{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 1100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 6.060606060606061e-06,
      "loss": 1.2458,
      "step": 1
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 1.0751,
      "step": 2
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 1.1931,
      "step": 3
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.4242424242424244e-05,
      "loss": 1.2385,
      "step": 4
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.0303030303030306e-05,
      "loss": 1.1864,
      "step": 5
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 1.1336,
      "step": 6
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.242424242424243e-05,
      "loss": 1.0835,
      "step": 7
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.848484848484849e-05,
      "loss": 1.1527,
      "step": 8
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.4545454545454546e-05,
      "loss": 1.103,
      "step": 9
    },
    {
      "epoch": 0.09,
      "learning_rate": 6.060606060606061e-05,
      "loss": 1.0819,
      "step": 10
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.666666666666667e-05,
      "loss": 1.1467,
      "step": 11
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.272727272727273e-05,
      "loss": 1.0989,
      "step": 12
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.878787878787879e-05,
      "loss": 1.0974,
      "step": 13
    },
    {
      "epoch": 0.13,
      "learning_rate": 8.484848484848486e-05,
      "loss": 1.0698,
      "step": 14
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.090909090909092e-05,
      "loss": 1.0169,
      "step": 15
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.696969696969698e-05,
      "loss": 0.9964,
      "step": 16
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00010303030303030303,
      "loss": 1.0856,
      "step": 17
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00010909090909090909,
      "loss": 1.0507,
      "step": 18
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00011515151515151516,
      "loss": 1.007,
      "step": 19
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00012121212121212122,
      "loss": 1.1084,
      "step": 20
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00012727272727272728,
      "loss": 1.0358,
      "step": 21
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00013333333333333334,
      "loss": 1.0582,
      "step": 22
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001393939393939394,
      "loss": 1.0544,
      "step": 23
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00014545454545454546,
      "loss": 1.0535,
      "step": 24
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00015151515151515152,
      "loss": 1.0582,
      "step": 25
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00015757575757575757,
      "loss": 1.0906,
      "step": 26
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00016363636363636366,
      "loss": 1.0108,
      "step": 27
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00016969696969696972,
      "loss": 1.0323,
      "step": 28
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00017575757575757578,
      "loss": 1.0126,
      "step": 29
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00018181818181818183,
      "loss": 1.0156,
      "step": 30
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001878787878787879,
      "loss": 1.0324,
      "step": 31
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019393939393939395,
      "loss": 1.0257,
      "step": 32
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002,
      "loss": 1.028,
      "step": 33
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019999956654841327,
      "loss": 1.0171,
      "step": 34
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001999982661974106,
      "loss": 1.0172,
      "step": 35
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019999609895826481,
      "loss": 1.0059,
      "step": 36
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019999306484976375,
      "loss": 1.0053,
      "step": 37
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019998916389821024,
      "loss": 1.0573,
      "step": 38
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001999843961374217,
      "loss": 1.0424,
      "step": 39
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019997876160873006,
      "loss": 1.0238,
      "step": 40
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019997226036098118,
      "loss": 1.0396,
      "step": 41
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001999648924505346,
      "loss": 1.0538,
      "step": 42
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019995665794126296,
      "loss": 1.0093,
      "step": 43
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019994755690455152,
      "loss": 1.0355,
      "step": 44
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019993758941929735,
      "loss": 1.0471,
      "step": 45
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019992675557190906,
      "loss": 1.0135,
      "step": 46
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019991505545630548,
      "loss": 1.011,
      "step": 47
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019990248917391536,
      "loss": 1.0681,
      "step": 48
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019988905683367616,
      "loss": 1.0026,
      "step": 49
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019987475855203326,
      "loss": 1.0069,
      "step": 50
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019985959445293895,
      "loss": 0.9912,
      "step": 51
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019984356466785127,
      "loss": 1.0128,
      "step": 52
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019982666933573295,
      "loss": 0.9888,
      "step": 53
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019980890860305013,
      "loss": 0.9872,
      "step": 54
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019979028262377118,
      "loss": 1.0294,
      "step": 55
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0001997707915593653,
      "loss": 0.9813,
      "step": 56
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019975043557880117,
      "loss": 1.0589,
      "step": 57
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019972921485854542,
      "loss": 1.0116,
      "step": 58
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019970712958256113,
      "loss": 1.0192,
      "step": 59
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00019968417994230625,
      "loss": 1.0511,
      "step": 60
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.000199660366136732,
      "loss": 0.9901,
      "step": 61
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00019963568837228095,
      "loss": 1.0191,
      "step": 62
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00019961014686288545,
      "loss": 1.0461,
      "step": 63
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019958374182996568,
      "loss": 1.0783,
      "step": 64
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00019955647350242765,
      "loss": 1.0415,
      "step": 65
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001995283421166614,
      "loss": 1.0162,
      "step": 66
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00019949934791653882,
      "loss": 1.0578,
      "step": 67
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0001994694911534115,
      "loss": 0.9896,
      "step": 68
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0001994387720861087,
      "loss": 1.0177,
      "step": 69
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00019940719098093497,
      "loss": 0.9988,
      "step": 70
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00019937474811166793,
      "loss": 1.0414,
      "step": 71
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0001993414437595559,
      "loss": 0.9831,
      "step": 72
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00019930727821331523,
      "loss": 1.0807,
      "step": 73
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019927225176912824,
      "loss": 0.9917,
      "step": 74
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00019923636473064024,
      "loss": 1.0385,
      "step": 75
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019919961740895708,
      "loss": 0.9846,
      "step": 76
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00019916201012264254,
      "loss": 0.9663,
      "step": 77
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0001991235431977153,
      "loss": 1.0256,
      "step": 78
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00019908421696764636,
      "loss": 0.9896,
      "step": 79
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00019904403177335608,
      "loss": 1.0321,
      "step": 80
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019900298796321123,
      "loss": 1.0409,
      "step": 81
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001989610858930218,
      "loss": 1.0255,
      "step": 82
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00019891832592603827,
      "loss": 1.0539,
      "step": 83
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019887470843294808,
      "loss": 1.0096,
      "step": 84
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00019883023379187265,
      "loss": 0.9989,
      "step": 85
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00019878490238836415,
      "loss": 1.0249,
      "step": 86
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00019873871461540185,
      "loss": 1.0171,
      "step": 87
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00019869167087338907,
      "loss": 1.0151,
      "step": 88
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00019864377157014952,
      "loss": 1.0435,
      "step": 89
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00019859501712092374,
      "loss": 1.0117,
      "step": 90
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0001985454079483656,
      "loss": 1.0114,
      "step": 91
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019849494448253863,
      "loss": 1.0297,
      "step": 92
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00019844362716091215,
      "loss": 0.9811,
      "step": 93
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00019839145642835772,
      "loss": 1.0065,
      "step": 94
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00019833843273714505,
      "loss": 1.0101,
      "step": 95
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0001982845565469382,
      "loss": 0.9774,
      "step": 96
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00019822982832479156,
      "loss": 0.9908,
      "step": 97
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019817424854514583,
      "loss": 1.0048,
      "step": 98
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0001981178176898239,
      "loss": 1.0564,
      "step": 99
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00019806053624802665,
      "loss": 1.0181,
      "step": 100
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001980024047163287,
      "loss": 1.0054,
      "step": 101
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00019794342359867413,
      "loss": 1.0615,
      "step": 102
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00019788359340637212,
      "loss": 1.0096,
      "step": 103
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00019782291465809254,
      "loss": 0.9647,
      "step": 104
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0001977613878798614,
      "loss": 1.0785,
      "step": 105
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00019769901360505623,
      "loss": 1.0377,
      "step": 106
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00019763579237440162,
      "loss": 1.0066,
      "step": 107
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00019757172473596445,
      "loss": 0.9941,
      "step": 108
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00019750681124514906,
      "loss": 1.0009,
      "step": 109
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00019744105246469263,
      "loss": 0.9947,
      "step": 110
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00019737444896466007,
      "loss": 0.7684,
      "step": 111
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00019730700132243923,
      "loss": 0.7996,
      "step": 112
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00019723871012273587,
      "loss": 0.7788,
      "step": 113
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0001971695759575686,
      "loss": 0.772,
      "step": 114
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00019709959942626365,
      "loss": 0.7489,
      "step": 115
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001970287811354498,
      "loss": 0.8119,
      "step": 116
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00019695712169905306,
      "loss": 0.8026,
      "step": 117
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00019688462173829142,
      "loss": 0.7304,
      "step": 118
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0001968112818816692,
      "loss": 0.7994,
      "step": 119
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0001967371027649721,
      "loss": 0.7524,
      "step": 120
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00019666208503126112,
      "loss": 0.7649,
      "step": 121
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00019658622933086745,
      "loss": 0.8006,
      "step": 122
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0001965095363213865,
      "loss": 0.7658,
      "step": 123
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0001964320066676725,
      "loss": 0.7576,
      "step": 124
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00019635364104183235,
      "loss": 0.7667,
      "step": 125
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00019627444012322026,
      "loss": 0.7654,
      "step": 126
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00019619440459843143,
      "loss": 0.7745,
      "step": 127
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00019611353516129642,
      "loss": 0.7684,
      "step": 128
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0001960318325128749,
      "loss": 0.7513,
      "step": 129
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00019594929736144976,
      "loss": 0.7768,
      "step": 130
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0001958659304225208,
      "loss": 0.7667,
      "step": 131
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00019578173241879872,
      "loss": 0.7559,
      "step": 132
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00019569670408019864,
      "loss": 0.7424,
      "step": 133
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00019561084614383396,
      "loss": 0.7191,
      "step": 134
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001955241593540098,
      "loss": 0.7827,
      "step": 135
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00019543664446221678,
      "loss": 0.8107,
      "step": 136
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001953483022271242,
      "loss": 0.7312,
      "step": 137
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001952591334145737,
      "loss": 0.7395,
      "step": 138
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00019516913879757258,
      "loss": 0.7216,
      "step": 139
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00019507831915628704,
      "loss": 0.7588,
      "step": 140
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0001949866752780354,
      "loss": 0.7738,
      "step": 141
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00019489420795728134,
      "loss": 0.7664,
      "step": 142
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00019480091799562704,
      "loss": 0.7325,
      "step": 143
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00019470680620180614,
      "loss": 0.7847,
      "step": 144
