loss": 1.0271,
      "step": 1614
    },
    {
      "epoch": 5.17,
      "learning_rate": 9.732159004864772e-06,
      "loss": 1.1255,
      "step": 1615
    },
    {
      "epoch": 5.17,
      "learning_rate": 9.657810685766454e-06,
      "loss": 1.1057,
      "step": 1616
    },
    {
      "epoch": 5.17,
      "learning_rate": 9.583733034714981e-06,
      "loss": 0.99,
      "step": 1617
    },
    {
      "epoch": 5.18,
      "learning_rate": 9.50992627364935e-06,
      "loss": 0.9934,
      "step": 1618
    },
    {
      "epoch": 5.18,
      "learning_rate": 9.436390623696911e-06,
      "loss": 1.139,
      "step": 1619
    },
    {
      "epoch": 5.18,
      "learning_rate": 9.363126305172831e-06,
      "loss": 1.2182,
      "step": 1620
    },
    {
      "epoch": 5.19,
      "learning_rate": 9.290133537579294e-06,
      "loss": 1.3155,
      "step": 1621
    },
    {
      "epoch": 5.19,
      "learning_rate": 9.217412539604942e-06,
      "loss": 1.2178,
      "step": 1622
    },
    {
      "epoch": 5.19,
      "learning_rate": 9.144963529124162e-06,
      "loss": 1.1029,
      "step": 1623
    },
    {
      "epoch": 5.2,
      "learning_rate": 9.072786723196468e-06,
      "loss": 1.0682,
      "step": 1624
    },
    {
      "epoch": 5.2,
      "learning_rate": 9.000882338065875e-06,
      "loss": 1.1096,
      "step": 1625
    },
    {
      "epoch": 5.2,
      "learning_rate": 8.929250589160166e-06,
      "loss": 1.0044,
      "step": 1626
    },
    {
      "epoch": 5.21,
      "learning_rate": 8.857891691090337e-06,
      "loss": 1.1437,
      "step": 1627
    },
    {
      "epoch": 5.21,
      "learning_rate": 8.78680585764987e-06,
      "loss": 0.9624,
      "step": 1628
    },
    {
      "epoch": 5.21,
      "learning_rate": 8.715993301814173e-06,
      "loss": 1.2713,
      "step": 1629
    },
    {
      "epoch": 5.22,
      "learning_rate": 8.645454235739903e-06,
      "loss": 1.0223,
      "step": 1630
    },
    {
      "epoch": 5.22,
      "learning_rate": 8.575188870764328e-06,
      "loss": 0.9563,
      "step": 1631
    },
    {
      "epoch": 5.22,
      "learning_rate": 8.505197417404687e-06,
      "loss": 1.3216,
      "step": 1632
    },
    {
      "epoch": 5.23,
      "learning_rate": 8.435480085357583e-06,
      "loss": 1.0972,
      "step": 1633
    },
    {
      "epoch": 5.23,
      "learning_rate": 8.366037083498335e-06,
      "loss": 1.194,
      "step": 1634
    },
    {
      "epoch": 5.23,
      "learning_rate": 8.296868619880372e-06,
      "loss": 1.2959,
      "step": 1635
    },
    {
      "epoch": 5.24,
      "learning_rate": 8.227974901734569e-06,
      "loss": 1.0076,
      "step": 1636
    },
    {
      "epoch": 5.24,
      "learning_rate": 8.15935613546872e-06,
      "loss": 1.301,
      "step": 1637
    },
    {
      "epoch": 5.24,
      "learning_rate": 8.091012526666796e-06,
      "loss": 0.8579,
      "step": 1638
    },
    {
      "epoch": 5.24,
      "learning_rate": 8.022944280088384e-06,
      "loss": 1.1015,
      "step": 1639
    },
    {
      "epoch": 5.25,
      "learning_rate": 7.955151599668121e-06,
      "loss": 1.1605,
      "step": 1640
    },
    {
      "epoch": 5.25,
      "learning_rate": 7.887634688515e-06,
      "loss": 1.0521,
      "step": 1641
    },
    {
      "epoch": 5.25,
      "learning_rate": 7.820393748911791e-06,
      "loss": 1.163,
      "step": 1642
    },
    {
      "epoch": 5.26,
      "learning_rate": 7.753428982314515e-06,
      "loss": 1.0387,
      "step": 1643
    },
    {
      "epoch": 5.26,
      "learning_rate": 7.686740589351704e-06,
      "loss": 1.2478,
      "step": 1644
    },
    {
      "epoch": 5.26,
      "learning_rate": 7.620328769823892e-06,
      "loss": 1.0635,
      "step": 1645
    },
    {
      "epoch": 5.27,
      "learning_rate": 7.554193722702962e-06,
      "loss": 1.2989,
      "step": 1646
    },
    {
      "epoch": 5.27,
      "learning_rate": 7.488335646131628e-06,
