{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.0,
  "global_step": 2088,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 10.858,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.1746031746031745e-05,
      "loss": 11.0409,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.761904761904762e-05,
      "loss": 10.0276,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.349206349206349e-05,
      "loss": 9.4209,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 7.936507936507937e-05,
      "loss": 9.0528,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.523809523809524e-05,
      "loss": 9.011,
      "step": 6
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001111111111111111,
      "loss": 8.6485,
      "step": 7
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00012698412698412698,
      "loss": 7.7262,
      "step": 8
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00014285714285714284,
      "loss": 7.4301,
      "step": 9
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00015873015873015873,
      "loss": 7.1522,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001746031746031746,
      "loss": 6.6217,
      "step": 11
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019047619047619048,
      "loss": 5.9846,
      "step": 12
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00020634920634920634,
      "loss": 5.5675,
      "step": 13
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002222222222222222,
      "loss": 5.0319,
      "step": 14
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0002380952380952381,
      "loss": 4.492,
      "step": 15
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00025396825396825396,
      "loss": 3.9963,
      "step": 16
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002698412698412698,
      "loss": 3.477,
      "step": 17
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0002857142857142857,
      "loss": 3.0997,
      "step": 18
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003015873015873016,
      "loss": 2.8757,
      "step": 19
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00031746031746031746,
      "loss": 2.7135,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003333333333333333,
      "loss": 2.5498,
      "step": 21
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0003492063492063492,
      "loss": 2.4897,
      "step": 22
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00036507936507936505,
      "loss": 2.2522,
      "step": 23
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00038095238095238096,
      "loss": 2.1594,
      "step": 24
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0003968253968253968,
      "loss": 2.147,
      "step": 25
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0004126984126984127,
      "loss": 2.0602,
      "step": 26
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00042857142857142855,
      "loss": 2.0704,
      "step": 27
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0004444444444444444,
      "loss": 1.9008,
      "step": 28
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00046031746031746033,
      "loss": 1.96,
      "step": 29
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0004761904761904762,
      "loss": 1.8308,
      "step": 30
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.000492063492063492,
      "loss": 2.003,
      "step": 31
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0005079365079365079,
      "loss": 1.9223,
      "step": 32
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0005238095238095238,
      "loss": 1.8372,
      "step": 33
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0005396825396825396,
      "loss": 1.9004,
      "step": 34
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0005555555555555556,
      "loss": 1.7686,
      "step": 35
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0005714285714285714,
      "loss": 1.6898,
      "step": 36
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0005873015873015874,
      "loss": 1.768,
      "step": 37
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0006031746031746032,
      "loss": 1.734,
      "step": 38
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0006190476190476191,
      "loss": 1.8663,
      "step": 39
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0006349206349206349,
      "loss": 1.7024,
      "step": 40
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0006507936507936508,
      "loss": 1.802,
      "step": 41
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0006666666666666666,
      "loss": 1.7592,
      "step": 42
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0006825396825396826,
      "loss": 1.8061,
      "step": 43
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0006984126984126984,
      "loss": 1.6674,
      "step": 44
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0007142857142857143,
      "loss": 1.7368,
      "step": 45
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0007301587301587301,
      "loss": 1.6775,
      "step": 46
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.000746031746031746,
      "loss": 1.6514,
      "step": 47
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0007619047619047619,
      "loss": 1.5911,
      "step": 48
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0007777777777777778,
      "loss": 1.6751,
      "step": 49
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0007936507936507937,
      "loss": 1.729,
      "step": 50
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0008095238095238096,
      "loss": 1.6754,
      "step": 51
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0008253968253968254,
      "loss": 1.6405,
      "step": 52
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0008412698412698413,
      "loss": 1.6685,
      "step": 53
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0008571428571428571,
      "loss": 1.6634,
      "step": 54
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.000873015873015873,
      "loss": 1.6521,
      "step": 55
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0008888888888888888,
      "loss": 1.6919,
      "step": 56
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0009047619047619047,
      "loss": 1.617,
      "step": 57
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0009206349206349207,
      "loss": 1.5883,
      "step": 58
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0009365079365079366,
      "loss": 1.6086,
      "step": 59
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0009523809523809524,
      "loss": 1.667,
      "step": 60
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0009682539682539683,
      "loss": 1.5496,
      "step": 61
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.000984126984126984,
      "loss": 1.5906,
      "step": 62
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.001,
      "loss": 1.6258,
      "step": 63
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0009999993982866989,
      "loss": 1.6176,
      "step": 64
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0009999975931482437,
      "loss": 1.6628,
      "step": 65
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0009999945845889794,
      "loss": 1.6323,
      "step": 66
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0009999903726161468,
      "loss": 1.5887,
      "step": 67
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.000999984957239884,
      "loss": 1.6214,
      "step": 68
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0009999783384732242,
      "loss": 1.6315,
      "step": 69
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0009999705163320986,
      "loss": 1.5949,
      "step": 70
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0009999614908353335,
      "loss": 1.5244,
      "step": 71
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0009999512620046521,