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00019461187339167671,
      "loss": 0.7444,
      "step": 145
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00019451612038821434,
      "loss": 0.7668,
      "step": 146
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00019441954802150484,
      "loss": 0.7823,
      "step": 147
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001943221571287371,
      "loss": 0.7661,
      "step": 148
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00019422394855419592,
      "loss": 0.7614,
      "step": 149
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00019412492314925452,
      "loss": 0.8043,
      "step": 150
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00019402508177236724,
      "loss": 0.7802,
      "step": 151
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00019392442528906218,
      "loss": 0.7813,
      "step": 152
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0001938229545719336,
      "loss": 0.7515,
      "step": 153
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00019372067050063438,
      "loss": 0.7397,
      "step": 154
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00019361757396186834,
      "loss": 0.7684,
      "step": 155
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00019351366584938263,
      "loss": 0.8109,
      "step": 156
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00019340894706396,
      "loss": 0.78,
      "step": 157
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00019330341851341097,
      "loss": 0.7766,
      "step": 158
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00019319708111256582,
      "loss": 0.7474,
      "step": 159
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0001930899357832669,
      "loss": 0.7223,
      "step": 160
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00019298198345436043,
      "loss": 0.8049,
      "step": 161
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00019287322506168861,
      "loss": 0.7267,
      "step": 162
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00019276366154808138,
      "loss": 0.7611,
      "step": 163
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00019265329386334833,
      "loss": 0.7677,
      "step": 164
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00019254212296427044,
      "loss": 0.8044,
      "step": 165
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0001924301498145917,
      "loss": 0.7477,
      "step": 166
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00019231737538501093,
      "loss": 0.7394,
      "step": 167
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00019220380065317327,
      "loss": 0.7248,
      "step": 168
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00019208942660366165,
      "loss": 0.7907,
      "step": 169
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00019197425422798833,
      "loss": 0.7903,
      "step": 170
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00019185828452458628,
      "loss": 0.8229,
      "step": 171
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00019174151849880056,
      "loss": 0.8371,
      "step": 172
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00019162395716287956,
      "loss": 0.7744,
      "step": 173
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00019150560153596618,
      "loss": 0.7723,
      "step": 174
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00019138645264408916,
      "loss": 0.7998,
      "step": 175
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00019126651152015403,
      "loss": 0.7477,
      "step": 176
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00019114577920393414,
      "loss": 0.8233,
      "step": 177
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00019102425674206183,
      "loss": 0.7602,
      "step": 178
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00019090194518801915,
      "loss": 0.7294,
      "step": 179
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00019077884560212885,
      "loss": 0.7563,
      "step": 180
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0001906549590515452,
      "loss": 0.7924,
      "step": 181
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0001905302866102445,
      "loss": 0.7367,
      "step": 182
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00019040482935901623,
      "loss": 0.7015,
      "step": 183
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00019027858838545323,
      "loss": 0.8093,
      "step": 184
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00019015156478394252,
      "loss": 0.7707,
      "step": 185
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0001900237596556557,
      "loss": 0.787,
      "step": 186
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00018989517410853955,
      "loss": 0.8094,
      "step": 187
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0001897658092573062,
      "loss": 0.8045,
      "step": 188
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00018963566622342366,
      "loss": 0.8224,
      "step": 189
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00018950474613510605,
      "loss": 0.781,
      "step": 190
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.00018937305012730373,
      "loss": 0.7438,
      "step": 191
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0001892405793416936,
      "loss": 0.8044,
      "step": 192
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00018910733492666907,
      "loss": 0.743,
      "step": 193
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.00018897331803733027,
      "loss": 0.818,
      "step": 194
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.00018883852983547382,
      "loss": 0.7227,
      "step": 195
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00018870297148958288,
      "loss": 0.7971,
      "step": 196
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00018856664417481715,
      "loss": 0.8214,
      "step": 197
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00018842954907300236,
      "loss": 0.7984,
      "step": 198
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00018829168737262037,
      "loss": 0.8087,
      "step": 199
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00018815306026879856,
      "loss": 0.762,
      "step": 200
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00018801366896329974,
      "loss": 0.8216,
      "step": 201
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00018787351466451155,
      "loss": 0.7769,
      "step": 202
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.00018773259858743607,
      "loss": 0.8061,
      "step": 203
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.00018759092195367924,
      "loss": 0.786,
      "step": 204
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.00018744848599144026,
      "loss": 0.7582,
      "step": 205
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00018730529193550102,
      "loss": 0.8243,
      "step": 206
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0001871613410272154,
      "loss": 0.7878,
      "step": 207
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.00018701663451449832,
      "loss": 0.7152,
      "step": 208
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00018687117365181512,
      "loss": 0.8343,
      "step": 209
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0001867249597001707,
      "loss": 0.7689,
      "step": 210
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00018657799392709837,
      "loss": 0.8025,
      "step": 211
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.00018643027760664916,
      "loss": 0.8049,
      "step": 212
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00018628181201938048,
      "loss": 0.7477,
      "step": 213
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00018613259845234523,
      "loss": 0.7493,
      "step": 214
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0001859826381990806,
      "loss": 0.7873,
      "step": 215
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00018583193255959673,
      "loss": 0.7504,
      "step": 216
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00018568048284036561,
      "loss": 0.7952,
      "step": 217
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.00018552829035430968,
      "loss": 0.8027,
      "step": 218
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00018537535642079042,
      "loss": 0.7545,
      "step": 219
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00018522168236559695,
      "loss": 0.6007,
      "step": 220
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0001850672695209345,
      "loss": 0.445,
      "step": 221
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.00018491211922541298,
      "loss": 0.3805,
      "step": 222
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0001847562328240352,
      "loss": 0.3903,
      "step": 223
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0001845996116681853,
      "loss": 0.4189,
      "step": 224
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00018444225711561712,
      "loss": 0.3914,
      "step": 225
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00018428417053044217,
      "loss": 0.3721,
      "step": 226
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00018412535328311814,
      "loss": 0.3809,
      "step": 227
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00018396580675043676,
      "loss": 0.4338,
      "step": 228
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.00018380553231551197,
      "loss": 0.4181,
      "step": 229
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.00018364453136776802,
      "loss": 0.379,
      "step": 230
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00018348280530292713,
      "loss": 0.3891,
      "step": 231
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00018332035552299773,
      "loss": 0.4007,
      "step": 232
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00018315718343626215,
      "loss": 0.3883,
      "step": 233
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00018299329045726433,
      "loss": 0.39,
      "step": 234
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00018282867800679774,
      "loss": 0.4199,
      "step": 235
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.00018266334751189294,
      "loss": 0.3938,
      "step": 236
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0001824973004058052,
      "loss": 0.4177,
      "step": 237
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0001823305381280022,
      "loss": 0.4017,
      "step": 238
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.0001821630621241514,
      "loss": 0.4202,
      "step": 239
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.00018199487384610758,
      "loss": 0.4298,
      "step": 240
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.00018182597475190028,
      "loss": 0.4002,
      "step": 241
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0001816563663057211,
      "loss": 0.3872,
      "step": 242
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.00018148604997791104,
      "loss": 0.3952,
      "step": 243
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00018131502724494776,
      "loss": 0.3956,
      "step": 244
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0001811432995894327,
      "loss": 0.3798,
      "step": 245
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00018097086850007847,
      "loss": 0.3606,
      "step": 246
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00018079773547169557,
      "loss": 0.3916,
      "step": 247
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00018062390200517973,
      "loss": 0.3831,
      "step": 248
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0001804493696074988,
      "loss": 0.3789,
      "step": 249
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.00018027413979167966,
      "loss": 0.4011,
      "step": 250
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0001800982140767952,
      "loss": 0.4575,
      "step": 251
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.00017992159398795093,
      "loss": 0.417,
      "step": 252
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.00017974428105627208,
      "loss": 0.3838,
      "step": 253
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.00017956627681889006,
      "loss": 0.3594,
      "step": 254
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.00017938758281892928,
      "loss": 0.4085,
      "step": 255
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.00017920820060549362,
      "loss": 0.4102,
      "step": 256
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.00017902813173365325,
      "loss": 0.4135,
      "step": 257
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.0001788473777644309,
      "loss": 0.416,
      "step": 258
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00017866594026478847,
      "loss": 0.4096,
      "step": 259
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00017848382080761342,
      "loss": 0.4117,
      "step": 260
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.00017830102097170513,
      "loss": 0.4426,
      "step": 261
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.0001781175423417611,
      "loss": 0.3994,
      "step": 262
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.00017793338650836345,
      "loss": 0.4522,
      "step": 263
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00017774855506796496,
      "loss": 0.4467,
      "step": 264
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.0001775630496228752,
      "loss": 0.4326,
      "step": 265
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00017737687178124678,
      "loss": 0.4128,
      "step": 266
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0001771900231570613,
      "loss": 0.4097,
      "step": 267
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0001770025053701155,
      "loss": 0.4273,
      "step": 268
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00017681432004600694,
      "loss": 0.4085,
      "step": 269
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00017662546881612017,
      "loss": 0.3865,
      "step": 270
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0001764359533176125,
      "loss": 0.4147,
      "step": 271
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.00017624577519339987,
      "loss": 0.3966,
      "step": 272
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.00017605493609214236,
      "loss": 0.4235,
      "step": 273
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.0001758634376682303,
      "loss": 0.4583,
      "step": 274
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.00017567128158176953,
      "loss": 0.3938,
      "step": 275
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0001754784694985673,
      "loss": 0.4009,
      "step": 276
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0001752850030901177,
      "loss": 0.4464,
      "step": 277
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.00017509088403358706,
      "loss": 0.4142,
      "step": 278
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00017489611401179977,
      "loss": 0.4072,
      "step": 279
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.00017470069471322325,
      "loss": 0.3988,
      "step": 280
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.00017450462783195363,
      "loss": 0.4494,
      "step": 281
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.00017430791506770096,
      "loss": 0.4328,
      "step": 282
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00017411055812577436,
      "loss": 0.4518,
      "step": 283
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00017391255871706746,
      "loss": 0.4066,
      "step": 284
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0001737139185580434,
      "loss": 0.3941,
      "step": 285
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00017351463937072004,
      "loss": 0.4568,
      "step": 286
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.000173314722882655,