      "loss": 1.0498,
      "step": 1647
    },
    {
      "epoch": 5.27,
      "learning_rate": 7.422754737422744e-06,
      "loss": 1.1006,
      "step": 1648
    },
    {
      "epoch": 5.28,
      "learning_rate": 7.357451193058773e-06,
      "loss": 1.0577,
      "step": 1649
    },
    {
      "epoch": 5.28,
      "learning_rate": 7.292425208691212e-06,
      "loss": 1.0495,
      "step": 1650
    },
    {
      "epoch": 5.28,
      "learning_rate": 7.227676979139953e-06,
      "loss": 1.0042,
      "step": 1651
    },
    {
      "epoch": 5.29,
      "learning_rate": 7.163206698392744e-06,
      "loss": 1.0271,
      "step": 1652
    },
    {
      "epoch": 5.29,
      "learning_rate": 7.0990145596045555e-06,
      "loss": 1.0745,
      "step": 1653
    },
    {
      "epoch": 5.29,
      "learning_rate": 7.035100755097068e-06,
      "loss": 1.0182,
      "step": 1654
    },
    {
      "epoch": 5.3,
      "learning_rate": 6.9714654763580635e-06,
      "loss": 1.0172,
      "step": 1655
    },
    {
      "epoch": 5.3,
      "learning_rate": 6.908108914040823e-06,
      "loss": 1.1802,
      "step": 1656
    },
    {
      "epoch": 5.3,
      "learning_rate": 6.845031257963619e-06,
      "loss": 1.1543,
      "step": 1657
    },
    {
      "epoch": 5.31,
      "learning_rate": 6.782232697109103e-06,
      "loss": 1.1352,
      "step": 1658
    },
    {
      "epoch": 5.31,
      "learning_rate": 6.7197134196237295e-06,
      "loss": 1.0852,
      "step": 1659
    },
    {
      "epoch": 5.31,
      "learning_rate": 6.657473612817244e-06,
      "loss": 1.0451,
      "step": 1660
    },
    {
      "epoch": 5.32,
      "learning_rate": 6.59551346316204e-06,
      "loss": 0.9291,
      "step": 1661
    },
    {
      "epoch": 5.32,
      "learning_rate": 6.533833156292679e-06,
      "loss": 1.0729,
      "step": 1662
    },
    {
      "epoch": 5.32,
      "learning_rate": 6.472432877005341e-06,
      "loss": 1.1238,
      "step": 1663
    },
    {
      "epoch": 5.32,
      "learning_rate": 6.4113128092571864e-06,
      "loss": 1.052,
      "step": 1664
    },
    {
      "epoch": 5.33,
      "learning_rate": 6.350473136165835e-06,
      "loss": 1.222,
      "step": 1665
    },
    {
      "epoch": 5.33,
      "learning_rate": 6.289914040008904e-06,
      "loss": 1.2064,
      "step": 1666
    },
    {
      "epoch": 5.33,
      "learning_rate": 6.229635702223324e-06,
      "loss": 1.1648,
      "step": 1667
    },
    {
      "epoch": 5.34,
      "learning_rate": 6.169638303404912e-06,
      "loss": 1.2552,
      "step": 1668
    },
    {
      "epoch": 5.34,
      "learning_rate": 6.109922023307746e-06,
      "loss": 1.224,
      "step": 1669
    },
    {
      "epoch": 5.34,
      "learning_rate": 6.050487040843711e-06,
      "loss": 1.106,
      "step": 1670
    },
    {
      "epoch": 5.35,
      "learning_rate": 5.991333534081878e-06,
      "loss": 1.1702,
      "step": 1671
    },
    {
      "epoch": 5.35,
      "learning_rate": 5.932461680248014e-06,
      "loss": 1.004,
      "step": 1672
    },
    {
      "epoch": 5.35,
      "learning_rate": 5.8738716557240505e-06,
      "loss": 1.1387,
      "step": 1673
    },
    {
      "epoch": 5.36,
      "learning_rate": 5.8155636360475385e-06,
      "loss": 1.1334,
      "step": 1674
    },
    {
      "epoch": 5.36,
      "learning_rate": 5.757537795911161e-06,
      "loss": 1.0456,
      "step": 1675
    },
    {
      "epoch": 5.36,
      "learning_rate": 5.699794309162132e-06,
      "loss": 1.1552,
      "step": 1676
    },
    {
      "epoch": 5.37,
      "learning_rate": 5.6423333488018095e-06,
      "loss": 1.1616,
      "step": 1677
    },
    {
      "epoch": 5.37,
      "learning_rate": 5.5851550869850055e-06,
      "loss": 1.0576,
      "step": 1678
    },
    {
      "epoch": 5.37,
      "learning_rate": 5.5282596950196e-06,
      "loss": 1.1009,
      "step": 1679
    },
    {
      "epoch": 5.38,