      "loss": 1.5821,
      "step": 72
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0009999398298646738,
      "loss": 1.5052,
      "step": 73
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0009999271944429139,
      "loss": 1.5884,
      "step": 74
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.000999913355769784,
      "loss": 1.5788,
      "step": 75
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0009998983138785917,
      "loss": 1.5456,
      "step": 76
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0009998820688055408,
      "loss": 1.6339,
      "step": 77
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.000999864620589731,
      "loss": 1.5465,
      "step": 78
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.000999845969273157,
      "loss": 1.5348,
      "step": 79
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0009998261149007104,
      "loss": 1.4573,
      "step": 80
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.000999805057520177,
      "loss": 1.5221,
      "step": 81
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0009997827971822398,
      "loss": 1.5501,
      "step": 82
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0009997593339404756,
      "loss": 1.4671,
      "step": 83
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0009997346678513569,
      "loss": 1.5013,
      "step": 84
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0009997087989742515,
      "loss": 1.4871,
      "step": 85
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.000999681727371422,
      "loss": 1.6084,
      "step": 86
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.000999653453108026,
      "loss": 1.5493,
      "step": 87
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.000999623976252115,
      "loss": 1.4833,
      "step": 88
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.000999593296874636,
      "loss": 1.4842,
      "step": 89
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0009995614150494292,
      "loss": 1.4867,
      "step": 90
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00099952833085323,
      "loss": 1.5356,
      "step": 91
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0009994940443656668,
      "loss": 1.5109,
      "step": 92
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0009994585556692623,
      "loss": 1.5295,
      "step": 93
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0009994218648494326,
      "loss": 1.5424,
      "step": 94
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0009993839719944872,
      "loss": 1.4976,
      "step": 95
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0009993448771956285,
      "loss": 1.4943,
      "step": 96
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.000999304580546952,
      "loss": 1.5548,
      "step": 97
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0009992630821454458,
      "loss": 1.4616,
      "step": 98
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0009992203820909905,
      "loss": 1.5114,
      "step": 99
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0009991764804863589,
      "loss": 1.4019,
      "step": 100
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0009991313774372155,
      "loss": 1.4938,
      "step": 101
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.000999085073052117,
      "loss": 1.5183,
      "step": 102
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0009990375674425109,
      "loss": 1.5439,
      "step": 103
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0009989888607227367,
      "loss": 1.6146,
      "step": 104
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.000998938953010024,
      "loss": 1.4827,
      "step": 105
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0009988878444244936,
      "loss": 1.5365,
      "step": 106
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0009988355350891557,
      "loss": 1.5265,
      "step": 107
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0009987820251299122,
      "loss": 1.4903,
      "step": 108
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0009987273146755528,
      "loss": 1.4756,
      "step": 109
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0009986714038577582,
      "loss": 1.5931,
      "step": 110
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.000998614292811097,
      "loss": 1.524,
      "step": 111
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0009985559816730277,
      "loss": 1.5054,
      "step": 112
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.000998496470583896,
      "loss": 1.3419,
      "step": 113
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0009984357596869368,
      "loss": 1.5558,
      "step": 114
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0009983738491282721,
      "loss": 1.5496,
      "step": 115
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0009983107390569117,
      "loss": 1.4731,
      "step": 116
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0009982464296247522,
      "loss": 1.4511,
      "step": 117
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.000998180920986577,
      "loss": 1.4041,
      "step": 118
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0009981142133000556,
      "loss": 1.5442,
      "step": 119
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0009980463067257438,
      "loss": 1.4804,
      "step": 120
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0009979772014270824,
      "loss": 1.4939,
      "step": 121
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0009979068975703982,
      "loss": 1.4178,
      "step": 122
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0009978353953249022,
      "loss": 1.4467,
      "step": 123
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0009977626948626897,
      "loss": 1.4355,
      "step": 124
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0009976887963587398,
      "loss": 1.4709,
      "step": 125
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0009976136999909156,
      "loss": 1.4561,
      "step": 126
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.000997537405939963,
      "loss": 1.4689,
      "step": 127
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0009974599143895107,
      "loss": 1.4292,
      "step": 128
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0009973812255260693,
      "loss": 1.4712,
      "step": 129
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0009973013395390314,
      "loss": 1.4202,
      "step": 130
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0009972202566206707,
      "loss": 1.4553,
      "step": 131
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0009971379769661423,
      "loss": 1.4914,
      "step": 132
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0009970545007734807,
      "loss": 1.5284,
      "step": 133
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0009969698282436012,
      "loss": 1.3419,
      "step": 134
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0009968839595802983,
      "loss": 1.4314,
      "step": 135
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0009967968949902448,
      "loss": 1.4597,
      "step": 136
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0009967086346829926,
      "loss": 1.4891,
      "step": 137
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0009966191788709714,
      "loss": 1.4204,
      "step": 138
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0009965285277694883,
      "loss": 1.3908,
      "step": 139
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.000996436681596727,
      "loss": 1.3788,
      "step": 140
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0009963436405737477,
      "loss": 1.4607,
      "step": 141
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0009962494049244864,