      "loss": 0.432,
      "step": 287
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.00017311417082693065,
      "loss": 0.4147,
      "step": 288
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0001729129849421391,
      "loss": 0.3969,
      "step": 289
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.00017271116697236717,
      "loss": 0.4232,
      "step": 290
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00017250871866718125,
      "loss": 0.4514,
      "step": 291
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00017230564178161219,
      "loss": 0.4338,
      "step": 292
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00017210193807613982,
      "loss": 0.463,
      "step": 293
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00017189760931667815,
      "loss": 0.4707,
      "step": 294
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00017169265727455963,
      "loss": 0.4521,
      "step": 295
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00017148708372652003,
      "loss": 0.4076,
      "step": 296
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.00017128089045468294,
      "loss": 0.4315,
      "step": 297
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0001710740792465444,
      "loss": 0.4305,
      "step": 298
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.0001708666518949573,
      "loss": 0.3851,
      "step": 299
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00017065861019811599,
      "loss": 0.3975,
      "step": 300
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00017044995595954048,
      "loss": 0.4555,
      "step": 301
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00017024069098806106,
      "loss": 0.4304,
      "step": 302
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00017003081709780237,
      "loss": 0.4225,
      "step": 303
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00016982033610816774,
      "loss": 0.4471,
      "step": 304
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.00016960924984382373,
      "loss": 0.4191,
      "step": 305
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.0001693975601346837,
      "loss": 0.4532,
      "step": 306
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00016918526881589253,
      "loss": 0.4192,
      "step": 307
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00016897237772781044,
      "loss": 0.4189,
      "step": 308
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00016875888871599695,
      "loss": 0.4091,
      "step": 309
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00016854480363119508,
      "loss": 0.4058,
      "step": 310
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0001683301243293153,
      "loss": 0.4004,
      "step": 311
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00016811485267141923,
      "loss": 0.4156,
      "step": 312
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.0001678989905237037,
      "loss": 0.4063,
      "step": 313
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00016768253975748456,
      "loss": 0.4625,
      "step": 314
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0001674655022491803,
      "loss": 0.3981,
      "step": 315
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0001672478798802961,
      "loss": 0.4275,
      "step": 316
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.000167029674537407,
      "loss": 0.4193,
      "step": 317
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00016681088811214222,
      "loss": 0.447,
      "step": 318
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.00016659152250116812,
      "loss": 0.4103,
      "step": 319
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00016637157960617218,
      "loss": 0.4075,
      "step": 320
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.00016615106133384627,
      "loss": 0.4245,
      "step": 321
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0001659299695958704,
      "loss": 0.4507,
      "step": 322
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0001657083063088958,
      "loss": 0.4194,
      "step": 323
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00016548607339452853,
      "loss": 0.4011,
      "step": 324
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00016526327277931278,
      "loss": 0.3921,
      "step": 325
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0001650399063947142,
      "loss": 0.4722,
      "step": 326
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.000164815976177103,
      "loss": 0.4405,
      "step": 327
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0001645914840677374,
      "loss": 0.4242,
      "step": 328
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.00016436643201274659,
      "loss": 0.4313,
      "step": 329
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.000164140821963114,
      "loss": 0.2972,
      "step": 330
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.00016391465587466028,
      "loss": 0.2008,
      "step": 331
    },
    {
      "epoch": 3.02,
      "learning_rate": 0.00016368793570802645,
      "loss": 0.1809,
      "step": 332
    },
    {
      "epoch": 3.03,
      "learning_rate": 0.00016346066342865683,
      "loss": 0.16,
      "step": 333
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.000163232841006782,
      "loss": 0.1762,
      "step": 334
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.0001630044704174018,
      "loss": 0.1897,
      "step": 335
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.00016277555364026806,
      "loss": 0.1826,
      "step": 336
    },
    {
      "epoch": 3.06,
      "learning_rate": 0.00016254609265986764,
      "loss": 0.1803,
      "step": 337
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.00016231608946540505,
      "loss": 0.1717,
      "step": 338
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.00016208554605078528,
      "loss": 0.1795,
      "step": 339
    },
    {
      "epoch": 3.09,
      "learning_rate": 0.0001618544644145965,
      "loss": 0.176,
      "step": 340
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.00016162284656009274,
      "loss": 0.1964,
      "step": 341
    },
    {
      "epoch": 3.11,
      "learning_rate": 0.00016139069449517653,
      "loss": 0.175,
      "step": 342
    },
    {
      "epoch": 3.12,
      "learning_rate": 0.00016115801023238152,
      "loss": 0.1669,
      "step": 343
    },
    {
      "epoch": 3.13,
      "learning_rate": 0.00016092479578885498,
      "loss": 0.1704,
      "step": 344
    },
    {
      "epoch": 3.14,
      "learning_rate": 0.00016069105318634023,
      "loss": 0.1726,
      "step": 345
    },
    {
      "epoch": 3.15,
      "learning_rate": 0.0001604567844511594,
      "loss": 0.1986,
      "step": 346
    },
    {
      "epoch": 3.15,
      "learning_rate": 0.00016022199161419556,
      "loss": 0.1611,
      "step": 347
    },
    {
      "epoch": 3.16,
      "learning_rate": 0.00015998667671087525,
      "loss": 0.1833,
      "step": 348
    },
    {
      "epoch": 3.17,
      "learning_rate": 0.00015975084178115084,
      "loss": 0.1504,
      "step": 349
    },
    {
      "epoch": 3.18,
      "learning_rate": 0.00015951448886948285,
      "loss": 0.1747,
      "step": 350
    },
    {
      "epoch": 3.19,
      "learning_rate": 0.00015927762002482213,
      "loss": 0.1881,
      "step": 351
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.00015904023730059228,
      "loss": 0.1685,
      "step": 352
    },
    {
      "epoch": 3.21,
      "learning_rate": 0.00015880234275467158,
      "loss": 0.1716,
      "step": 353
    },
    {
      "epoch": 3.22,
      "learning_rate": 0.00015856393844937544,
      "loss": 0.1474,
      "step": 354
    },
    {
      "epoch": 3.23,
      "learning_rate": 0.00015832502645143836,
      "loss": 0.1751,
      "step": 355
    },
    {
      "epoch": 3.24,
      "learning_rate": 0.00015808560883199602,
      "loss": 0.1492,
      "step": 356
    },
    {
      "epoch": 3.25,
      "learning_rate": 0.00015784568766656734,
      "loss": 0.1611,
      "step": 357
    },
    {
      "epoch": 3.25,
      "learning_rate": 0.0001576052650350366,
      "loss": 0.2032,
      "step": 358
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.0001573643430216351,
      "loss": 0.1789,
      "step": 359
    },
    {
      "epoch": 3.27,
      "learning_rate": 0.0001571229237149235,
      "loss": 0.1883,
      "step": 360
    },
    {
      "epoch": 3.28,
      "learning_rate": 0.00015688100920777342,
      "loss": 0.1783,
      "step": 361
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.00015663860159734937,
      "loss": 0.1613,
      "step": 362
    },
    {
      "epoch": 3.3,
      "learning_rate": 0.00015639570298509064,
      "loss": 0.1597,
      "step": 363
    },
    {
      "epoch": 3.31,
      "learning_rate": 0.00015615231547669303,
      "loss": 0.1806,
      "step": 364
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.00015590844118209055,
      "loss": 0.1768,
      "step": 365
    },
    {
      "epoch": 3.33,
      "learning_rate": 0.0001556640822154372,
      "loss": 0.1819,
      "step": 366
    },
    {
      "epoch": 3.34,
      "learning_rate": 0.0001554192406950886,
      "loss": 0.168,
      "step": 367
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.0001551739187435837,
      "loss": 0.1874,
      "step": 368
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.00015492811848762626,
      "loss": 0.1817,
      "step": 369
    },
    {
      "epoch": 3.36,
      "learning_rate": 0.00015468184205806647,
      "loss": 0.1737,
      "step": 370
    },
    {
      "epoch": 3.37,
      "learning_rate": 0.00015443509158988254,
      "loss": 0.1636,
      "step": 371
    },
    {
      "epoch": 3.38,
      "learning_rate": 0.00015418786922216215,
      "loss": 0.1528,
      "step": 372
    },
    {
      "epoch": 3.39,
      "learning_rate": 0.00015394017709808382,
      "loss": 0.199,
      "step": 373
    },
    {
      "epoch": 3.4,
      "learning_rate": 0.0001536920173648984,
      "loss": 0.2006,
      "step": 374
    },
    {
      "epoch": 3.41,
      "learning_rate": 0.0001534433921739105,
      "loss": 0.1793,
      "step": 375
    },
    {
      "epoch": 3.42,
      "learning_rate": 0.00015319430368045987,
      "loss": 0.188,
      "step": 376
    },
    {
      "epoch": 3.43,
      "learning_rate": 0.00015294475404390245,
      "loss": 0.176,
      "step": 377
    },
    {
      "epoch": 3.44,
      "learning_rate": 0.00015269474542759206,
      "loss": 0.1753,
      "step": 378
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.00015244427999886123,
      "loss": 0.1695,
      "step": 379
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.00015219335992900282,
      "loss": 0.1911,
      "step": 380
    },
    {
      "epoch": 3.46,
      "learning_rate": 0.0001519419873932508,
      "loss": 0.1624,
      "step": 381
    },
    {
      "epoch": 3.47,
      "learning_rate": 0.0001516901645707617,
      "loss": 0.1628,
      "step": 382
    },
    {
      "epoch": 3.48,
      "learning_rate": 0.0001514378936445955,
      "loss": 0.1587,
      "step": 383
    },
    {
      "epoch": 3.49,
      "learning_rate": 0.00015118517680169695,
      "loss": 0.1748,
      "step": 384
    },
    {
      "epoch": 3.5,
      "learning_rate": 0.00015093201623287631,
      "loss": 0.1657,
      "step": 385
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.00015067841413279062,
      "loss": 0.178,
      "step": 386
    },
    {
      "epoch": 3.52,
      "learning_rate": 0.00015042437269992452,
      "loss": 0.1824,
      "step": 387
    },
    {
      "epoch": 3.53,
      "learning_rate": 0.00015016989413657128,
      "loss": 0.1836,
      "step": 388
    },
    {
      "epoch": 3.54,
      "learning_rate": 0.0001499149806488136,
      "loss": 0.1714,
      "step": 389
    },
    {
      "epoch": 3.55,
      "learning_rate": 0.00014965963444650465,
      "loss": 0.1756,
      "step": 390
    },
    {
      "epoch": 3.55,
      "learning_rate": 0.0001494038577432487,
      "loss": 0.1934,
      "step": 391
    },
    {
      "epoch": 3.56,
      "learning_rate": 0.00014914765275638215,
      "loss": 0.166,
      "step": 392
    },
    {
      "epoch": 3.57,
      "learning_rate": 0.00014889102170695414,
      "loss": 0.1658,
      "step": 393
    },
    {
      "epoch": 3.58,
      "learning_rate": 0.0001486339668197074,
      "loss": 0.1732,
      "step": 394
    },
    {
      "epoch": 3.59,
      "learning_rate": 0.00014837649032305885,
      "loss": 0.1843,
      "step": 395
    },
    {
      "epoch": 3.6,
      "learning_rate": 0.00014811859444908052,
      "loss": 0.2113,
      "step": 396
    },
    {
      "epoch": 3.61,
      "learning_rate": 0.00014786028143347983,
      "loss": 0.1672,
      "step": 397
    },
    {
      "epoch": 3.62,
      "learning_rate": 0.00014760155351558054,
      "loss": 0.1796,
      "step": 398
    },
    {
      "epoch": 3.63,
      "learning_rate": 0.0001473424129383032,
      "loss": 0.1806,
      "step": 399
    },
    {
      "epoch": 3.64,
      "learning_rate": 0.00014708286194814566,
      "loss": 0.1811,
      "step": 400
    },
    {
      "epoch": 3.65,
      "learning_rate": 0.0001468229027951637,
      "loss": 0.1776,
      "step": 401
    },
    {
      "epoch": 3.65,
      "learning_rate": 0.0001465625377329515,
      "loss": 0.1757,
      "step": 402
    },
    {
      "epoch": 3.66,
      "learning_rate": 0.00014630176901862205,
      "loss": 0.1771,
      "step": 403
    },
    {
      "epoch": 3.67,
      "learning_rate": 0.00014604059891278753,
      "loss": 0.1821,
      "step": 404
    },
    {
      "epoch": 3.68,
      "learning_rate": 0.00014577902967953995,
      "loss": 0.1822,
      "step": 405
    },
    {
      "epoch": 3.69,
      "learning_rate": 0.0001455170635864313,
      "loss": 0.1928,
      "step": 406
    },
    {
      "epoch": 3.7,
      "learning_rate": 0.00014525470290445392,
      "loss": 0.185,
      "step": 407
    },
    {
      "epoch": 3.71,
      "learning_rate": 0.00014499194990802086,
      "loss": 0.1628,
      "step": 408
    },
    {
      "epoch": 3.72,
      "learning_rate": 0.00014472880687494625,
      "loss": 0.1686,
      "step": 409
    },
    {
      "epoch": 3.73,
      "learning_rate": 0.00014446527608642538,
      "loss": 0.2042,
      "step": 410
    },
    {
      "epoch": 3.74,
      "learning_rate": 0.000144201359827015,
      "loss": 0.183,
      "step": 411
    },
    {
      "epoch": 3.75,
      "learning_rate": 0.00014393706038461353,
      "loss": 0.204,
      "step": 412
    },
    {
      "epoch": 3.75,
      "learning_rate": 0.00014367238005044125,
      "loss": 0.1639,
      "step": 413
    },
    {
      "epoch": 3.76,
      "learning_rate": 0.00014340732111902034,
      "loss": 0.1812,
      "step": 414
    },
    {
      "epoch": 3.77,
      "learning_rate": 0.00014314188588815513,
      "loss": 0.1792,
      "step": 415
    },
    {
      "epoch": 3.78,
      "learning_rate": 0.00014287607665891203,
      "loss": 0.1976,
      "step": 416
    },
    {
      "epoch": 3.79,
      "learning_rate": 0.00014260989573559974,
      "loss": 0.1548,
      "step": 417
    },
    {
      "epoch": 3.8,
      "learning_rate": 0.00014234334542574906,
      "loss": 0.1933,
      "step": 418
    },
    {
      "epoch": 3.81,
      "learning_rate": 0.00014207642804009308,
      "loss": 0.2162,
      "step": 419
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.00014180914589254716,
      "loss": 0.176,
      "step": 420
    },
    {
      "epoch": 3.83,
      "learning_rate": 0.00014154150130018866,
      "loss": 0.1758,
      "step": 421
    },
    {
      "epoch": 3.84,
      "learning_rate": 0.00014127349658323707,
      "loss": 0.19,
      "step": 422
    },
    {
      "epoch": 3.85,
      "learning_rate": 0.00014100513406503377,
      "loss": 0.183,
      "step": 423
    },
    {
      "epoch": 3.85,
      "learning_rate": 0.00014073641607202197,
      "loss": 0.1919,
      "step": 424
    },
    {
      "epoch": 3.86,
      "learning_rate": 0.00014046734493372646,