      "learning_rate": 5.471647343365982e-06,
      "loss": 1.057,
      "step": 1680
    },
    {
      "epoch": 5.38,
      "learning_rate": 5.4153182016365435e-06,
      "loss": 1.071,
      "step": 1681
    },
    {
      "epoch": 5.38,
      "learning_rate": 5.359272438595153e-06,
      "loss": 1.1713,
      "step": 1682
    },
    {
      "epoch": 5.39,
      "learning_rate": 5.303510222156716e-06,
      "loss": 0.9717,
      "step": 1683
    },
    {
      "epoch": 5.39,
      "learning_rate": 5.2480317193865705e-06,
      "loss": 1.0091,
      "step": 1684
    },
    {
      "epoch": 5.39,
      "learning_rate": 5.192837096500058e-06,
      "loss": 0.9723,
      "step": 1685
    },
    {
      "epoch": 5.4,
      "learning_rate": 5.137926518862013e-06,
      "loss": 1.111,
      "step": 1686
    },
    {
      "epoch": 5.4,
      "learning_rate": 5.083300150986259e-06,
      "loss": 1.1653,
      "step": 1687
    },
    {
      "epoch": 5.4,
      "learning_rate": 5.028958156535113e-06,
      "loss": 1.048,
      "step": 1688
    },
    {
      "epoch": 5.4,
      "learning_rate": 4.974900698318885e-06,
      "loss": 1.1723,
      "step": 1689
    },
    {
      "epoch": 5.41,
      "learning_rate": 4.921127938295456e-06,
      "loss": 1.1396,
      "step": 1690
    },
    {
      "epoch": 5.41,
      "learning_rate": 4.867640037569698e-06,
      "loss": 1.1174,
      "step": 1691
    },
    {
      "epoch": 5.41,
      "learning_rate": 4.8144371563930476e-06,
      "loss": 1.1017,
      "step": 1692
    },
    {
      "epoch": 5.42,
      "learning_rate": 4.76151945416301e-06,
      "loss": 1.0861,
      "step": 1693
    },
    {
      "epoch": 5.42,
      "learning_rate": 4.708887089422686e-06,
      "loss": 1.0885,
      "step": 1694
    },
    {
      "epoch": 5.42,
      "learning_rate": 4.656540219860317e-06,
      "loss": 1.146,
      "step": 1695
    },
    {
      "epoch": 5.43,
      "learning_rate": 4.604479002308737e-06,
      "loss": 1.3095,
      "step": 1696
    },
    {
      "epoch": 5.43,
      "learning_rate": 4.552703592745033e-06,
      "loss": 1.0145,
      "step": 1697
    },
    {
      "epoch": 5.43,
      "learning_rate": 4.501214146289956e-06,
      "loss": 1.2758,
      "step": 1698
    },
    {
      "epoch": 5.44,
      "learning_rate": 4.450010817207495e-06,
      "loss": 1.2866,
      "step": 1699
    },
    {
      "epoch": 5.44,
      "learning_rate": 4.3990937589044605e-06,
      "loss": 0.9988,
      "step": 1700
    },
    {
      "epoch": 5.44,
      "learning_rate": 4.3484631239299356e-06,
      "loss": 1.0628,
      "step": 1701
    },
    {
      "epoch": 5.45,
      "learning_rate": 4.298119063974914e-06,
      "loss": 1.2719,
      "step": 1702
    },
    {
      "epoch": 5.45,
      "learning_rate": 4.2480617298718e-06,
      "loss": 1.1494,
      "step": 1703
    },
    {
      "epoch": 5.45,
      "learning_rate": 4.198291271593924e-06,
      "loss": 1.0897,
      "step": 1704
    },
    {
      "epoch": 5.46,
      "learning_rate": 4.148807838255164e-06,
      "loss": 1.2933,
      "step": 1705
    },
    {
      "epoch": 5.46,
      "learning_rate": 4.09961157810943e-06,
      "loss": 1.1768,
      "step": 1706
    },
    {
      "epoch": 5.46,
      "learning_rate": 4.050702638550275e-06,
      "loss": 1.1135,
      "step": 1707
    },
    {
      "epoch": 5.47,
      "learning_rate": 4.0020811661104165e-06,
      "loss": 1.1571,
      "step": 1708
    },
    {
      "epoch": 5.47,
      "learning_rate": 3.953747306461331e-06,
      "loss": 0.9912,
      "step": 1709
    },
    {
      "epoch": 5.47,
      "learning_rate": 3.9057012044127815e-06,
      "loss": 1.1516,
      "step": 1710
    },
    {
      "epoch": 5.48,
      "learning_rate": 3.857943003912401e-06,
      "loss": 1.1868,
      "step": 1711
    },
    {
      "epoch": 5.48,
      "learning_rate": 3.810472848045266e-06,
      "loss": 1.1438,
      "step": 1712
    },
    {
      "epoch": 5.48,
      "learning_rate": 3.7632908790334655e-06,
      "loss": 0.9461,
      "step": 1713
    },
    {
      "epoch": 5.48,
      "learning_rate": 3.716397238235636e-06,
      "loss": 1.1444,
      "step": 1714
    },
    {
      "epoch": 5.49,
      "learning_rate": 3.6697920661466487e-06,
      "loss": 1.1059,
      "step": 1715
    },
    {
      "epoch": 5.49,
      "learning_rate": 3.6234755023970446e-06,
      "loss": 1.2671,
      "step": 1716
    },
    {
      "epoch": 5.49,
      "learning_rate": 3.5774476857527107e-06,
      "loss": 1.11,
      "step": 1717
    },
    {
      "epoch": 5.5,
      "learning_rate": 3.5317087541144377e-06,
      "loss": 0.9635,
      "step": 1718
    },
    {
      "epoch": 5.5,
      "learning_rate": 3.4862588445174984e-06,
      "loss": 1.1287,
      "step": 1719
    },
    {
      "epoch": 5.5,
      "learning_rate": 3.4410980931312564e-06,
      "loss": 1.2909,
      "step": 1720
    },
    {
      "epoch": 5.51,
      "learning_rate": 3.396226635258759e-06,
      "loss": 1.1126,
      "step": 1721
    },
    {
      "epoch": 5.51,
      "learning_rate": 3.3516446053363015e-06,
      "loss": 1.2494,
      "step": 1722
    },
    {
      "epoch": 5.51,
      "learning_rate": 3.3073521369330617e-06,
      "loss": 1.0905,
      "step": 1723
    },
    {
      "epoch": 5.52,
      "learning_rate": 3.263349362750645e-06,
      "loss": 1.2346,
      "step": 1724
    },
    {
      "epoch": 5.52,
      "learning_rate": 3.2196364146227507e-06,
      "loss": 1.0789,
      "step": 1725
    },
    {
      "epoch": 5.52,
      "learning_rate": 3.1762134235147846e-06,
      "loss": 1.0356,
      "step": 1726
    },
    {
      "epoch": 5.53,
      "learning_rate": 3.133080519523368e-06,
      "loss": 1.109,
      "step": 1727
    },
    {
      "epoch": 5.53,
      "learning_rate": 3.090237831876053e-06,
      "loss": 1.1461,
      "step": 1728
    },
    {
      "epoch": 5.53,
      "learning_rate": 3.047685488930874e-06,
      "loss": 0.8361,
      "step": 1729
    },
    {
      "epoch": 5.54,
      "learning_rate": 3.005423618175984e-06,
      "loss": 1.0393,
      "step": 1730
    },
    {
      "epoch": 5.54,
      "learning_rate": 2.9634523462293005e-06,
      "loss": 1.1356,
      "step": 1731
    },
    {
      "epoch": 5.54,
      "learning_rate": 2.921771798838069e-06,
      "loss": 0.9565,
      "step": 1732
    },
    {
      "epoch": 5.55,
      "learning_rate": 2.8803821008785204e-06,
      "loss": 1.239,
      "step": 1733
    },
    {
      "epoch": 5.55,
      "learning_rate": 2.839283376355506e-06,
      "loss": 1.1006,
      "step": 1734
    },
    {
      "epoch": 5.55,
      "learning_rate": 2.798475748402074e-06,
      "loss": 0.978,
      "step": 1735
    },
    {
      "epoch": 5.56,
      "learning_rate": 2.7579593392791813e-06,
      "loss": 1.1736,
      "step": 1736
    },
    {
      "epoch": 5.56,
      "learning_rate": 2.717734270375272e-06,
      "loss": 1.0379,
      "step": 1737
    },
    {
      "epoch": 5.56,
      "learning_rate": 2.677800662205909e-06,
      "loss": 1.2612,
      "step": 1738
    },
    {
      "epoch": 5.56,
      "learning_rate": 2.6381586344134458e-06,
      "loss": 1.0783,
      "step": 1739
    },
    {
      "epoch": 5.57,
      "learning_rate": 2.5988083057666533e-06,
      "loss": 1.2208,
      "step": 1740
    },
    {
      "epoch": 5.57,
      "learning_rate": 2.5597497941603487e-06,
      "loss": 1.026,
      "step": 1741
    },
    {
      "epoch": 5.57,
      "learning_rate": 2.520983216615047e-06,
      "loss": 1.0977,
      "step": 1742
    },
    {
      "epoch": 5.58,
      "learning_rate": 2.482508689276675e-06,
      "loss": 1.0289,
      "step": 1743
    },
    {
      "epoch": 5.58,
      "learning_rate": 2.4443263274161023e-06,
      "loss": 0.988,
      "step": 1744
    },
    {
      "epoch": 5.58,
      "learning_rate": 2.4064362454289002e-06,