      "loss": 0.1728,
      "step": 425
    },
    {
      "epoch": 3.87,
      "learning_rate": 0.00014019792298273348,
      "loss": 0.1691,
      "step": 426
    },
    {
      "epoch": 3.88,
      "learning_rate": 0.00013992815255467052,
      "loss": 0.2098,
      "step": 427
    },
    {
      "epoch": 3.89,
      "learning_rate": 0.00013965803598818587,
      "loss": 0.2139,
      "step": 428
    },
    {
      "epoch": 3.9,
      "learning_rate": 0.00013938757562492873,
      "loss": 0.176,
      "step": 429
    },
    {
      "epoch": 3.91,
      "learning_rate": 0.00013911677380952853,
      "loss": 0.1935,
      "step": 430
    },
    {
      "epoch": 3.92,
      "learning_rate": 0.00013884563288957478,
      "loss": 0.1999,
      "step": 431
    },
    {
      "epoch": 3.93,
      "learning_rate": 0.00013857415521559676,
      "loss": 0.1779,
      "step": 432
    },
    {
      "epoch": 3.94,
      "learning_rate": 0.00013830234314104298,
      "loss": 0.1669,
      "step": 433
    },
    {
      "epoch": 3.95,
      "learning_rate": 0.000138030199022261,
      "loss": 0.1623,
      "step": 434
    },
    {
      "epoch": 3.95,
      "learning_rate": 0.00013775772521847684,
      "loss": 0.1798,
      "step": 435
    },
    {
      "epoch": 3.96,
      "learning_rate": 0.00013748492409177448,
      "loss": 0.1718,
      "step": 436
    },
    {
      "epoch": 3.97,
      "learning_rate": 0.00013721179800707555,
      "loss": 0.189,
      "step": 437
    },
    {
      "epoch": 3.98,
      "learning_rate": 0.0001369383493321188,
      "loss": 0.1841,
      "step": 438
    },
    {
      "epoch": 3.99,
      "learning_rate": 0.00013666458043743948,
      "loss": 0.19,
      "step": 439
    },
    {
      "epoch": 4.0,
      "learning_rate": 0.00013639049369634876,
      "loss": 0.1205,
      "step": 440
    },
    {
      "epoch": 4.01,
      "learning_rate": 0.00013611609148491332,
      "loss": 0.0822,
      "step": 441
    },
    {
      "epoch": 4.02,
      "learning_rate": 0.00013584137618193469,
      "loss": 0.0676,
      "step": 442
    },
    {
      "epoch": 4.03,
      "learning_rate": 0.00013556635016892846,
      "loss": 0.0721,
      "step": 443
    },
    {
      "epoch": 4.04,
      "learning_rate": 0.00013529101583010394,
      "loss": 0.0752,
      "step": 444
    },
    {
      "epoch": 4.05,
      "learning_rate": 0.00013501537555234323,
      "loss": 0.0806,
      "step": 445
    },
    {
      "epoch": 4.05,
      "learning_rate": 0.00013473943172518066,
      "loss": 0.0743,
      "step": 446
    },
    {
      "epoch": 4.06,
      "learning_rate": 0.00013446318674078194,
      "loss": 0.0779,
      "step": 447
    },
    {
      "epoch": 4.07,
      "learning_rate": 0.0001341866429939237,
      "loss": 0.0655,
      "step": 448
    },
    {
      "epoch": 4.08,
      "learning_rate": 0.00013390980288197242,
      "loss": 0.0669,
      "step": 449
    },
    {
      "epoch": 4.09,
      "learning_rate": 0.00013363266880486388,
      "loss": 0.063,
      "step": 450
    },
    {
      "epoch": 4.1,
      "learning_rate": 0.00013335524316508208,
      "loss": 0.0632,
      "step": 451
    },
    {
      "epoch": 4.11,
      "learning_rate": 0.00013307752836763873,
      "loss": 0.0682,
      "step": 452
    },
    {
      "epoch": 4.12,
      "learning_rate": 0.00013279952682005228,
      "loss": 0.0855,
      "step": 453
    },
    {
      "epoch": 4.13,
      "learning_rate": 0.0001325212409323269,
      "loss": 0.0633,
      "step": 454
    },
    {
      "epoch": 4.14,
      "learning_rate": 0.00013224267311693185,
      "loss": 0.0779,
      "step": 455
    },
    {
      "epoch": 4.15,
      "learning_rate": 0.00013196382578878032,
      "loss": 0.0815,
      "step": 456
    },
    {
      "epoch": 4.15,
      "learning_rate": 0.00013168470136520864,
      "loss": 0.0694,
      "step": 457
    },
    {
      "epoch": 4.16,
      "learning_rate": 0.0001314053022659553,
      "loss": 0.068,
      "step": 458
    },
    {
      "epoch": 4.17,
      "learning_rate": 0.00013112563091313999,
      "loss": 0.0566,
      "step": 459
    },
    {
      "epoch": 4.18,
      "learning_rate": 0.00013084568973124248,
      "loss": 0.0692,
      "step": 460
    },
    {
      "epoch": 4.19,
      "learning_rate": 0.00013056548114708183,
      "loss": 0.0748,
      "step": 461
    },
    {
      "epoch": 4.2,
      "learning_rate": 0.00013028500758979506,
      "loss": 0.0681,
      "step": 462
    },
    {
      "epoch": 4.21,
      "learning_rate": 0.00013000427149081645,
      "loss": 0.0726,
      "step": 463
    },
    {
      "epoch": 4.22,
      "learning_rate": 0.0001297232752838561,
      "loss": 0.0671,
      "step": 464
    },
    {
      "epoch": 4.23,
      "learning_rate": 0.00012944202140487905,
      "loss": 0.0626,
      "step": 465
    },
    {
      "epoch": 4.24,
      "learning_rate": 0.00012916051229208407,
      "loss": 0.065,
      "step": 466
    },
    {
      "epoch": 4.25,
      "learning_rate": 0.0001288787503858826,
      "loss": 0.0591,
      "step": 467
    },
    {
      "epoch": 4.25,
      "learning_rate": 0.00012859673812887757,
      "loss": 0.0656,
      "step": 468
    },
    {
      "epoch": 4.26,
      "learning_rate": 0.00012831447796584223,
      "loss": 0.063,
      "step": 469
    },
    {
      "epoch": 4.27,
      "learning_rate": 0.00012803197234369877,
      "loss": 0.0702,
      "step": 470
    },
    {
      "epoch": 4.28,
      "learning_rate": 0.0001277492237114975,
      "loss": 0.0632,
      "step": 471
    },
    {
      "epoch": 4.29,
      "learning_rate": 0.00012746623452039525,
      "loss": 0.0711,
      "step": 472
    },
    {
      "epoch": 4.3,
      "learning_rate": 0.0001271830072236343,
      "loss": 0.0637,
      "step": 473
    },
    {
      "epoch": 4.31,
      "learning_rate": 0.00012689954427652108,
      "loss": 0.0558,
      "step": 474
    },
    {
      "epoch": 4.32,
      "learning_rate": 0.00012661584813640484,
      "loss": 0.0602,
      "step": 475
    },
    {
      "epoch": 4.33,
      "learning_rate": 0.00012633192126265645,
      "loss": 0.0755,
      "step": 476
    },
    {
      "epoch": 4.34,
      "learning_rate": 0.00012604776611664702,
      "loss": 0.0597,
      "step": 477
    },
    {
      "epoch": 4.35,
      "learning_rate": 0.00012576338516172644,
      "loss": 0.0695,
      "step": 478
    },
    {
      "epoch": 4.35,
      "learning_rate": 0.00012547878086320235,
      "loss": 0.0541,
      "step": 479
    },
    {
      "epoch": 4.36,
      "learning_rate": 0.00012519395568831836,
      "loss": 0.0697,
      "step": 480
    },
    {
      "epoch": 4.37,
      "learning_rate": 0.00012490891210623298,
      "loss": 0.0747,
      "step": 481
    },
    {
      "epoch": 4.38,
      "learning_rate": 0.00012462365258799802,
      "loss": 0.0729,
      "step": 482
    },
    {
      "epoch": 4.39,
      "learning_rate": 0.0001243381796065374,
      "loss": 0.0681,
      "step": 483
    },
    {
      "epoch": 4.4,
      "learning_rate": 0.00012405249563662537,
      "loss": 0.0563,
      "step": 484
    },
    {
      "epoch": 4.41,
      "learning_rate": 0.00012376660315486533,
      "loss": 0.0736,
      "step": 485
    },
    {
      "epoch": 4.42,
      "learning_rate": 0.00012348050463966833,
      "loss": 0.0598,
      "step": 486
    },
    {
      "epoch": 4.43,
      "learning_rate": 0.0001231942025712315,
      "loss": 0.0671,
      "step": 487
    },
    {
      "epoch": 4.44,
      "learning_rate": 0.00012290769943151646,
      "loss": 0.078,
      "step": 488
    },
    {
      "epoch": 4.45,
      "learning_rate": 0.0001226209977042281,
      "loss": 0.0667,
      "step": 489
    },
    {
      "epoch": 4.45,
      "learning_rate": 0.0001223340998747927,
      "loss": 0.0706,
      "step": 490
    },
    {
      "epoch": 4.46,
      "learning_rate": 0.00012204700843033674,
      "loss": 0.063,
      "step": 491
    },
    {
      "epoch": 4.47,
      "learning_rate": 0.00012175972585966504,
      "loss": 0.0683,
      "step": 492
    },
    {
      "epoch": 4.48,
      "learning_rate": 0.00012147225465323932,
      "loss": 0.0646,
      "step": 493
    },
    {
      "epoch": 4.49,
      "learning_rate": 0.00012118459730315659,
      "loss": 0.0676,
      "step": 494
    },
    {
      "epoch": 4.5,
      "learning_rate": 0.00012089675630312754,
      "loss": 0.0707,
      "step": 495
    },
    {
      "epoch": 4.51,
      "learning_rate": 0.0001206087341484549,
      "loss": 0.065,
      "step": 496
    },
    {
      "epoch": 4.52,
      "learning_rate": 0.00012032053333601194,
      "loss": 0.0683,
      "step": 497
    },
    {
      "epoch": 4.53,
      "learning_rate": 0.00012003215636422062,
      "loss": 0.0653,
      "step": 498
    },
    {
      "epoch": 4.54,
      "learning_rate": 0.00011974360573303005,
      "loss": 0.0746,
      "step": 499
    },
    {
      "epoch": 4.55,
      "learning_rate": 0.00011945488394389479,
      "loss": 0.0681,
      "step": 500
    },
    {
      "epoch": 4.55,
      "learning_rate": 0.00011916599349975327,
      "loss": 0.0706,
      "step": 501
    },
    {
      "epoch": 4.56,
      "learning_rate": 0.00011887693690500583,
      "loss": 0.0633,
      "step": 502
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.00011858771666549329,
      "loss": 0.0585,
      "step": 503
    },
    {
      "epoch": 4.58,
      "learning_rate": 0.00011829833528847511,
      "loss": 0.0669,
      "step": 504
    },
    {
      "epoch": 4.59,
      "learning_rate": 0.0001180087952826076,
      "loss": 0.0784,
      "step": 505
    },
    {
      "epoch": 4.6,
      "learning_rate": 0.0001177190991579223,
      "loss": 0.0678,
      "step": 506
    },
    {
      "epoch": 4.61,
      "learning_rate": 0.00011742924942580404,
      "loss": 0.0525,
      "step": 507
    },
    {
      "epoch": 4.62,
      "learning_rate": 0.00011713924859896943,
      "loss": 0.0575,
      "step": 508
    },
    {
      "epoch": 4.63,
      "learning_rate": 0.00011684909919144478,
      "loss": 0.0633,
      "step": 509
    },
    {
      "epoch": 4.64,
      "learning_rate": 0.00011655880371854454,
      "loss": 0.0657,
      "step": 510
    },
    {
      "epoch": 4.65,
      "learning_rate": 0.00011626836469684934,
      "loss": 0.0613,
      "step": 511
    },
    {
      "epoch": 4.65,
      "learning_rate": 0.00011597778464418433,
      "loss": 0.064,
      "step": 512
    },
    {
      "epoch": 4.66,
      "learning_rate": 0.0001156870660795972,
      "loss": 0.0656,
      "step": 513
    },
    {
      "epoch": 4.67,
      "learning_rate": 0.00011539621152333635,
      "loss": 0.0619,
      "step": 514
    },
    {
      "epoch": 4.68,
      "learning_rate": 0.00011510522349682921,
      "loss": 0.0686,
      "step": 515
    },
    {
      "epoch": 4.69,
      "learning_rate": 0.00011481410452266026,
      "loss": 0.0702,
      "step": 516
    },
    {
      "epoch": 4.7,
      "learning_rate": 0.00011452285712454904,
      "loss": 0.0642,
      "step": 517
    },
    {
      "epoch": 4.71,
      "learning_rate": 0.00011423148382732853,
      "loss": 0.0691,
      "step": 518
    },
    {
      "epoch": 4.72,
      "learning_rate": 0.00011393998715692307,
      "loss": 0.0605,
      "step": 519
    },
    {
      "epoch": 4.73,
      "learning_rate": 0.00011364836964032657,
      "loss": 0.0557,
      "step": 520
    },
    {
      "epoch": 4.74,
      "learning_rate": 0.00011335663380558053,
      "loss": 0.0595,
      "step": 521
    },
    {
      "epoch": 4.75,
      "learning_rate": 0.00011306478218175216,
      "loss": 0.0614,
      "step": 522
    },
    {
      "epoch": 4.75,
      "learning_rate": 0.00011277281729891241,
      "loss": 0.0578,
      "step": 523
    },
    {
      "epoch": 4.76,
      "learning_rate": 0.00011248074168811416,
      "loss": 0.0668,
      "step": 524
    },
    {
      "epoch": 4.77,
      "learning_rate": 0.00011218855788137015,
      "loss": 0.0694,
      "step": 525
    },
    {
      "epoch": 4.78,
      "learning_rate": 0.00011189626841163108,
      "loss": 0.0637,
      "step": 526
    },
    {
      "epoch": 4.79,
      "learning_rate": 0.00011160387581276362,
      "loss": 0.0683,
      "step": 527
    },
    {
      "epoch": 4.8,
      "learning_rate": 0.00011131138261952845,
      "loss": 0.065,
      "step": 528
    },
    {
      "epoch": 4.81,
      "learning_rate": 0.00011101879136755837,
      "loss": 0.0595,
      "step": 529
    },
    {
      "epoch": 4.82,
      "learning_rate": 0.0001107261045933363,
      "loss": 0.054,
      "step": 530
    },
    {
      "epoch": 4.83,
      "learning_rate": 0.0001104333248341731,
      "loss": 0.0657,
      "step": 531
    },
    {
      "epoch": 4.84,
      "learning_rate": 0.00011014045462818579,
      "loss": 0.0554,
      "step": 532
    },
    {
      "epoch": 4.85,
      "learning_rate": 0.0001098474965142755,
      "loss": 0.0683,
      "step": 533
    },
    {
      "epoch": 4.85,
      "learning_rate": 0.0001095544530321054,
      "loss": 0.0578,
      "step": 534
    },
    {
      "epoch": 4.86,
      "learning_rate": 0.0001092613267220788,
      "loss": 0.0637,
      "step": 535
    },
    {
      "epoch": 4.87,
      "learning_rate": 0.00010896812012531692,
      "loss": 0.0746,
      "step": 536
    },
    {
      "epoch": 4.88,
      "learning_rate": 0.00010867483578363707,
      "loss": 0.0705,
      "step": 537
    },
    {
      "epoch": 4.89,
      "learning_rate": 0.0001083814762395305,
      "loss": 0.0713,
      "step": 538
    },
    {
      "epoch": 4.9,
      "learning_rate": 0.00010808804403614043,
      "loss": 0.0582,
      "step": 539
    },
    {
      "epoch": 4.91,
      "learning_rate": 0.00010779454171723993,
      "loss": 0.0624,
      "step": 540
    },
    {
      "epoch": 4.92,
      "learning_rate": 0.00010750097182720993,
      "loss": 0.0659,
      "step": 541
    },
    {
      "epoch": 4.93,
      "learning_rate": 0.00010720733691101712,
      "loss": 0.0746,
      "step": 542
    },
    {
      "epoch": 4.94,
      "learning_rate": 0.00010691363951419189,
      "loss": 0.0651,
      "step": 543
    },
    {
      "epoch": 4.95,
      "learning_rate": 0.0001066198821828063,
      "loss": 0.0692,
      "step": 544
    },
    {
      "epoch": 4.95,
      "learning_rate": 0.00010632606746345202,
      "loss": 0.07,
      "step": 545
    },
    {
      "epoch": 4.96,
      "learning_rate": 0.00010603219790321812,
      "loss": 0.0686,
      "step": 546
    },
    {
      "epoch": 4.97,
      "learning_rate": 0.00010573827604966918,
      "loss": 0.0621,
      "step": 547
    },
    {
      "epoch": 4.98,
      "learning_rate": 0.00010544430445082301,
      "loss": 0.0705,
      "step": 548
    },
    {
      "epoch": 4.99,
      "learning_rate": 0.0001051502856551288,
      "loss": 0.0538,
      "step": 549
    },
    {
      "epoch": 5.0,
      "learning_rate": 0.00010485622221144484,
      "loss": 0.0445,
      "step": 550
    },
    {
      "epoch": 5.01,
      "learning_rate": 0.00010456211666901639,
      "loss": 0.0315,
      "step": 551
    },
    {
      "epoch": 5.02,
      "learning_rate": 0.00010426797157745375,
      "loss": 0.031,
      "step": 552
    },
    {
      "epoch": 5.03,
      "learning_rate": 0.00010397378948671006,
      "loss": 0.0254,
      "step": 553
    },
    {
      "epoch": 5.04,
      "learning_rate": 0.00010367957294705922,
      "loss": 0.03,
      "step": 554
    },
    {
      "epoch": 5.05,
      "learning_rate": 0.00010338532450907372,
      "loss": 0.0244,
      "step": 555
    },
    {
      "epoch": 5.05,
      "learning_rate": 0.00010309104672360264,
      "loss": 0.0179,
      "step": 556
    },
    {
      "epoch": 5.06,
      "learning_rate": 0.00010279674214174936,
      "loss": 0.0292,
      "step": 557
    },
    {
      "epoch": 5.07,
      "learning_rate": 0.00010250241331484973,
      "loss": 0.031,
      "step": 558
    },
    {
      "epoch": 5.08,
      "learning_rate": 0.00010220806279444967,
      "loss": 0.0288,
      "step": 559
    },
    {
      "epoch": 5.09,
      "learning_rate": 0.00010191369313228319,
      "loss": 0.0282,
      "step": 560
    },
    {
      "epoch": 5.1,