      "loss": 0.952,
      "step": 1745
    },
    {
      "epoch": 5.59,
      "learning_rate": 2.3688385568349515e-06,
      "loss": 1.2469,
      "step": 1746
    },
    {
      "epoch": 5.59,
      "learning_rate": 2.3315333742780942e-06,
      "loss": 1.1064,
      "step": 1747
    },
    {
      "epoch": 5.59,
      "learning_rate": 2.2945208095258573e-06,
      "loss": 0.9623,
      "step": 1748
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.2578009734690265e-06,
      "loss": 1.1957,
      "step": 1749
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.2213739761214104e-06,
      "loss": 1.1854,
      "step": 1750
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.1852399266194314e-06,
      "loss": 1.0602,
      "step": 1751
    },
    {
      "epoch": 5.61,
      "learning_rate": 2.1493989332218468e-06,
      "loss": 1.0628,
      "step": 1752
    },
    {
      "epoch": 5.61,
      "learning_rate": 2.1138511033093944e-06,
      "loss": 1.0294,
      "step": 1753
    },
    {
      "epoch": 5.61,
      "learning_rate": 2.0785965433845036e-06,
      "loss": 1.0544,
      "step": 1754
    },
    {
      "epoch": 5.62,
      "learning_rate": 2.043635359070928e-06,
      "loss": 1.0697,
      "step": 1755
    },
    {
      "epoch": 5.62,
      "learning_rate": 2.0089676551134917e-06,
      "loss": 0.973,
      "step": 1756
    },
    {
      "epoch": 5.62,
      "learning_rate": 1.974593535377722e-06,
      "loss": 1.0353,
      "step": 1757
    },
    {
      "epoch": 5.63,
      "learning_rate": 1.9405131028495836e-06,
      "loss": 1.1184,
      "step": 1758
    },
    {
      "epoch": 5.63,
      "learning_rate": 1.9067264596351109e-06,
      "loss": 1.0999,
      "step": 1759
    },
    {
      "epoch": 5.63,
      "learning_rate": 1.8732337069601425e-06,
      "loss": 1.0941,
      "step": 1760
    },
    {
      "epoch": 5.64,
      "learning_rate": 1.8400349451700438e-06,
      "loss": 1.3801,
      "step": 1761
    },
    {
      "epoch": 5.64,
      "learning_rate": 1.8071302737293295e-06,
      "loss": 1.3229,
      "step": 1762
    },
    {
      "epoch": 5.64,
      "learning_rate": 1.7745197912214517e-06,
      "loss": 1.0711,
      "step": 1763
    },
    {
      "epoch": 5.64,
      "learning_rate": 1.742203595348435e-06,
      "loss": 1.0437,
      "step": 1764
    },
    {
      "epoch": 5.65,
      "learning_rate": 1.71018178293062e-06,
      "loss": 1.1303,
      "step": 1765
    },
    {
      "epoch": 5.65,
      "learning_rate": 1.6784544499063637e-06,
      "loss": 1.1888,
      "step": 1766
    },
    {
      "epoch": 5.65,
      "learning_rate": 1.6470216913317626e-06,
      "loss": 1.197,
      "step": 1767
    },
    {
      "epoch": 5.66,
      "learning_rate": 1.6158836013803525e-06,
      "loss": 1.191,
      "step": 1768
    },
    {
      "epoch": 5.66,
      "learning_rate": 1.5850402733428194e-06,
      "loss": 1.1267,
      "step": 1769
    },
    {
      "epoch": 5.66,
      "learning_rate": 1.554491799626756e-06,
      "loss": 1.206,
      "step": 1770
    },
    {
      "epoch": 5.67,
      "learning_rate": 1.5242382717563619e-06,
      "loss": 1.1139,
      "step": 1771
    },
    {
      "epoch": 5.67,
      "learning_rate": 1.4942797803721543e-06,
      "loss": 1.0769,
      "step": 1772
    },
    {
      "epoch": 5.67,
      "learning_rate": 1.4646164152307018e-06,
      "loss": 1.4499,
      "step": 1773
    },
    {
      "epoch": 5.68,
      "learning_rate": 1.4352482652044031e-06,
      "loss": 1.0935,
      "step": 1774
    },
    {
      "epoch": 5.68,
      "learning_rate": 1.4061754182811414e-06,
      "loss": 1.136,
      "step": 1775
    },
    {
      "epoch": 5.68,
      "learning_rate": 1.3773979615640975e-06,
      "loss": 1.1535,
      "step": 1776
    },
    {
      "epoch": 5.69,
      "learning_rate": 1.348915981271437e-06,
      "loss": 1.0039,
      "step": 1777
    },
    {
      "epoch": 5.69,
      "learning_rate": 1.3207295627360672e-06,
      "loss": 1.0813,
      "step": 1778
    },
    {
      "epoch": 5.69,
      "learning_rate": 1.292838790405393e-06,
      "loss": 1.0042,
      "step": 1779
    },
    {
      "epoch": 5.7,
      "learning_rate": 1.2652437478410272e-06,
      "loss": 1.0996,
      "step": 1780
    },
    {
      "epoch": 5.7,
      "learning_rate": 1.2379445177186143e-06,
      "loss": 1.0002,
      "step": 1781
    },
    {
      "epoch": 5.7,
      "learning_rate": 1.2109411818274852e-06,
      "loss": 1.171,
      "step": 1782
    },
    {
      "epoch": 5.71,
      "learning_rate": 1.1842338210705017e-06,
      "loss": 1.0376,
      "step": 1783
    },
    {
      "epoch": 5.71,
      "learning_rate": 1.157822515463758e-06,
      "loss": 1.0963,
      "step": 1784
    },
    {
      "epoch": 5.71,
      "learning_rate": 1.1317073441363457e-06,
      "loss": 1.1816,
      "step": 1785
    },
    {
      "epoch": 5.72,
      "learning_rate": 1.105888385330156e-06,
      "loss": 1.1051,
      "step": 1786
    },
    {
      "epoch": 5.72,
      "learning_rate": 1.0803657163995895e-06,
      "loss": 0.9638,
      "step": 1787
    },
    {
      "epoch": 5.72,
      "learning_rate": 1.055139413811379e-06,
      "loss": 1.0063,
      "step": 1788
    },
    {
      "epoch": 5.72,
      "learning_rate": 1.0302095531443013e-06,
      "loss": 1.0294,
      "step": 1789
    },
    {
      "epoch": 5.73,
      "learning_rate": 1.005576209089032e-06,
      "loss": 1.1779,
      "step": 1790
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.812394554478355e-07,
      "loss": 1.0854,
      "step": 1791
    },
    {
      "epoch": 5.73,
      "learning_rate": 9.57199365134387e-07,
      "loss": 1.1852,
      "step": 1792
    },
    {
      "epoch": 5.74,
      "learning_rate": 9.334560101735612e-07,
      "loss": 1.1297,
      "step": 1793
    },
    {
      "epoch": 5.74,
      "learning_rate": 9.10009461701189e-07,
      "loss": 1.1593,
      "step": 1794
    },
    {
      "epoch": 5.74,
      "learning_rate": 8.868597899638898e-07,
      "loss": 1.3273,
      "step": 1795
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.640070643187836e-07,
      "loss": 1.0802,
      "step": 1796
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.41451353233369e-07,
      "loss": 0.9034,
      "step": 1797
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.19192724285267e-07,
      "loss": 1.108,
      "step": 1798
    },
    {
      "epoch": 5.76,
      "learning_rate": 7.972312441620444e-07,
      "loss": 1.0961,
      "step": 1799
    },
    {
      "epoch": 5.76,
      "learning_rate": 7.755669786609687e-07,
      "loss": 1.0567,
      "step": 1800
    },
    {
      "epoch": 5.76,
      "learning_rate": 7.541999926888644e-07,
      "loss": 1.1284,
      "step": 1801
    },
    {
      "epoch": 5.77,
      "learning_rate": 7.331303502618903e-07,
      "loss": 1.1834,
      "step": 1802
    },
    {
      "epoch": 5.77,
      "learning_rate": 7.123581145053848e-07,
      "loss": 0.9692,
      "step": 1803
    },
    {
      "epoch": 5.77,
      "learning_rate": 6.91883347653599e-07,
      "loss": 1.1749,
      "step": 1804
    },
    {
      "epoch": 5.78,
      "learning_rate": 6.717061110495859e-07,
      "loss": 1.1236,
      "step": 1805
    },
    {
      "epoch": 5.78,
      "learning_rate": 6.518264651449779e-07,
      "loss": 0.9471,
      "step": 1806
    },
    {
      "epoch": 5.78,
      "learning_rate": 6.322444694998319e-07,
      "loss": 0.9857,
      "step": 1807
    },
    {
      "epoch": 5.79,
      "learning_rate": 6.129601827824183e-07,
      "loss": 1.4305,
      "step": 1808
    },
    {
      "epoch": 5.79,
      "learning_rate": 5.939736627690539e-07,
      "loss": 1.1852,
      "step": 1809
    },
    {
      "epoch": 5.79,
      "learning_rate": 5.752849663439697e-07,
      "loss": 1.0886,
      "step": 1810
    },