      "learning_rate": 0.00010161930688025017,
      "loss": 0.0216,
      "step": 561
    },
    {
      "epoch": 5.11,
      "learning_rate": 0.00010132490659039443,
      "loss": 0.0316,
      "step": 562
    },
    {
      "epoch": 5.12,
      "learning_rate": 0.00010103049481488141,
      "loss": 0.0308,
      "step": 563
    },
    {
      "epoch": 5.13,
      "learning_rate": 0.00010073607410597616,
      "loss": 0.0199,
      "step": 564
    },
    {
      "epoch": 5.14,
      "learning_rate": 0.0001004416470160211,
      "loss": 0.0256,
      "step": 565
    },
    {
      "epoch": 5.15,
      "learning_rate": 0.00010014721609741404,
      "loss": 0.0245,
      "step": 566
    },
    {
      "epoch": 5.15,
      "learning_rate": 9.985278390258597e-05,
      "loss": 0.0211,
      "step": 567
    },
    {
      "epoch": 5.16,
      "learning_rate": 9.95583529839789e-05,
      "loss": 0.0261,
      "step": 568
    },
    {
      "epoch": 5.17,
      "learning_rate": 9.926392589402387e-05,
      "loss": 0.026,
      "step": 569
    },
    {
      "epoch": 5.18,
      "learning_rate": 9.896950518511861e-05,
      "loss": 0.0222,
      "step": 570
    },
    {
      "epoch": 5.19,
      "learning_rate": 9.867509340960561e-05,
      "loss": 0.0245,
      "step": 571
    },
    {
      "epoch": 5.2,
      "learning_rate": 9.838069311974986e-05,
      "loss": 0.0225,
      "step": 572
    },
    {
      "epoch": 5.21,
      "learning_rate": 9.808630686771687e-05,
      "loss": 0.0245,
      "step": 573
    },
    {
      "epoch": 5.22,
      "learning_rate": 9.779193720555038e-05,
      "loss": 0.0254,
      "step": 574
    },
    {
      "epoch": 5.23,
      "learning_rate": 9.749758668515027e-05,
      "loss": 0.0195,
      "step": 575
    },
    {
      "epoch": 5.24,
      "learning_rate": 9.720325785825065e-05,
      "loss": 0.0255,
      "step": 576
    },
    {
      "epoch": 5.25,
      "learning_rate": 9.69089532763974e-05,
      "loss": 0.024,
      "step": 577
    },
    {
      "epoch": 5.25,
      "learning_rate": 9.66146754909263e-05,
      "loss": 0.0225,
      "step": 578
    },
    {
      "epoch": 5.26,
      "learning_rate": 9.632042705294081e-05,
      "loss": 0.0215,
      "step": 579
    },
    {
      "epoch": 5.27,
      "learning_rate": 9.602621051328996e-05,
      "loss": 0.0276,
      "step": 580
    },
    {
      "epoch": 5.28,
      "learning_rate": 9.573202842254627e-05,
      "loss": 0.024,
      "step": 581
    },
    {
      "epoch": 5.29,
      "learning_rate": 9.543788333098365e-05,
      "loss": 0.0227,
      "step": 582
    },
    {
      "epoch": 5.3,
      "learning_rate": 9.514377778855521e-05,
      "loss": 0.028,
      "step": 583
    },
    {
      "epoch": 5.31,
      "learning_rate": 9.484971434487119e-05,
      "loss": 0.0245,
      "step": 584
    },
    {
      "epoch": 5.32,
      "learning_rate": 9.4555695549177e-05,
      "loss": 0.0247,
      "step": 585
    },
    {
      "epoch": 5.33,
      "learning_rate": 9.426172395033085e-05,
      "loss": 0.0305,
      "step": 586
    },
    {
      "epoch": 5.34,
      "learning_rate": 9.396780209678191e-05,
      "loss": 0.026,
      "step": 587
    },
    {
      "epoch": 5.35,
      "learning_rate": 9.367393253654802e-05,
      "loss": 0.0262,
      "step": 588
    },
    {
      "epoch": 5.35,
      "learning_rate": 9.338011781719372e-05,
      "loss": 0.0247,
      "step": 589
    },
    {
      "epoch": 5.36,
      "learning_rate": 9.308636048580814e-05,
      "loss": 0.0249,
      "step": 590
    },
    {
      "epoch": 5.37,
      "learning_rate": 9.279266308898292e-05,
      "loss": 0.0266,
      "step": 591
    },
    {
      "epoch": 5.38,
      "learning_rate": 9.249902817279007e-05,
      "loss": 0.0234,
      "step": 592
    },
    {
      "epoch": 5.39,
      "learning_rate": 9.220545828276008e-05,
      "loss": 0.0218,
      "step": 593
    },
    {
      "epoch": 5.4,
      "learning_rate": 9.19119559638596e-05,
      "loss": 0.0238,
      "step": 594
    },
    {
      "epoch": 5.41,
      "learning_rate": 9.161852376046953e-05,
      "loss": 0.0252,
      "step": 595
    },
    {
      "epoch": 5.42,
      "learning_rate": 9.132516421636296e-05,
      "loss": 0.0218,
      "step": 596
    },
    {
      "epoch": 5.43,
      "learning_rate": 9.103187987468311e-05,
      "loss": 0.0309,
      "step": 597
    },
    {
      "epoch": 5.44,
      "learning_rate": 9.073867327792123e-05,
      "loss": 0.023,
      "step": 598
    },
    {
      "epoch": 5.45,
      "learning_rate": 9.044554696789462e-05,
      "loss": 0.0218,
      "step": 599
    },
    {
      "epoch": 5.45,
      "learning_rate": 9.015250348572452e-05,
      "loss": 0.0265,
      "step": 600
    },
    {
      "epoch": 5.46,
      "learning_rate": 8.985954537181422e-05,
      "loss": 0.0245,
      "step": 601
    },
    {
      "epoch": 5.47,
      "learning_rate": 8.956667516582694e-05,
      "loss": 0.0265,
      "step": 602
    },
    {
      "epoch": 5.48,
      "learning_rate": 8.927389540666373e-05,
      "loss": 0.0247,
      "step": 603
    },
    {
      "epoch": 5.49,
      "learning_rate": 8.898120863244164e-05,
      "loss": 0.0236,
      "step": 604
    },
    {
      "epoch": 5.5,
      "learning_rate": 8.868861738047158e-05,
      "loss": 0.0226,
      "step": 605
    },
    {
      "epoch": 5.51,
      "learning_rate": 8.839612418723643e-05,
      "loss": 0.0278,
      "step": 606
    },
    {
      "epoch": 5.52,
      "learning_rate": 8.810373158836897e-05,
      "loss": 0.027,
      "step": 607
    },
    {
      "epoch": 5.53,
      "learning_rate": 8.781144211862984e-05,
      "loss": 0.0296,
      "step": 608
    },
    {
      "epoch": 5.54,
      "learning_rate": 8.751925831188584e-05,
      "loss": 0.0207,
      "step": 609
    },
    {
      "epoch": 5.55,
      "learning_rate": 8.72271827010876e-05,
      "loss": 0.0294,
      "step": 610
    },
    {
      "epoch": 5.55,
      "learning_rate": 8.693521781824787e-05,
      "loss": 0.0179,
      "step": 611
    },
    {
      "epoch": 5.56,
      "learning_rate": 8.664336619441949e-05,
      "loss": 0.0246,
      "step": 612
    },
    {
      "epoch": 5.57,
      "learning_rate": 8.635163035967346e-05,
      "loss": 0.0219,
      "step": 613
    },
    {
      "epoch": 5.58,
      "learning_rate": 8.606001284307694e-05,
      "loss": 0.0239,
      "step": 614
    },
    {
      "epoch": 5.59,
      "learning_rate": 8.57685161726715e-05,
      "loss": 0.0293,
      "step": 615
    },
    {
      "epoch": 5.6,
      "learning_rate": 8.5477142875451e-05,
      "loss": 0.0264,
      "step": 616
    },
    {
      "epoch": 5.61,
      "learning_rate": 8.518589547733975e-05,
      "loss": 0.0196,
      "step": 617
    },
    {
      "epoch": 5.62,
      "learning_rate": 8.489477650317077e-05,
      "loss": 0.019,
      "step": 618
    },
    {
      "epoch": 5.63,
      "learning_rate": 8.460378847666366e-05,
      "loss": 0.0259,
      "step": 619
    },
    {
      "epoch": 5.64,
      "learning_rate": 8.431293392040282e-05,
      "loss": 0.0189,
      "step": 620
    },
    {
      "epoch": 5.65,
      "learning_rate": 8.402221535581568e-05,
      "loss": 0.0191,
      "step": 621
    },
    {
      "epoch": 5.65,
      "learning_rate": 8.373163530315068e-05,
      "loss": 0.0207,
      "step": 622
    },
    {
      "epoch": 5.66,
      "learning_rate": 8.344119628145552e-05,
      "loss": 0.0204,
      "step": 623
    },
    {
      "epoch": 5.67,
      "learning_rate": 8.315090080855526e-05,
      "loss": 0.0194,
      "step": 624
    },
    {
      "epoch": 5.68,
      "learning_rate": 8.286075140103058e-05,
      "loss": 0.0241,
      "step": 625
    },
    {
      "epoch": 5.69,
      "learning_rate": 8.257075057419595e-05,
      "loss": 0.0187,
      "step": 626
    },
    {
      "epoch": 5.7,
      "learning_rate": 8.228090084207774e-05,
      "loss": 0.0241,
      "step": 627
    },
    {
      "epoch": 5.71,
      "learning_rate": 8.199120471739242e-05,
      "loss": 0.0183,
      "step": 628
    },
    {
      "epoch": 5.72,
      "learning_rate": 8.170166471152491e-05,
      "loss": 0.0231,
      "step": 629
    },
    {
      "epoch": 5.73,
      "learning_rate": 8.141228333450673e-05,
      "loss": 0.0159,
      "step": 630
    },
    {
      "epoch": 5.74,
      "learning_rate": 8.112306309499422e-05,
      "loss": 0.0203,
      "step": 631
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.08340065002468e-05,
      "loss": 0.0296,
      "step": 632
    },
    {
      "epoch": 5.75,
      "learning_rate": 8.05451160561052e-05,
      "loss": 0.0246,
      "step": 633
    },
    {
      "epoch": 5.76,
      "learning_rate": 8.025639426696998e-05,
      "loss": 0.0215,
      "step": 634
    },
    {
      "epoch": 5.77,
      "learning_rate": 7.996784363577939e-05,
      "loss": 0.026,
      "step": 635
    },
    {
      "epoch": 5.78,
      "learning_rate": 7.967946666398807e-05,
      "loss": 0.0277,
      "step": 636
    },
    {
      "epoch": 5.79,
      "learning_rate": 7.939126585154511e-05,
      "loss": 0.0255,
      "step": 637
    },
    {
      "epoch": 5.8,
      "learning_rate": 7.91032436968725e-05,
      "loss": 0.0238,
      "step": 638
    },
    {
      "epoch": 5.81,
      "learning_rate": 7.881540269684344e-05,
      "loss": 0.0241,
      "step": 639
    },
    {
      "epoch": 5.82,
      "learning_rate": 7.852774534676071e-05,
      "loss": 0.0164,
      "step": 640
    },
    {
      "epoch": 5.83,
      "learning_rate": 7.824027414033495e-05,
      "loss": 0.0197,
      "step": 641
    },
    {
      "epoch": 5.84,
      "learning_rate": 7.795299156966327e-05,
      "loss": 0.0237,
      "step": 642
    },
    {
      "epoch": 5.85,
      "learning_rate": 7.766590012520732e-05,
      "loss": 0.0184,
      "step": 643
    },
    {
      "epoch": 5.85,
      "learning_rate": 7.737900229577194e-05,
      "loss": 0.022,
      "step": 644
    },
    {
      "epoch": 5.86,
      "learning_rate": 7.709230056848357e-05,
      "loss": 0.0274,
      "step": 645
    },
    {
      "epoch": 5.87,
      "learning_rate": 7.680579742876855e-05,
      "loss": 0.0219,
      "step": 646
    },
    {
      "epoch": 5.88,
      "learning_rate": 7.651949536033167e-05,
      "loss": 0.02,
      "step": 647
    },
    {
      "epoch": 5.89,
      "learning_rate": 7.623339684513466e-05,
      "loss": 0.0204,
      "step": 648
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.594750436337467e-05,
      "loss": 0.0228,
      "step": 649
    },
    {
      "epoch": 5.91,
      "learning_rate": 7.56618203934626e-05,
      "loss": 0.0229,
      "step": 650
    },
    {
      "epoch": 5.92,
      "learning_rate": 7.537634741200196e-05,
      "loss": 0.0262,
      "step": 651
    },
    {
      "epoch": 5.93,
      "learning_rate": 7.509108789376706e-05,
      "loss": 0.017,
      "step": 652
    },
    {
      "epoch": 5.94,
      "learning_rate": 7.480604431168166e-05,
      "loss": 0.0233,
      "step": 653
    },
    {
      "epoch": 5.95,
      "learning_rate": 7.452121913679768e-05,
      "loss": 0.0235,
      "step": 654
    },
    {
      "epoch": 5.95,
      "learning_rate": 7.423661483827357e-05,
      "loss": 0.0154,
      "step": 655
    },
    {
      "epoch": 5.96,
      "learning_rate": 7.395223388335305e-05,
      "loss": 0.0271,
      "step": 656
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.366807873734358e-05,
      "loss": 0.0233,
      "step": 657
    },
    {
      "epoch": 5.98,
      "learning_rate": 7.338415186359517e-05,
      "loss": 0.0189,
      "step": 658
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.310045572347895e-05,
      "loss": 0.0191,
      "step": 659
    },
    {
      "epoch": 6.0,
      "learning_rate": 7.281699277636572e-05,
      "loss": 0.0152,
      "step": 660
    },
    {
      "epoch": 6.01,
      "learning_rate": 7.253376547960477e-05,
      "loss": 0.0088,
      "step": 661
    },
    {
      "epoch": 6.02,
      "learning_rate": 7.225077628850254e-05,
      "loss": 0.0114,
      "step": 662
    },
    {
      "epoch": 6.03,
      "learning_rate": 7.196802765630124e-05,
      "loss": 0.0111,
      "step": 663
    },
    {
      "epoch": 6.04,
      "learning_rate": 7.168552203415783e-05,
      "loss": 0.0095,
      "step": 664
    },
    {
      "epoch": 6.05,
      "learning_rate": 7.140326187112246e-05,
      "loss": 0.011,
      "step": 665
    },
    {
      "epoch": 6.05,
      "learning_rate": 7.112124961411742e-05,
      "loss": 0.0094,
      "step": 666
    },
    {
      "epoch": 6.06,
      "learning_rate": 7.083948770791598e-05,
      "loss": 0.0065,
      "step": 667
    },
    {
      "epoch": 6.07,
      "learning_rate": 7.055797859512098e-05,
      "loss": 0.0092,
      "step": 668
    },
    {
      "epoch": 6.08,
      "learning_rate": 7.027672471614391e-05,
      "loss": 0.0066,
      "step": 669
    },
    {
      "epoch": 6.09,
      "learning_rate": 6.999572850918357e-05,
      "loss": 0.0085,
      "step": 670
    },
    {
      "epoch": 6.1,
      "learning_rate": 6.971499241020495e-05,
      "loss": 0.0095,
      "step": 671
    },
    {
      "epoch": 6.11,
      "learning_rate": 6.94345188529182e-05,
      "loss": 0.0092,
      "step": 672
    },
    {
      "epoch": 6.12,
      "learning_rate": 6.915431026875755e-05,
      "loss": 0.0086,
      "step": 673
    },
    {
      "epoch": 6.13,
      "learning_rate": 6.887436908686007e-05,
      "loss": 0.0068,
      "step": 674
    },
    {
      "epoch": 6.14,
      "learning_rate": 6.85946977340447e-05,
      "loss": 0.009,
      "step": 675
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.831529863479137e-05,
      "loss": 0.0096,
      "step": 676
    },
    {
      "epoch": 6.15,
      "learning_rate": 6.803617421121968e-05,
      "loss": 0.0085,
      "step": 677
    },
    {
      "epoch": 6.16,
      "learning_rate": 6.775732688306815e-05,
      "loss": 0.0075,
      "step": 678
    },
    {
      "epoch": 6.17,
      "learning_rate": 6.747875906767309e-05,
      "loss": 0.0068,
      "step": 679
    },
    {
      "epoch": 6.18,
      "learning_rate": 6.720047317994775e-05,
      "loss": 0.0077,
      "step": 680
    },
    {
      "epoch": 6.19,
      "learning_rate": 6.692247163236128e-05,
      "loss": 0.0068,
      "step": 681
    },
    {
      "epoch": 6.2,
      "learning_rate": 6.664475683491796e-05,
      "loss": 0.0079,
      "step": 682
    },
    {
      "epoch": 6.21,
      "learning_rate": 6.636733119513615e-05,
      "loss": 0.0098,
      "step": 683
    },
    {
      "epoch": 6.22,
      "learning_rate": 6.609019711802757e-05,
      "loss": 0.0083,
      "step": 684
    },
    {
      "epoch": 6.23,
      "learning_rate": 6.581335700607631e-05,
      "loss": 0.0088,
      "step": 685
    },
    {
      "epoch": 6.24,
      "learning_rate": 6.553681325921808e-05,
      "loss": 0.0059,
      "step": 686
    },
    {
      "epoch": 6.25,
      "learning_rate": 6.52605682748194e-05,
      "loss": 0.0056,
      "step": 687
    },
    {
      "epoch": 6.25,
      "learning_rate": 6.498462444765679e-05,
      "loss": 0.0081,
      "step": 688
    },
    {
      "epoch": 6.26,
      "learning_rate": 6.470898416989608e-05,
      "loss": 0.0078,
      "step": 689
    },
    {
      "epoch": 6.27,
      "learning_rate": 6.443364983107156e-05,
      "loss": 0.0079,
      "step": 690
    },
    {
      "epoch": 6.28,
      "learning_rate": 6.415862381806534e-05,
      "loss": 0.0088,
      "step": 691
    },
    {
      "epoch": 6.29,
      "learning_rate": 6.388390851508666e-05,
      "loss": 0.0067,
      "step": 692
    },
    {
      "epoch": 6.3,
      "learning_rate": 6.360950630365126e-05,
      "loss": 0.008,
      "step": 693
    },
    {
      "epoch": 6.31,
      "learning_rate": 6.333541956256055e-05,
      "loss": 0.0044,
      "step": 694
    },
    {
      "epoch": 6.32,
      "learning_rate": 6.306165066788121e-05,
      "loss": 0.0067,
      "step": 695
    },
    {
      "epoch": 6.33,
      "learning_rate": 6.278820199292446e-05,
      "loss": 0.0062,
      "step": 696
    },
    {
      "epoch": 6.34,
      "learning_rate": 6.251507590822557e-05,
      "loss": 0.0081,