    {
      "epoch": 5.8,
      "learning_rate": 5.568941494990876e-07,
      "loss": 1.2561,
      "step": 1811
    },
    {
      "epoch": 5.8,
      "learning_rate": 5.388012673338661e-07,
      "loss": 1.133,
      "step": 1812
    },
    {
      "epoch": 5.8,
      "learning_rate": 5.210063740551552e-07,
      "loss": 1.154,
      "step": 1813
    },
    {
      "epoch": 5.8,
      "learning_rate": 5.035095229770192e-07,
      "loss": 1.1475,
      "step": 1814
    },
    {
      "epoch": 5.81,
      "learning_rate": 4.863107665205701e-07,
      "loss": 1.1182,
      "step": 1815
    },
    {
      "epoch": 5.81,
      "learning_rate": 4.694101562138231e-07,
      "loss": 1.3077,
      "step": 1816
    },
    {
      "epoch": 5.81,
      "learning_rate": 4.5280774269154115e-07,
      "loss": 1.0226,
      "step": 1817
    },
    {
      "epoch": 5.82,
      "learning_rate": 4.365035756950797e-07,
      "loss": 1.0565,
      "step": 1818
    },
    {
      "epoch": 5.82,
      "learning_rate": 4.204977040722424e-07,
      "loss": 1.1543,
      "step": 1819
    },
    {
      "epoch": 5.82,
      "learning_rate": 4.047901757771144e-07,
      "loss": 1.1388,
      "step": 1820
    },
    {
      "epoch": 5.83,
      "learning_rate": 3.893810378699514e-07,
      "loss": 1.0846,
      "step": 1821
    },
    {
      "epoch": 5.83,
      "learning_rate": 3.742703365170241e-07,
      "loss": 1.2273,
      "step": 1822
    },
    {
      "epoch": 5.83,
      "learning_rate": 3.5945811699045205e-07,
      "loss": 1.0894,
      "step": 1823
    },
    {
      "epoch": 5.84,
      "learning_rate": 3.449444236681254e-07,
      "loss": 1.0088,
      "step": 1824
    },
    {
      "epoch": 5.84,
      "learning_rate": 3.3072930003351656e-07,
      "loss": 1.1068,
      "step": 1825
    },
    {
      "epoch": 5.84,
      "learning_rate": 3.1681278867560226e-07,
      "loss": 0.97,
      "step": 1826
    },
    {
      "epoch": 5.85,
      "learning_rate": 3.0319493128866396e-07,
      "loss": 1.1623,
      "step": 1827
    },
    {
      "epoch": 5.85,
      "learning_rate": 2.898757686722542e-07,
      "loss": 1.0805,
      "step": 1828
    },
    {
      "epoch": 5.85,
      "learning_rate": 2.7685534073099705e-07,
      "loss": 1.0928,
      "step": 1829
    },
    {
      "epoch": 5.86,
      "learning_rate": 2.6413368647449923e-07,
      "loss": 1.1096,
      "step": 1830
    },
    {
      "epoch": 5.86,
      "learning_rate": 2.517108440172389e-07,
      "loss": 1.1753,
      "step": 1831
    },
    {
      "epoch": 5.86,
      "learning_rate": 2.395868505784438e-07,
      "loss": 1.1579,
      "step": 1832
    },
    {
      "epoch": 5.87,
      "learning_rate": 2.2776174248199112e-07,
      "loss": 1.1085,
      "step": 1833
    },
    {
      "epoch": 5.87,
      "learning_rate": 2.1623555515626337e-07,
      "loss": 0.9672,
      "step": 1834
    },
    {
      "epoch": 5.87,
      "learning_rate": 2.0500832313409269e-07,
      "loss": 1.0313,
      "step": 1835
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.9408008005260548e-07,
      "loss": 1.0946,
      "step": 1836
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.83450858653178e-07,
      "loss": 1.3363,
      "step": 1837
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.73120690781281e-07,
      "loss": 1.1878,
      "step": 1838
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.630896073864352e-07,
      "loss": 1.0907,
      "step": 1839
    },
    {
      "epoch": 5.89,
      "learning_rate": 1.5335763852206698e-07,
      "loss": 1.2107,
      "step": 1840
    },
    {
      "epoch": 5.89,
      "learning_rate": 1.4392481334547514e-07,
      "loss": 1.1802,
      "step": 1841
    },
    {
      "epoch": 5.89,
      "learning_rate": 1.3479116011769767e-07,
      "loss": 1.1346,
      "step": 1842
    },
    {
      "epoch": 5.9,
      "learning_rate": 1.2595670620344502e-07,
      "loss": 1.207,