      "step": 697
    },
    {
      "epoch": 6.35,
      "learning_rate": 6.224227478152322e-05,
      "loss": 0.0069,
      "step": 698
    },
    {
      "epoch": 6.35,
      "learning_rate": 6.1969800977739e-05,
      "loss": 0.0081,
      "step": 699
    },
    {
      "epoch": 6.36,
      "learning_rate": 6.169765685895703e-05,
      "loss": 0.0044,
      "step": 700
    },
    {
      "epoch": 6.37,
      "learning_rate": 6.142584478440328e-05,
      "loss": 0.0082,
      "step": 701
    },
    {
      "epoch": 6.38,
      "learning_rate": 6.115436711042523e-05,
      "loss": 0.0079,
      "step": 702
    },
    {
      "epoch": 6.39,
      "learning_rate": 6.088322619047151e-05,
      "loss": 0.0075,
      "step": 703
    },
    {
      "epoch": 6.4,
      "learning_rate": 6.061242437507131e-05,
      "loss": 0.0081,
      "step": 704
    },
    {
      "epoch": 6.41,
      "learning_rate": 6.034196401181413e-05,
      "loss": 0.0072,
      "step": 705
    },
    {
      "epoch": 6.42,
      "learning_rate": 6.007184744532952e-05,
      "loss": 0.0079,
      "step": 706
    },
    {
      "epoch": 6.43,
      "learning_rate": 5.980207701726654e-05,
      "loss": 0.0088,
      "step": 707
    },
    {
      "epoch": 6.44,
      "learning_rate": 5.953265506627355e-05,
      "loss": 0.0052,
      "step": 708
    },
    {
      "epoch": 6.45,
      "learning_rate": 5.926358392797804e-05,
      "loss": 0.0072,
      "step": 709
    },
    {
      "epoch": 6.45,
      "learning_rate": 5.899486593496625e-05,
      "loss": 0.0086,
      "step": 710
    },
    {
      "epoch": 6.46,
      "learning_rate": 5.8726503416762954e-05,
      "loss": 0.008,
      "step": 711
    },
    {
      "epoch": 6.47,
      "learning_rate": 5.845849869981137e-05,
      "loss": 0.0074,
      "step": 712
    },
    {
      "epoch": 6.48,
      "learning_rate": 5.8190854107452885e-05,
      "loss": 0.0108,
      "step": 713
    },
    {
      "epoch": 6.49,
      "learning_rate": 5.792357195990694e-05,
      "loss": 0.0098,
      "step": 714
    },
    {
      "epoch": 6.5,
      "learning_rate": 5.765665457425102e-05,
      "loss": 0.006,
      "step": 715
    },
    {
      "epoch": 6.51,
      "learning_rate": 5.739010426440028e-05,
      "loss": 0.0072,
      "step": 716
    },
    {
      "epoch": 6.52,
      "learning_rate": 5.712392334108799e-05,
      "loss": 0.0073,
      "step": 717
    },
    {
      "epoch": 6.53,
      "learning_rate": 5.6858114111844895e-05,
      "loss": 0.0072,
      "step": 718
    },
    {
      "epoch": 6.54,
      "learning_rate": 5.6592678880979674e-05,
      "loss": 0.0081,
      "step": 719
    },
    {
      "epoch": 6.55,
      "learning_rate": 5.632761994955881e-05,
      "loss": 0.0065,
      "step": 720
    },
    {
      "epoch": 6.55,
      "learning_rate": 5.6062939615386514e-05,
      "loss": 0.0058,
      "step": 721
    },
    {
      "epoch": 6.56,
      "learning_rate": 5.579864017298504e-05,
      "loss": 0.0056,
      "step": 722
    },
    {
      "epoch": 6.57,
      "learning_rate": 5.553472391357468e-05,
      "loss": 0.0088,
      "step": 723
    },
    {
      "epoch": 6.58,
      "learning_rate": 5.527119312505373e-05,
      "loss": 0.0078,
      "step": 724
    },
    {
      "epoch": 6.59,
      "learning_rate": 5.5008050091979155e-05,
      "loss": 0.0068,
      "step": 725
    },
    {
      "epoch": 6.6,
      "learning_rate": 5.474529709554612e-05,
      "loss": 0.0088,
      "step": 726
    },
    {
      "epoch": 6.61,
      "learning_rate": 5.448293641356872e-05,
      "loss": 0.0067,
      "step": 727
    },
    {
      "epoch": 6.62,
      "learning_rate": 5.4220970320460075e-05,
      "loss": 0.0083,
      "step": 728
    },
    {
      "epoch": 6.63,
      "learning_rate": 5.395940108721249e-05,
      "loss": 0.0068,
      "step": 729
    },
    {
      "epoch": 6.64,
      "learning_rate": 5.3698230981378026e-05,
      "loss": 0.009,
      "step": 730
    },
    {
      "epoch": 6.65,
      "learning_rate": 5.343746226704853e-05,
      "loss": 0.0054,
      "step": 731
    },
    {
      "epoch": 6.65,
      "learning_rate": 5.317709720483631e-05,
      "loss": 0.0063,
      "step": 732
    },
    {
      "epoch": 6.66,
      "learning_rate": 5.291713805185437e-05,
      "loss": 0.009,
      "step": 733
    },
    {
      "epoch": 6.67,
      "learning_rate": 5.265758706169681e-05,
      "loss": 0.006,
      "step": 734
    },
    {
      "epoch": 6.68,
      "learning_rate": 5.239844648441949e-05,
      "loss": 0.0055,
      "step": 735
    },
    {
      "epoch": 6.69,
      "learning_rate": 5.213971856652019e-05,
      "loss": 0.0105,
      "step": 736
    },
    {
      "epoch": 6.7,
      "learning_rate": 5.1881405550919493e-05,
      "loss": 0.0071,
      "step": 737
    },
    {
      "epoch": 6.71,
      "learning_rate": 5.1623509676941165e-05,
      "loss": 0.0076,
      "step": 738
    },
    {
      "epoch": 6.72,
      "learning_rate": 5.1366033180292636e-05,
      "loss": 0.0077,
      "step": 739
    },
    {
      "epoch": 6.73,
      "learning_rate": 5.1108978293045916e-05,
      "loss": 0.0093,
      "step": 740
    },
    {
      "epoch": 6.74,
      "learning_rate": 5.085234724361786e-05,
      "loss": 0.0096,
      "step": 741
    },
    {
      "epoch": 6.75,
      "learning_rate": 5.059614225675129e-05,
      "loss": 0.0104,
      "step": 742
    },
    {
      "epoch": 6.75,
      "learning_rate": 5.0340365553495374e-05,
      "loss": 0.0073,
      "step": 743
    },
    {
      "epoch": 6.76,
      "learning_rate": 5.008501935118639e-05,
      "loss": 0.007,
      "step": 744
    },
    {
      "epoch": 6.77,
      "learning_rate": 4.983010586342876e-05,
      "loss": 0.0076,
      "step": 745
    },
    {
      "epoch": 6.78,
      "learning_rate": 4.95756273000755e-05,
      "loss": 0.0093,
      "step": 746
    },
    {
      "epoch": 6.79,
      "learning_rate": 4.93215858672094e-05,
      "loss": 0.0065,
      "step": 747
    },
    {
      "epoch": 6.8,
      "learning_rate": 4.9067983767123736e-05,
      "loss": 0.0047,
      "step": 748
    },
    {
      "epoch": 6.81,
      "learning_rate": 4.881482319830306e-05,
      "loss": 0.0076,
      "step": 749
    },
    {
      "epoch": 6.82,
      "learning_rate": 4.856210635540451e-05,
      "loss": 0.0059,
      "step": 750
    },
    {
      "epoch": 6.83,
      "learning_rate": 4.830983542923833e-05,
      "loss": 0.0061,
      "step": 751
    },
    {
      "epoch": 6.84,
      "learning_rate": 4.80580126067492e-05,
      "loss": 0.0065,
      "step": 752
    },
    {
      "epoch": 6.85,
      "learning_rate": 4.780664007099721e-05,
      "loss": 0.0084,
      "step": 753
    },
    {
      "epoch": 6.85,
      "learning_rate": 4.755572000113877e-05,
      "loss": 0.0073,
      "step": 754
    },
    {
      "epoch": 6.86,
      "learning_rate": 4.730525457240796e-05,
      "loss": 0.007,
      "step": 755
    },
    {
      "epoch": 6.87,
      "learning_rate": 4.705524595609758e-05,
      "loss": 0.0074,
      "step": 756
    },
    {
      "epoch": 6.88,
      "learning_rate": 4.680569631954014e-05,
      "loss": 0.0074,
      "step": 757
    },
    {
      "epoch": 6.89,
      "learning_rate": 4.65566078260895e-05,
      "loss": 0.0071,
      "step": 758
    },
    {
      "epoch": 6.9,
      "learning_rate": 4.630798263510162e-05,
      "loss": 0.0068,
      "step": 759
    },
    {
      "epoch": 6.91,
      "learning_rate": 4.6059822901916226e-05,
      "loss": 0.0074,
      "step": 760
    },
    {
      "epoch": 6.92,
      "learning_rate": 4.581213077783787e-05,
      "loss": 0.009,
      "step": 761
    },
    {
      "epoch": 6.93,
      "learning_rate": 4.556490841011746e-05,
      "loss": 0.0074,
      "step": 762
    },
    {
      "epoch": 6.94,
      "learning_rate": 4.531815794193357e-05,
      "loss": 0.0051,
      "step": 763
    },
    {
      "epoch": 6.95,
      "learning_rate": 4.507188151237379e-05,
      "loss": 0.0103,
      "step": 764
    },
    {
      "epoch": 6.95,
      "learning_rate": 4.482608125641633e-05,
      "loss": 0.0076,
      "step": 765
    },
    {
      "epoch": 6.96,
      "learning_rate": 4.45807593049114e-05,
      "loss": 0.0084,
      "step": 766
    },
    {
      "epoch": 6.97,
      "learning_rate": 4.433591778456281e-05,
      "loss": 0.0059,
      "step": 767
    },
    {
      "epoch": 6.98,
      "learning_rate": 4.4091558817909476e-05,
      "loss": 0.0076,
      "step": 768
    },
    {
      "epoch": 6.99,
      "learning_rate": 4.384768452330698e-05,
      "loss": 0.0088,
      "step": 769
    },
    {
      "epoch": 7.0,
      "learning_rate": 4.360429701490934e-05,
      "loss": 0.0047,
      "step": 770
    },
    {
      "epoch": 7.01,
      "learning_rate": 4.336139840265065e-05,
      "loss": 0.0032,
      "step": 771
    },
    {
      "epoch": 7.02,
      "learning_rate": 4.311899079222661e-05,
      "loss": 0.0027,
      "step": 772
    },
    {
      "epoch": 7.03,
      "learning_rate": 4.2877076285076546e-05,
      "loss": 0.0023,
      "step": 773
    },
    {
      "epoch": 7.04,
      "learning_rate": 4.26356569783649e-05,
      "loss": 0.0025,
      "step": 774
    },
    {
      "epoch": 7.05,
      "learning_rate": 4.239473496496344e-05,
      "loss": 0.0031,
      "step": 775
    },
    {
      "epoch": 7.05,
      "learning_rate": 4.2154312333432646e-05,
      "loss": 0.0025,
      "step": 776
    },
    {
      "epoch": 7.06,
      "learning_rate": 4.191439116800397e-05,
      "loss": 0.0027,
      "step": 777
    },
    {
      "epoch": 7.07,
      "learning_rate": 4.167497354856166e-05,
      "loss": 0.004,
      "step": 778
    },
    {
      "epoch": 7.08,
      "learning_rate": 4.143606155062457e-05,
      "loss": 0.0029,
      "step": 779
    },
    {
      "epoch": 7.09,
      "learning_rate": 4.1197657245328425e-05,
      "loss": 0.0026,
      "step": 780
    },
    {
      "epoch": 7.1,
      "learning_rate": 4.0959762699407766e-05,
      "loss": 0.0029,
      "step": 781
    },
    {
      "epoch": 7.11,
      "learning_rate": 4.0722379975177835e-05,
      "loss": 0.0026,
      "step": 782
    },
    {
      "epoch": 7.12,
      "learning_rate": 4.048551113051716e-05,
      "loss": 0.0034,
      "step": 783
    },
    {
      "epoch": 7.13,
      "learning_rate": 4.024915821884916e-05,
      "loss": 0.0021,
      "step": 784
    },
    {
      "epoch": 7.14,
      "learning_rate": 4.001332328912475e-05,
      "loss": 0.0032,
      "step": 785
    },
    {
      "epoch": 7.15,
      "learning_rate": 3.977800838580448e-05,
      "loss": 0.0028,
      "step": 786
    },
    {
      "epoch": 7.15,
      "learning_rate": 3.9543215548840614e-05,
      "loss": 0.0028,
      "step": 787
    },
    {
      "epoch": 7.16,
      "learning_rate": 3.930894681365981e-05,
      "loss": 0.0022,
      "step": 788
    },
    {
      "epoch": 7.17,
      "learning_rate": 3.907520421114508e-05,
      "loss": 0.0028,
      "step": 789
    },
    {
      "epoch": 7.18,
      "learning_rate": 3.884198976761846e-05,
      "loss": 0.0022,
      "step": 790
    },
    {
      "epoch": 7.19,
      "learning_rate": 3.860930550482347e-05,
      "loss": 0.0026,
      "step": 791
    },
    {
      "epoch": 7.2,
      "learning_rate": 3.8377153439907266e-05,
      "loss": 0.0028,
      "step": 792
    },
    {
      "epoch": 7.21,
      "learning_rate": 3.8145535585403537e-05,
      "loss": 0.0029,
      "step": 793
    },
    {
      "epoch": 7.22,
      "learning_rate": 3.791445394921474e-05,
      "loss": 0.0032,
      "step": 794
    },
    {
      "epoch": 7.23,
      "learning_rate": 3.768391053459496e-05,
      "loss": 0.0022,
      "step": 795
    },
    {
      "epoch": 7.24,
      "learning_rate": 3.7453907340132385e-05,
      "loss": 0.0023,
      "step": 796
    },
    {
      "epoch": 7.25,
      "learning_rate": 3.722444635973196e-05,
      "loss": 0.0027,
      "step": 797
    },
    {
      "epoch": 7.25,
      "learning_rate": 3.6995529582598234e-05,
      "loss": 0.002,
      "step": 798
    },
    {
      "epoch": 7.26,
      "learning_rate": 3.676715899321801e-05,
      "loss": 0.0046,
      "step": 799
    },
    {
      "epoch": 7.27,
      "learning_rate": 3.6539336571343174e-05,
      "loss": 0.0025,
      "step": 800
    },
    {
      "epoch": 7.28,
      "learning_rate": 3.631206429197357e-05,
      "loss": 0.0027,
      "step": 801
    },
    {
      "epoch": 7.29,
      "learning_rate": 3.6085344125339726e-05,
      "loss": 0.0024,
      "step": 802
    },
    {
      "epoch": 7.3,
      "learning_rate": 3.585917803688603e-05,
      "loss": 0.0037,
      "step": 803
    },
    {
      "epoch": 7.31,
      "learning_rate": 3.5633567987253424e-05,
      "loss": 0.0022,
      "step": 804
    },
    {
      "epoch": 7.32,
      "learning_rate": 3.540851593226261e-05,
      "loss": 0.0027,
      "step": 805
    },
    {
      "epoch": 7.33,
      "learning_rate": 3.518402382289704e-05,
      "loss": 0.0025,
      "step": 806
    },
    {
      "epoch": 7.34,
      "learning_rate": 3.496009360528581e-05,
      "loss": 0.002,
      "step": 807
    },
    {
      "epoch": 7.35,
      "learning_rate": 3.4736727220687234e-05,
      "loss": 0.003,
      "step": 808
    },
    {
      "epoch": 7.35,
      "learning_rate": 3.45139266054715e-05,
      "loss": 0.002,
      "step": 809
    },
    {
      "epoch": 7.36,
      "learning_rate": 3.429169369110422e-05,
      "loss": 0.0023,
      "step": 810
    },
    {
      "epoch": 7.37,
      "learning_rate": 3.407003040412964e-05,
      "loss": 0.0022,
      "step": 811
    },
    {
      "epoch": 7.38,
      "learning_rate": 3.384893866615374e-05,
      "loss": 0.0026,
      "step": 812
    },
    {
      "epoch": 7.39,
      "learning_rate": 3.362842039382785e-05,
      "loss": 0.0029,
      "step": 813
    },
    {
      "epoch": 7.4,
      "learning_rate": 3.340847749883191e-05,
      "loss": 0.0023,
      "step": 814
    },
    {
      "epoch": 7.41,
      "learning_rate": 3.3189111887857774e-05,
      "loss": 0.0035,
      "step": 815
    },
    {
      "epoch": 7.42,
      "learning_rate": 3.297032546259299e-05,
      "loss": 0.0025,
      "step": 816
    },
    {
      "epoch": 7.43,
      "learning_rate": 3.2752120119703925e-05,
      "loss": 0.002,
      "step": 817
    },
    {
      "epoch": 7.44,
      "learning_rate": 3.253449775081967e-05,
      "loss": 0.0024,
      "step": 818
    },
    {
      "epoch": 7.45,
      "learning_rate": 3.2317460242515474e-05,
      "loss": 0.0029,
      "step": 819
    },
    {
      "epoch": 7.45,
      "learning_rate": 3.2101009476296304e-05,
      "loss": 0.0018,
      "step": 820
    },
    {
      "epoch": 7.46,
      "learning_rate": 3.1885147328580824e-05,
      "loss": 0.0026,
      "step": 821
    },
    {
      "epoch": 7.47,
      "learning_rate": 3.166987567068473e-05,
      "loss": 0.0026,
      "step": 822
    },
    {
      "epoch": 7.48,
      "learning_rate": 3.145519636880493e-05,
      "loss": 0.0035,
      "step": 823
    },
    {
      "epoch": 7.49,
      "learning_rate": 3.124111128400308e-05,
      "loss": 0.0022,
      "step": 824
    },
    {
      "epoch": 7.5,
      "learning_rate": 3.102762227218957e-05,
      "loss": 0.0028,
      "step": 825
    },
    {
      "epoch": 7.51,
      "learning_rate": 3.081473118410747e-05,
      "loss": 0.0025,
      "step": 826
    },
    {
      "epoch": 7.52,
      "learning_rate": 3.060243986531631e-05,
      "loss": 0.0024,
      "step": 827
    },
    {
      "epoch": 7.53,
      "learning_rate": 3.0390750156176294e-05,
      "loss": 0.0023,
      "step": 828
    },
    {
      "epoch": 7.54,
      "learning_rate": 3.0179663891832266e-05,
      "loss": 0.0031,
      "step": 829
    },
    {
      "epoch": 7.55,
      "learning_rate": 2.996918290219769e-05,
      "loss": 0.0021,
      "step": 830
    },
    {
      "epoch": 7.55,
      "learning_rate": 2.9759309011938997e-05,
      "loss": 0.0033,
      "step": 831
    },
    {
      "epoch": 7.56,
      "learning_rate": 2.955004404045951e-05,
      "loss": 0.0028,
      "step": 832
    },
    {
      "epoch": 7.57,
      "learning_rate": 2.934138980188401e-05,
      "loss": 0.0032,
      "step": 833
    },
    {
      "epoch": 7.58,
      "learning_rate": 2.9133348105042713e-05,
      "loss": 0.0022,
      "step": 834
    },
    {
      "epoch": 7.59,
      "learning_rate": 2.8925920753455625e-05,
      "loss": 0.0027,
      "step": 835
    },
    {
      "epoch": 7.6,
      "learning_rate": 2.8719109545317103e-05,
      "loss": 0.0021,
      "step": 836
    },
    {
      "epoch": 7.61,
      "learning_rate": 2.8512916273479996e-05,
      "loss": 0.0021,
      "step": 837
    },
    {
      "epoch": 7.62,
      "learning_rate": 2.830734272544038e-05,
      "loss": 0.0046,
      "step": 838
    },
    {
      "epoch": 7.63,
      "learning_rate": 2.8102390683321878e-05,
      "loss": 0.0022,
      "step": 839
    },
    {
      "epoch": 7.64,
      "learning_rate": 2.7898061923860153e-05,
      "loss": 0.002,
      "step": 840
    },
    {
      "epoch": 7.65,
      "learning_rate": 2.769435821838785e-05,
      "loss": 0.0018,
      "step": 841
    },
    {
      "epoch": 7.65,
      "learning_rate": 2.749128133281874e-05,
      "loss": 0.002,
      "step": 842
    },
    {
      "epoch": 7.66,
      "learning_rate": 2.7288833027632833e-05,
      "loss": 0.0021,
      "step": 843
    },
    {
      "epoch": 7.67,
      "learning_rate": 2.7087015057860944e-05,
      "loss": 0.0028,
      "step": 844
    },
    {
      "epoch": 7.68,
      "learning_rate": 2.6885829173069375e-05,
      "loss": 0.0021,
      "step": 845
    },
    {
      "epoch": 7.69,
      "learning_rate": 2.6685277117345008e-05,
      "loss": 0.0021,
      "step": 846
    },
    {
      "epoch": 7.7,
      "learning_rate": 2.6485360629279987e-05,
      "loss": 0.0017,
      "step": 847
    },
    {
      "epoch": 7.71,
      "learning_rate": 2.62860814419566e-05,
      "loss": 0.0023,
      "step": 848
    },
    {
      "epoch": 7.72,
      "learning_rate": 2.6087441282932578e-05,
      "loss": 0.0023,
      "step": 849
    },
    {
      "epoch": 7.73,
      "learning_rate": 2.588944187422567e-05,
      "loss": 0.002,
      "step": 850
    },
    {
      "epoch": 7.74,
      "learning_rate": 2.56920849322991e-05,
      "loss": 0.0019,
      "step": 851
    },
    {
      "epoch": 7.75,
      "learning_rate": 2.5495372168046383e-05,
      "loss": 0.0025,
      "step": 852
    },
    {
      "epoch": 7.75,
      "learning_rate": 2.5299305286776764e-05,
      "loss": 0.0019,
      "step": 853
    },
    {
      "epoch": 7.76,
      "learning_rate": 2.5103885988200272e-05,
      "loss": 0.0023,
      "step": 854
    },
    {
      "epoch": 7.77,
      "learning_rate": 2.4909115966412954e-05,
      "loss": 0.0019,
      "step": 855
    },
    {
      "epoch": 7.78,
      "learning_rate": 2.471499690988235e-05,
      "loss": 0.0022,
      "step": 856
    },
    {
      "epoch": 7.79,
      "learning_rate": 2.452153050143271e-05,
      "loss": 0.0021,
      "step": 857
    },
    {
      "epoch": 7.8,
      "learning_rate": 2.432871841823047e-05,
      "loss": 0.0018,
      "step": 858
    },
    {
      "epoch": 7.81,
      "learning_rate": 2.413656233176973e-05,
      "loss": 0.0026,
      "step": 859
    },
    {
      "epoch": 7.82,
      "learning_rate": 2.3945063907857646e-05,
      "loss": 0.0021,
      "step": 860
    },
    {
      "epoch": 7.83,
      "learning_rate": 2.3754224806600157e-05,
      "loss": 0.0021,
      "step": 861
    },
    {
      "epoch": 7.84,
      "learning_rate": 2.356404668238751e-05,
      "loss": 0.0021,
      "step": 862
    },
    {
      "epoch": 7.85,
      "learning_rate": 2.3374531183879854e-05,
      "loss": 0.0021,
      "step": 863
    },
    {
      "epoch": 7.85,
      "learning_rate": 2.318567995399311e-05,
      "loss": 0.0022,
      "step": 864
    },
    {
      "epoch": 7.86,
      "learning_rate": 2.299749462988451e-05,
      "loss": 0.0023,
      "step": 865
    },
    {
      "epoch": 7.87,
      "learning_rate": 2.2809976842938696e-05,
      "loss": 0.0029,
      "step": 866
    },
    {
      "epoch": 7.88,
      "learning_rate": 2.2623128218753242e-05,
      "loss": 0.0018,
      "step": 867
    },
    {
      "epoch": 7.89,
      "learning_rate": 2.243695037712482e-05,
      "loss": 0.0024,
      "step": 868
    },
    {
      "epoch": 7.9,
      "learning_rate": 2.2251444932035094e-05,
      "loss": 0.0018,
      "step": 869
    },
    {
      "epoch": 7.91,
      "learning_rate": 2.2066613491636566e-05,
      "loss": 0.0023,
      "step": 870
    },
    {
      "epoch": 7.92,
      "learning_rate": 2.1882457658238907e-05,
      "loss": 0.0021,
      "step": 871
    },
    {
      "epoch": 7.93,
      "learning_rate": 2.1698979028294918e-05,
      "loss": 0.0024,
      "step": 872
    },
    {
      "epoch": 7.94,
      "learning_rate": 2.1516179192386564e-05,
      "loss": 0.0025,
      "step": 873
    },
    {
      "epoch": 7.95,
      "learning_rate": 2.133405973521154e-05,
      "loss": 0.002,
      "step": 874
    },
    {
      "epoch": 7.95,
      "learning_rate": 2.115262223556912e-05,
      "loss": 0.0017,
      "step": 875
    },
    {
      "epoch": 7.96,
      "learning_rate": 2.0971868266346773e-05,
      "loss": 0.002,
      "step": 876
    },
    {
      "epoch": 7.97,
      "learning_rate": 2.0791799394506415e-05,
      "loss": 0.0024,
      "step": 877
    },
    {
      "epoch": 7.98,
      "learning_rate": 2.0612417181070764e-05,
      "loss": 0.0021,
      "step": 878
    },
    {
      "epoch": 7.99,
      "learning_rate": 2.0433723181109966e-05,
      "loss": 0.0018,
      "step": 879
    },
    {
      "epoch": 8.0,
      "learning_rate": 2.025571894372794e-05,
      "loss": 0.0019,
      "step": 880
    },
    {
      "epoch": 8.01,
      "learning_rate": 2.0078406012049054e-05,
      "loss": 0.0015,
      "step": 881
    },
    {
      "epoch": 8.02,
      "learning_rate": 1.990178592320482e-05,
      "loss": 0.0014,
      "step": 882
    },
    {
      "epoch": 8.03,
      "learning_rate": 1.9725860208320324e-05,
      "loss": 0.0014,
      "step": 883
    },
    {
      "epoch": 8.04,
      "learning_rate": 1.9550630392501225e-05,
      "loss": 0.0013,
      "step": 884
    },
    {
      "epoch": 8.05,
      "learning_rate": 1.9376097994820285e-05,
      "loss": 0.0017,
      "step": 885
    },
    {
      "epoch": 8.05,
      "learning_rate": 1.9202264528304447e-05,
      "loss": 0.0015,
      "step": 886
    },
    {
      "epoch": 8.06,
      "learning_rate": 1.902913149992156e-05,
      "loss": 0.0016,
      "step": 887
    },
    {
      "epoch": 8.07,
      "learning_rate": 1.8856700410567297e-05,
      "loss": 0.0018,
      "step": 888
    },
    {
      "epoch": 8.08,
      "learning_rate": 1.8684972755052266e-05,
      "loss": 0.0015,
      "step": 889
    },
    {
      "epoch": 8.09,
      "learning_rate": 1.8513950022088965e-05,
      "loss": 0.0014,
      "step": 890
    },
    {
      "epoch": 8.1,
      "learning_rate": 1.8343633694278895e-05,
      "loss": 0.0015,
      "step": 891
    },
    {
      "epoch": 8.11,
      "learning_rate": 1.8174025248099736e-05,
      "loss": 0.0014,
      "step": 892
    },
    {
      "epoch": 8.12,
      "learning_rate": 1.8005126153892428e-05,
      "loss": 0.0014,
      "step": 893
    },
    {
      "epoch": 8.13,
      "learning_rate": 1.7836937875848625e-05,
      "loss": 0.0015,
      "step": 894
    },
    {
      "epoch": 8.14,
      "learning_rate": 1.7669461871997818e-05,
      "loss": 0.0015,
      "step": 895
    },
    {
      "epoch": 8.15,
      "learning_rate": 1.7502699594194803e-05,
      "loss": 0.0015,
      "step": 896
    },
    {
      "epoch": 8.15,
      "learning_rate": 1.7336652488107108e-05,
      "loss": 0.0014,
      "step": 897
    },
    {
      "epoch": 8.16,
      "learning_rate": 1.7171321993202272e-05,
      "loss": 0.0014,
      "step": 898
    },
    {
      "epoch": 8.17,
      "learning_rate": 1.70067095427357e-05,
      "loss": 0.0016,
      "step": 899
    },
    {
      "epoch": 8.18,
      "learning_rate": 1.6842816563737873e-05,
      "loss": 0.0015,
      "step": 900
    },
    {
      "epoch": 8.19,
      "learning_rate": 1.6679644477002266e-05,
      "loss": 0.0014,
      "step": 901
    },
    {
      "epoch": 8.2,
      "learning_rate": 1.65171946970729e-05,
      "loss": 0.0014,
      "step": 902
    },
    {
      "epoch": 8.21,
      "learning_rate": 1.6355468632232007e-05,
      "loss": 0.0013,
      "step": 903
    },
    {
      "epoch": 8.22,
      "learning_rate": 1.6194467684488025e-05,
      "loss": 0.0015,
      "step": 904
    },
    {
      "epoch": 8.23,
      "learning_rate": 1.603419324956328e-05,
      "loss": 0.0016,
      "step": 905
    },
    {
      "epoch": 8.24,
      "learning_rate": 1.587464671688187e-05,
      "loss": 0.0014,
      "step": 906
    },
    {
      "epoch": 8.25,
      "learning_rate": 1.5715829469557853e-05,
      "loss": 0.0013,
      "step": 907
    },
    {
      "epoch": 8.25,
      "learning_rate": 1.555774288438291e-05,
      "loss": 0.0014,
      "step": 908
    },
    {
      "epoch": 8.26,
      "learning_rate": 1.5400388331814687e-05,
      "loss": 0.0015,
      "step": 909
    },
    {
      "epoch": 8.27,
      "learning_rate": 1.5243767175964818e-05,
      "loss": 0.0012,
      "step": 910
    },
    {
      "epoch": 8.28,
      "learning_rate": 1.5087880774587026e-05,
      "loss": 0.0015,
      "step": 911
    },
    {
      "epoch": 8.29,
      "learning_rate": 1.4932730479065515e-05,
      "loss": 0.0013,
      "step": 912
    },
    {
      "epoch": 8.3,
      "learning_rate": 1.4778317634403083e-05,
      "loss": 0.0016,
      "step": 913
    },
    {
      "epoch": 8.31,
      "learning_rate": 1.4624643579209606e-05,
      "loss": 0.0014,
      "step": 914
    },
    {
      "epoch": 8.32,
      "learning_rate": 1.4471709645690334e-05,
      "loss": 0.0015,
      "step": 915
    },
    {
      "epoch": 8.33,
      "learning_rate": 1.4319517159634388e-05,
      "loss": 0.0014,
      "step": 916
    },
    {
      "epoch": 8.34,
      "learning_rate": 1.4168067440403299e-05,
      "loss": 0.0014,
      "step": 917
    },
    {
      "epoch": 8.35,
      "learning_rate": 1.4017361800919426e-05,
      "loss": 0.0014,
      "step": 918
    },
    {
      "epoch": 8.35,
      "learning_rate": 1.386740154765478e-05,
      "loss": 0.0013,
      "step": 919
    },
    {
      "epoch": 8.36,
      "learning_rate": 1.3718187980619557e-05,
      "loss": 0.0025,
      "step": 920
    },
    {
      "epoch": 8.37,
      "learning_rate": 1.3569722393350881e-05,
      "loss": 0.0015,
      "step": 921
    },
    {
      "epoch": 8.38,
      "learning_rate": 1.342200607290165e-05,
      "loss": 0.0014,
      "step": 922
    },
    {
      "epoch": 8.39,
      "learning_rate": 1.3275040299829333e-05,
      "loss": 0.0013,
      "step": 923
    },
    {
      "epoch": 8.4,
      "learning_rate": 1.3128826348184887e-05,
      "loss": 0.0014,
      "step": 924
    },
    {
      "epoch": 8.41,
      "learning_rate": 1.2983365485501719e-05,
      "loss": 0.0014,
      "step": 925
    },
    {
      "epoch": 8.42,
      "learning_rate": 1.2838658972784622e-05,
      "loss": 0.0014,
      "step": 926
    },
    {
      "epoch": 8.43,
      "learning_rate": 1.2694708064498983e-05,
      "loss": 0.0014,
      "step": 927
    },
    {
      "epoch": 8.44,
      "learning_rate": 1.255151400855975e-05,
      "loss": 0.0013,
      "step": 928
    },
    {
      "epoch": 8.45,
      "learning_rate": 1.2409078046320777e-05,
      "loss": 0.0013,
      "step": 929
    },
    {
      "epoch": 8.45,
      "learning_rate": 1.226740141256395e-05,
      "loss": 0.0014,
      "step": 930
    },
    {
      "epoch": 8.46,
      "learning_rate": 1.2126485335488436e-05,
      "loss": 0.0014,
      "step": 931
    },
    {
      "epoch": 8.47,
      "learning_rate": 1.1986331036700271e-05,
      "loss": 0.0016,
      "step": 932
    },
    {
      "epoch": 8.48,
      "learning_rate": 1.1846939731201456e-05,
      "loss": 0.0015,
      "step": 933
    },
    {
      "epoch": 8.49,
      "learning_rate": 1.1708312627379647e-05,
      "loss": 0.0015,
      "step": 934
    },
    {
      "epoch": 8.5,
      "learning_rate": 1.1570450926997655e-05,
      "loss": 0.0014,
      "step": 935
    },
    {
      "epoch": 8.51,
      "learning_rate": 1.1433355825182868e-05,
      "loss": 0.0013,
      "step": 936
    },
    {
      "epoch": 8.52,
      "learning_rate": 1.1297028510417108e-05,
      "loss": 0.0015,
      "step": 937
    },
    {
      "epoch": 8.53,
      "learning_rate": 1.1161470164526222e-05,
      "loss": 0.0013,
      "step": 938
    },
    {
      "epoch": 8.54,
      "learning_rate": 1.1026681962669726e-05,
      "loss": 0.0013,
      "step": 939
    },
    {
      "epoch": 8.55,
      "learning_rate": 1.0892665073330931e-05,
      "loss": 0.0013,
      "step": 940
    },
    {
      "epoch": 8.55,
      "learning_rate": 1.0759420658306418e-05,
      "loss": 0.0012,
      "step": 941
    },
    {
      "epoch": 8.56,
      "learning_rate": 1.0626949872696301e-05,
      "loss": 0.0013,
      "step": 942
    },
    {
      "epoch": 8.57,
      "learning_rate": 1.0495253864893983e-05,
      "loss": 0.0012,
      "step": 943
    },
    {
      "epoch": 8.58,
      "learning_rate": 1.0364333776576352e-05,
      "loss": 0.0014,
      "step": 944
    },
    {
      "epoch": 8.59,
      "learning_rate": 1.0234190742693838e-05,
      "loss": 0.0014,
      "step": 945
    },
    {
      "epoch": 8.6,
      "learning_rate": 1.010482589146048e-05,
      "loss": 0.0015,
      "step": 946
    },
    {
      "epoch": 8.61,
      "learning_rate": 9.976240344344301e-06,
      "loss": 0.0013,
      "step": 947
    },
    {
      "epoch": 8.62,
      "learning_rate": 9.848435216057494e-06,
      "loss": 0.0013,
      "step": 948
    },
    {
      "epoch": 8.63,
      "learning_rate": 9.721411614546772e-06,
      "loss": 0.0014,
      "step": 949
    },
    {
      "epoch": 8.64,
      "learning_rate": 9.595170640983786e-06,
      "loss": 0.0016,
      "step": 950
    },
    {
      "epoch": 8.65,
      "learning_rate": 9.469713389755508e-06,
      "loss": 0.0014,
      "step": 951
    },
    {
      "epoch": 8.65,
      "learning_rate": 9.345040948454842e-06,
      "loss": 0.0014,
      "step": 952
    },
    {
      "epoch": 8.66,
      "learning_rate": 9.221154397871157e-06,
      "loss": 0.0013,
      "step": 953
    },
    {
      "epoch": 8.67,
      "learning_rate": 9.09805481198085e-06,
      "loss": 0.0014,
      "step": 954
    },
    {
      "epoch": 8.68,
      "learning_rate": 8.975743257938186e-06,
      "loss": 0.0012,
      "step": 955
    },
    {
      "epoch": 8.69,
      "learning_rate": 8.854220796065849e-06,
      "loss": 0.0016,
      "step": 956
    },
    {
      "epoch": 8.7,
      "learning_rate": 8.733488479845997e-06,
      "loss": 0.0015,
      "step": 957
    },
    {
      "epoch": 8.71,
      "learning_rate": 8.613547355910846e-06,
      "loss": 0.0014,
      "step": 958
    },
    {
      "epoch": 8.72,
      "learning_rate": 8.494398464033814e-06,
      "loss": 0.0012,
      "step": 959
    },
    {
      "epoch": 8.73,
      "learning_rate": 8.376042837120479e-06,
      "loss": 0.0014,
      "step": 960
    },
    {
      "epoch": 8.74,
      "learning_rate": 8.258481501199455e-06,
      "loss": 0.0014,
      "step": 961
    },
    {
      "epoch": 8.75,
      "learning_rate": 8.141715475413725e-06,
      "loss": 0.0015,
      "step": 962
    },
    {
      "epoch": 8.75,
      "learning_rate": 8.025745772011695e-06,
      "loss": 0.0014,
      "step": 963
    },
    {
      "epoch": 8.76,
      "learning_rate": 7.91057339633835e-06,
      "loss": 0.0016,
      "step": 964
    },
    {
      "epoch": 8.77,
      "learning_rate": 7.796199346826726e-06,
      "loss": 0.0012,
      "step": 965
    },
    {
      "epoch": 8.78,
      "learning_rate": 7.682624614989065e-06,
      "loss": 0.0016,
      "step": 966
    },
    {
      "epoch": 8.79,
      "learning_rate": 7.5698501854083006e-06,
      "loss": 0.0014,
      "step": 967
    },
    {
      "epoch": 8.8,
      "learning_rate": 7.457877035729588e-06,
      "loss": 0.0014,
      "step": 968
    },
    {
      "epoch": 8.81,
      "learning_rate": 7.346706136651671e-06,