      "step": 1843
    },
    {
      "epoch": 5.9,
      "learning_rate": 1.1742147807101145e-07,
      "loss": 1.1664,
      "step": 1844
    },
    {
      "epoch": 5.9,
      "learning_rate": 1.0918550129223048e-07,
      "loss": 1.1745,
      "step": 1845
    },
    {
      "epoch": 5.91,
      "learning_rate": 1.0124880054233066e-07,
      "loss": 0.9993,
      "step": 1846
    },
    {
      "epoch": 5.91,
      "learning_rate": 9.361139959993549e-08,
      "loss": 1.0658,
      "step": 1847
    },
    {
      "epoch": 5.91,
      "learning_rate": 8.627332134690802e-08,
      "loss": 1.1743,
      "step": 1848
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.923458776839531e-08,
      "loss": 1.2076,
      "step": 1849
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.249521995263964e-08,
      "loss": 1.0994,
      "step": 1850
    },
    {
      "epoch": 5.92,
      "learning_rate": 6.605523809102287e-08,
      "loss": 1.1313,
      "step": 1851
    },
    {
      "epoch": 5.93,
      "learning_rate": 5.991466147791113e-08,
      "loss": 1.1256,
      "step": 1852
    },
    {
      "epoch": 5.93,
      "learning_rate": 5.407350851067694e-08,
      "loss": 0.8941,
      "step": 1853
    },
    {
      "epoch": 5.93,
      "learning_rate": 4.853179668959928e-08,
      "loss": 0.995,
      "step": 1854
    },
    {
      "epoch": 5.94,
      "learning_rate": 4.328954261781926e-08,
      "loss": 1.0701,
      "step": 1855
    },
    {
      "epoch": 5.94,
      "learning_rate": 3.834676200129561e-08,
      "loss": 1.0905,
      "step": 1856
    },
    {
      "epoch": 5.94,
      "learning_rate": 3.370346964876036e-08,
      "loss": 1.0473,
      "step": 1857
    },
    {
      "epoch": 5.95,
      "learning_rate": 2.935967947165219e-08,
      "loss": 1.2042,
      "step": 1858
    },
    {
      "epoch": 5.95,
      "learning_rate": 2.5315404484105298e-08,
      "loss": 1.1862,
      "step": 1859
    },
    {
      "epoch": 5.95,
      "learning_rate": 2.1570656802905043e-08,
      "loss": 1.0005,
      "step": 1860
    },
    {
      "epoch": 5.96,
      "learning_rate": 1.81254476474213e-08,
      "loss": 0.9977,
      "step": 1861
    },
    {
      "epoch": 5.96,
      "learning_rate": 1.4979787339619578e-08,
      "loss": 1.321,
      "step": 1862
    },
    {
      "epoch": 5.96,
      "learning_rate": 1.2133685303994391e-08,
      "loss": 1.0475,
      "step": 1863
    },
    {
      "epoch": 5.96,
      "learning_rate": 9.58715006755817e-09,
      "loss": 1.089,
      "step": 1864
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.340189259807951e-09,
      "loss": 0.9875,
      "step": 1865
    },
    {
      "epoch": 5.97,
      "learning_rate": 5.3928096127031644e-09,
      "loss": 1.2451,
      "step": 1866
    },
    {
      "epoch": 5.97,
      "learning_rate": 3.745016960665648e-09,
      "loss": 0.9982,
      "step": 1867
    },
    {
      "epoch": 5.98,
      "learning_rate": 2.3968162405352266e-09,
      "loss": 1.0194,
      "step": 1868
    },
    {
      "epoch": 5.98,
      "learning_rate": 1.3482114915475131e-09,
      "loss": 1.1749,
      "step": 1869
    },
    {
      "epoch": 5.98,
      "learning_rate": 5.992058553672131e-10,
      "loss": 1.2044,
      "step": 1870
    },
    {
      "epoch": 5.99,
      "learning_rate": 1.498015760437177e-10,
      "loss": 1.0756,
      "step": 1871
    },
    {
      "epoch": 5.99,
      "learning_rate": 0.0,
      "loss": 1.197,
      "step": 1872
    },
    {
      "epoch": 5.99,
      "step": 1872,
      "total_flos": 1.9118475836810854e+17,
      "train_loss": 1.3473664614825676,
      "train_runtime": 11846.7676,
      "train_samples_per_second": 2.532,
      "train_steps_per_second": 0.158
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1872,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 50000,
  "total_flos": 1.9118475836810854e+17,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}