      "loss": 0.0014,
      "step": 969
    },
    {
      "epoch": 8.82,
      "learning_rate": 7.236338451918634e-06,
      "loss": 0.0015,
      "step": 970
    },
    {
      "epoch": 8.83,
      "learning_rate": 7.126774938311409e-06,
      "loss": 0.0012,
      "step": 971
    },
    {
      "epoch": 8.84,
      "learning_rate": 7.018016545639572e-06,
      "loss": 0.0014,
      "step": 972
    },
    {
      "epoch": 8.85,
      "learning_rate": 6.910064216733137e-06,
      "loss": 0.0015,
      "step": 973
    },
    {
      "epoch": 8.85,
      "learning_rate": 6.802918887434195e-06,
      "loss": 0.0024,
      "step": 974
    },
    {
      "epoch": 8.86,
      "learning_rate": 6.696581486589071e-06,
      "loss": 0.0014,
      "step": 975
    },
    {
      "epoch": 8.87,
      "learning_rate": 6.5910529360400094e-06,
      "loss": 0.0013,
      "step": 976
    },
    {
      "epoch": 8.88,
      "learning_rate": 6.48633415061739e-06,
      "loss": 0.0014,
      "step": 977
    },
    {
      "epoch": 8.89,
      "learning_rate": 6.382426038131706e-06,
      "loss": 0.0015,
      "step": 978
    },
    {
      "epoch": 8.9,
      "learning_rate": 6.2793294993656494e-06,
      "loss": 0.0014,
      "step": 979
    },
    {
      "epoch": 8.91,
      "learning_rate": 6.177045428066397e-06,
      "loss": 0.0013,
      "step": 980
    },
    {
      "epoch": 8.92,
      "learning_rate": 6.075574710937803e-06,
      "loss": 0.0014,
      "step": 981
    },
    {
      "epoch": 8.93,
      "learning_rate": 5.974918227632753e-06,
      "loss": 0.0013,
      "step": 982
    },
    {
      "epoch": 8.94,
      "learning_rate": 5.875076850745509e-06,
      "loss": 0.0014,
      "step": 983
    },
    {
      "epoch": 8.95,
      "learning_rate": 5.776051445804087e-06,
      "loss": 0.0013,
      "step": 984
    },
    {
      "epoch": 8.95,
      "learning_rate": 5.677842871262895e-06,
      "loss": 0.0013,
      "step": 985
    },
    {
      "epoch": 8.96,
      "learning_rate": 5.5804519784951825e-06,
      "loss": 0.0014,
      "step": 986
    },
    {
      "epoch": 8.97,
      "learning_rate": 5.483879611785681e-06,
      "loss": 0.0013,
      "step": 987
    },
    {
      "epoch": 8.98,
      "learning_rate": 5.3881266083233185e-06,
      "loss": 0.0014,
      "step": 988
    },
    {
      "epoch": 8.99,
      "learning_rate": 5.293193798193874e-06,
      "loss": 0.0014,
      "step": 989
    },
    {
      "epoch": 9.0,
      "learning_rate": 5.199082004372957e-06,
      "loss": 0.0012,
      "step": 990
    },
    {
      "epoch": 9.01,
      "learning_rate": 5.105792042718649e-06,
      "loss": 0.0013,
      "step": 991
    },
    {
      "epoch": 9.02,
      "learning_rate": 5.013324721964607e-06,
      "loss": 0.0012,
      "step": 992
    },
    {
      "epoch": 9.03,
      "learning_rate": 4.921680843712972e-06,
      "loss": 0.0013,
      "step": 993
    },
    {
      "epoch": 9.04,
      "learning_rate": 4.8308612024274165e-06,
      "loss": 0.0013,
      "step": 994
    },
    {
      "epoch": 9.05,
      "learning_rate": 4.740866585426307e-06,
      "loss": 0.0012,
      "step": 995
    },
    {
      "epoch": 9.05,
      "learning_rate": 4.651697772875829e-06,
      "loss": 0.0012,
      "step": 996
    },
    {
      "epoch": 9.06,
      "learning_rate": 4.563355537783231e-06,
      "loss": 0.0012,
      "step": 997
    },
    {
      "epoch": 9.07,
      "learning_rate": 4.475840645990193e-06,
      "loss": 0.0012,
      "step": 998
    },
    {
      "epoch": 9.08,
      "learning_rate": 4.38915385616605e-06,
      "loss": 0.0014,
      "step": 999
    },
    {
      "epoch": 9.09,
      "learning_rate": 4.303295919801364e-06,
      "loss": 0.0013,
      "step": 1000
    },
    {
      "epoch": 9.1,
      "learning_rate": 4.2182675812012965e-06,
      "loss": 0.0012,
      "step": 1001
    },
    {
      "epoch": 9.11,
      "learning_rate": 4.134069577479205e-06,
      "loss": 0.0013,
      "step": 1002
    },
    {
      "epoch": 9.12,
      "learning_rate": 4.050702638550275e-06,
      "loss": 0.0013,
      "step": 1003
    },
    {
      "epoch": 9.13,
      "learning_rate": 3.9681674871251185e-06,
      "loss": 0.0013,
      "step": 1004
    },
    {
      "epoch": 9.14,
      "learning_rate": 3.886464838703607e-06,
      "loss": 0.0012,
      "step": 1005
    },
    {
      "epoch": 9.15,
      "learning_rate": 3.8055954015685736e-06,
      "loss": 0.0012,
      "step": 1006
    },
    {
      "epoch": 9.15,
      "learning_rate": 3.7255598767797494e-06,
      "loss": 0.0013,
      "step": 1007
    },
    {
      "epoch": 9.16,
      "learning_rate": 3.6463589581676506e-06,
      "loss": 0.0012,
      "step": 1008
    },
    {
      "epoch": 9.17,
      "learning_rate": 3.567993332327524e-06,
      "loss": 0.0012,
      "step": 1009
    },
    {
      "epoch": 9.18,
      "learning_rate": 3.490463678613487e-06,
      "loss": 0.0012,
      "step": 1010
    },
    {
      "epoch": 9.19,
      "learning_rate": 3.4137706691325654e-06,
      "loss": 0.0014,
      "step": 1011
    },
    {
      "epoch": 9.2,
      "learning_rate": 3.3379149687388867e-06,
      "loss": 0.0012,
      "step": 1012
    },
    {
      "epoch": 9.21,
      "learning_rate": 3.2628972350279173e-06,
      "loss": 0.0012,
      "step": 1013
    },
    {
      "epoch": 9.22,
      "learning_rate": 3.1887181183307913e-06,
      "loss": 0.0013,
      "step": 1014
    },
    {
      "epoch": 9.23,
      "learning_rate": 3.1153782617086126e-06,
      "loss": 0.0013,
      "step": 1015
    },
    {
      "epoch": 9.24,
      "learning_rate": 3.0428783009469385e-06,
      "loss": 0.0013,
      "step": 1016
    },
    {
      "epoch": 9.25,
      "learning_rate": 2.971218864550207e-06,
      "loss": 0.0012,
      "step": 1017
    },
    {
      "epoch": 9.25,
      "learning_rate": 2.900400573736373e-06,
      "loss": 0.0012,
      "step": 1018
    },
    {
      "epoch": 9.26,
      "learning_rate": 2.8304240424314233e-06,
      "loss": 0.0013,
      "step": 1019
    },
    {
      "epoch": 9.27,
      "learning_rate": 2.7612898772641393e-06,
      "loss": 0.0013,
      "step": 1020
    },
    {
      "epoch": 9.28,
      "learning_rate": 2.6929986775607985e-06,
      "loss": 0.0013,
      "step": 1021
    },
    {
      "epoch": 9.29,
      "learning_rate": 2.625551035339957e-06,
      "loss": 0.0012,
      "step": 1022
    },
    {
      "epoch": 9.3,
      "learning_rate": 2.5589475353073988e-06,
      "loss": 0.0013,
      "step": 1023
    },
    {
      "epoch": 9.31,
      "learning_rate": 2.4931887548509502e-06,
      "loss": 0.0012,
      "step": 1024
    },
    {
      "epoch": 9.32,
      "learning_rate": 2.4282752640355844e-06,
      "loss": 0.0012,
      "step": 1025
    },
    {
      "epoch": 9.33,
      "learning_rate": 2.3642076255984023e-06,
      "loss": 0.0013,
      "step": 1026
    },
    {
      "epoch": 9.34,
      "learning_rate": 2.300986394943794e-06,
      "loss": 0.0012,
      "step": 1027
    },
    {
      "epoch": 9.35,
      "learning_rate": 2.238612120138617e-06,
      "loss": 0.0012,
      "step": 1028
    },
    {
      "epoch": 9.35,
      "learning_rate": 2.177085341907459e-06,
      "loss": 0.0012,
      "step": 1029
    },
    {
      "epoch": 9.36,
      "learning_rate": 2.116406593627873e-06,
      "loss": 0.0013,
      "step": 1030
    },
    {
      "epoch": 9.37,
      "learning_rate": 2.0565764013258826e-06,
      "loss": 0.0013,
      "step": 1031
    },
    {
      "epoch": 9.38,
      "learning_rate": 1.9975952836713165e-06,
      "loss": 0.0012,
      "step": 1032
    },
    {
      "epoch": 9.39,
      "learning_rate": 1.9394637519733583e-06,
      "loss": 0.0012,
      "step": 1033
    },
    {
      "epoch": 9.4,
      "learning_rate": 1.882182310176095e-06,
      "loss": 0.0013,
      "step": 1034
    },
    {
      "epoch": 9.41,
      "learning_rate": 1.8257514548541633e-06,
      "loss": 0.0013,
      "step": 1035
    },
    {
      "epoch": 9.42,
      "learning_rate": 1.770171675208454e-06,
      "loss": 0.0012,
      "step": 1036
    },
    {
      "epoch": 9.43,
      "learning_rate": 1.7154434530618157e-06,
      "loss": 0.0012,
      "step": 1037
    },
    {
      "epoch": 9.44,
      "learning_rate": 1.6615672628549573e-06,
      "loss": 0.0013,
      "step": 1038
    },
    {
      "epoch": 9.45,
      "learning_rate": 1.6085435716422847e-06,
      "loss": 0.0012,
      "step": 1039
    },
    {
      "epoch": 9.45,
      "learning_rate": 1.5563728390878497e-06,
      "loss": 0.0013,
      "step": 1040
    },
    {
      "epoch": 9.46,
      "learning_rate": 1.5050555174613957e-06,
      "loss": 0.0013,
      "step": 1041
    },
    {
      "epoch": 9.47,
      "learning_rate": 1.454592051634407e-06,
      "loss": 0.0014,
      "step": 1042
    },
    {
      "epoch": 9.48,
      "learning_rate": 1.4049828790762776e-06,
      "loss": 0.0012,
      "step": 1043
    },
    {
      "epoch": 9.49,
      "learning_rate": 1.3562284298505034e-06,
      "loss": 0.0012,
      "step": 1044
    },
    {
      "epoch": 9.5,
      "learning_rate": 1.30832912661093e-06,
      "loss": 0.0012,
      "step": 1045
    },
    {
      "epoch": 9.51,
      "learning_rate": 1.2612853845981654e-06,
      "loss": 0.0014,
      "step": 1046
    },
    {
      "epoch": 9.52,
      "learning_rate": 1.2150976116358737e-06,
      "loss": 0.0012,
      "step": 1047
    },
    {
      "epoch": 9.53,
      "learning_rate": 1.169766208127332e-06,
      "loss": 0.0013,
      "step": 1048
    },
    {
      "epoch": 9.54,
      "learning_rate": 1.1252915670519448e-06,
      "loss": 0.0012,
      "step": 1049
    },
    {
      "epoch": 9.55,
      "learning_rate": 1.081674073961747e-06,
      "loss": 0.0012,
      "step": 1050
    },
    {
      "epoch": 9.55,
      "learning_rate": 1.038914106978195e-06,
      "loss": 0.0013,
      "step": 1051
    },
    {
      "epoch": 9.56,
      "learning_rate": 9.970120367887914e-07,
      "loss": 0.0013,
      "step": 1052
    },
    {
      "epoch": 9.57,
      "learning_rate": 9.559682266439108e-07,
      "loss": 0.0012,
      "step": 1053
    },
    {
      "epoch": 9.58,
      "learning_rate": 9.157830323536565e-07,
      "loss": 0.0012,
      "step": 1054
    },
    {
      "epoch": 9.59,
      "learning_rate": 8.764568022847197e-07,
      "loss": 0.0013,
      "step": 1055
    },
    {
      "epoch": 9.6,
      "learning_rate": 8.379898773574924e-07,
      "loss": 0.0013,
      "step": 1056
    },
    {
      "epoch": 9.61,
      "learning_rate": 8.003825910429252e-07,
      "loss": 0.0013,
      "step": 1057
    },
    {
      "epoch": 9.62,
      "learning_rate": 7.636352693597859e-07,
      "loss": 0.0012,
      "step": 1058
    },
    {
      "epoch": 9.63,
      "learning_rate": 7.277482308717831e-07,
      "loss": 0.0013,
      "step": 1059
    },
    {
      "epoch": 9.64,
      "learning_rate": 6.927217866847691e-07,
      "loss": 0.0012,
      "step": 1060
    },
    {
      "epoch": 9.65,
      "learning_rate": 6.585562404441303e-07,
      "loss": 0.0013,
      "step": 1061
    },
    {
      "epoch": 9.65,
      "learning_rate": 6.252518883320569e-07,
      "loss": 0.0013,
      "step": 1062
    },
    {
      "epoch": 9.66,
      "learning_rate": 5.928090190650215e-07,
      "loss": 0.0013,
      "step": 1063
    },
    {
      "epoch": 9.67,
      "learning_rate": 5.612279138913157e-07,
      "loss": 0.0012,
      "step": 1064
    },
    {
      "epoch": 9.68,
      "learning_rate": 5.305088465885067e-07,
      "loss": 0.0012,
      "step": 1065
    },
    {
      "epoch": 9.69,
      "learning_rate": 5.006520834612061e-07,
      "loss": 0.0013,
      "step": 1066
    },
    {
      "epoch": 9.7,
      "learning_rate": 4.7165788333860536e-07,
      "loss": 0.0013,
      "step": 1067
    },
    {
      "epoch": 9.71,
      "learning_rate": 4.435264975723552e-07,
      "loss": 0.0012,
      "step": 1068
    },
    {
      "epoch": 9.72,
      "learning_rate": 4.162581700343449e-07,
      "loss": 0.0012,
      "step": 1069
    },
    {
      "epoch": 9.73,
      "learning_rate": 3.8985313711455973e-07,
      "loss": 0.0013,
      "step": 1070
    },
    {
      "epoch": 9.74,
      "learning_rate": 3.643116277190606e-07,
      "loss": 0.0012,
      "step": 1071
    },
    {
      "epoch": 9.75,
      "learning_rate": 3.396338632680185e-07,
      "loss": 0.0012,
      "step": 1072
    },
    {
      "epoch": 9.75,
      "learning_rate": 3.1582005769374976e-07,
      "loss": 0.0013,
      "step": 1073
    },
    {
      "epoch": 9.76,
      "learning_rate": 2.928704174388952e-07,
      "loss": 0.0012,
      "step": 1074
    },
    {
      "epoch": 9.77,
      "learning_rate": 2.707851414545992e-07,
      "loss": 0.0013,
      "step": 1075
    },
    {
      "epoch": 9.78,
      "learning_rate": 2.495644211988335e-07,
      "loss": 0.0013,
      "step": 1076
    },
    {
      "epoch": 9.79,
      "learning_rate": 2.2920844063470948e-07,
      "loss": 0.0013,
      "step": 1077
    },
    {
      "epoch": 9.8,
      "learning_rate": 2.0971737622883515e-07,
      "loss": 0.0012,
      "step": 1078
    },
    {
      "epoch": 9.81,
      "learning_rate": 1.9109139694989396e-07,
      "loss": 0.0013,
      "step": 1079
    },
    {
      "epoch": 9.82,
      "learning_rate": 1.7333066426706846e-07,
      "loss": 0.0012,
      "step": 1080
    },
    {
      "epoch": 9.83,
      "learning_rate": 1.5643533214874107e-07,
      "loss": 0.0012,
      "step": 1081
    },
    {
      "epoch": 9.84,
      "learning_rate": 1.4040554706105103e-07,
      "loss": 0.0013,
      "step": 1082
    },
    {
      "epoch": 9.85,
      "learning_rate": 1.2524144796673965e-07,
      "loss": 0.0013,
      "step": 1083
    },
    {
      "epoch": 9.85,
      "learning_rate": 1.1094316632386247e-07,
      "loss": 0.0012,
      "step": 1084
    },
    {
      "epoch": 9.86,
      "learning_rate": 9.751082608465689e-08,
      "loss": 0.0013,
      "step": 1085
    },
    {
      "epoch": 9.87,
      "learning_rate": 8.494454369452065e-08,
      "loss": 0.0013,
      "step": 1086
    },
    {
      "epoch": 9.88,
      "learning_rate": 7.324442809095721e-08,
      "loss": 0.0013,
      "step": 1087
    },
    {
      "epoch": 9.89,
      "learning_rate": 6.241058070263205e-08,
      "loss": 0.0013,
      "step": 1088
    },
    {
      "epoch": 9.9,
      "learning_rate": 5.2443095448506674e-08,
      "loss": 0.0013,
      "step": 1089
    },
    {
      "epoch": 9.91,
      "learning_rate": 4.334205873705033e-08,
      "loss": 0.0013,
      "step": 1090
    },
    {
      "epoch": 9.92,
      "learning_rate": 3.51075494654074e-08,
      "loss": 0.0012,
      "step": 1091
    },
    {
      "epoch": 9.93,
      "learning_rate": 2.7739639018831143e-08,
      "loss": 0.0014,
      "step": 1092
    },
    {
      "epoch": 9.94,
      "learning_rate": 2.1238391269950975e-08,
      "loss": 0.0013,
      "step": 1093
    },
    {
      "epoch": 9.95,
      "learning_rate": 1.5603862578295047e-08,
      "loss": 0.0013,
      "step": 1094
    },
    {
      "epoch": 9.95,
      "learning_rate": 1.083610178976846e-08,
      "loss": 0.0013,
      "step": 1095
    },
    {
      "epoch": 9.96,
      "learning_rate": 6.935150236253574e-09,
      "loss": 0.0014,
      "step": 1096
    },
    {
      "epoch": 9.97,
      "learning_rate": 3.901041735210331e-09,
      "loss": 0.0012,
      "step": 1097
    },
    {
      "epoch": 9.98,
      "learning_rate": 1.7338025894209075e-09,
      "loss": 0.0013,
      "step": 1098
    },
    {
      "epoch": 9.99,
      "learning_rate": 4.3345158675656137e-10,
      "loss": 0.0012,
      "step": 1099
    },
    {
      "epoch": 10.0,
      "learning_rate": 0.0,
      "loss": 0.0012,
      "step": 1100
    },
    {
      "epoch": 10.0,
      "step": 1100,
      "total_flos": 81867413422080.0,
      "train_loss": 0.25055770045273346,
      "train_runtime": 8133.9591,
      "train_samples_per_second": 8.606,
      "train_steps_per_second": 0.135
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1100,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 81867413422080.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}