      "loss": 1.4215,
      "step": 142
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0009961539748757548,
      "loss": 1.4706,
      "step": 143
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.000996057350657239,
      "loss": 1.4528,
      "step": 144
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0009959595325014989,
      "loss": 1.4364,
      "step": 145
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.000995860520643969,
      "loss": 1.4469,
      "step": 146
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.000995760315322956,
      "loss": 1.4317,
      "step": 147
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0009956589167796391,
      "loss": 1.4125,
      "step": 148
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0009955563252580703,
      "loss": 1.4262,
      "step": 149
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.000995452541005172,
      "loss": 1.3844,
      "step": 150
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0009953475642707377,
      "loss": 1.361,
      "step": 151
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0009952413953074311,
      "loss": 1.4443,
      "step": 152
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0009951340343707852,
      "loss": 1.4266,
      "step": 153
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.000995025481719202,
      "loss": 1.4816,
      "step": 154
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0009949157376139518,
      "loss": 1.4021,
      "step": 155
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0009948048023191726,
      "loss": 1.4975,
      "step": 156
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0009946926761018695,
      "loss": 1.5046,
      "step": 157
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0009945793592319135,
      "loss": 1.5169,
      "step": 158
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.000994464851982042,
      "loss": 1.4054,
      "step": 159
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.000994349154627857,
      "loss": 1.4676,
      "step": 160
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.000994232267447825,
      "loss": 1.3813,
      "step": 161
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0009941141907232765,
      "loss": 1.437,
      "step": 162
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0009939949247384046,
      "loss": 1.4665,
      "step": 163
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.000993874469780265,
      "loss": 1.4689,
      "step": 164
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0009937528261387753,
      "loss": 1.483,
      "step": 165
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0009936299941067138,
      "loss": 1.4469,
      "step": 166
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0009935059739797191,
      "loss": 1.4709,
      "step": 167
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0009933807660562897,
      "loss": 1.4012,
      "step": 168
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0009932543706377822,
      "loss": 1.4211,
      "step": 169
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0009931267880284125,
      "loss": 1.4526,
      "step": 170
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0009929980185352525,
      "loss": 1.4187,
      "step": 171
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0009928680624682318,
      "loss": 1.3645,
      "step": 172
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0009927369201401357,
      "loss": 1.4388,
      "step": 173
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0009926045918666044,
      "loss": 1.407,
      "step": 174
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0009924710779661327,
      "loss": 1.3987,
      "step": 175
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0009923363787600688,
      "loss": 1.4674,
      "step": 176
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.000992200494572614,
      "loss": 1.3855,
      "step": 177
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0009920634257308217,
      "loss": 1.4326,
      "step": 178
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0009919251725645963,
      "loss": 1.4156,
      "step": 179
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.000991785735406693,
      "loss": 1.4109,
      "step": 180
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0009916451145927164,
      "loss": 1.489,
      "step": 181
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0009915033104611205,
      "loss": 1.407,
      "step": 182
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0009913603233532068,
      "loss": 1.3062,
      "step": 183
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.000991216153613124,
      "loss": 1.2807,
      "step": 184
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.000991070801587868,
      "loss": 1.427,
      "step": 185
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0009909242676272795,
      "loss": 1.3424,
      "step": 186
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0009907765520840446,
      "loss": 1.4024,
      "step": 187
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0009906276553136924,
      "loss": 1.353,
      "step": 188
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0009904775776745957,
      "loss": 1.4293,
      "step": 189
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0009903263195279698,
      "loss": 1.4562,
      "step": 190
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0009901738812378702,
      "loss": 1.3866,
      "step": 191
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.000990020263171194,
      "loss": 1.4124,
      "step": 192
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.000989865465697677,
      "loss": 1.4368,
      "step": 193
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0009897094891898942,
      "loss": 1.426,
      "step": 194
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0009895523340232582,
      "loss": 1.3765,
      "step": 195
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.000989394000576018,
      "loss": 1.4103,
      "step": 196
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0009892344892292596,
      "loss": 1.4581,
      "step": 197
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0009890738003669028,
      "loss": 1.4795,
      "step": 198
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0009889119343757025,
      "loss": 1.4765,
      "step": 199
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0009887488916452464,
      "loss": 1.3762,
      "step": 200
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0009885846725679538,
      "loss": 1.4633,
      "step": 201
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.000988419277539077,
      "loss": 1.3823,
      "step": 202
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0009882527069566965,
      "loss": 1.4494,
      "step": 203
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0009880849612217237,
      "loss": 1.3239,
      "step": 204
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0009879160407378982,
      "loss": 1.3756,
      "step": 205
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0009877459459117863,
      "loss": 1.411,
      "step": 206
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0009875746771527815,
      "loss": 1.3943,
      "step": 207
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0009874022348731028,
      "loss": 1.4134,
      "step": 208
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.000987228619487793,
      "loss": 1.3435,
      "step": 209
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0009870538314147193,
      "loss": 1.3784,
      "step": 210
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0009868778710745707,
      "loss": 1.4834,
      "step": 211
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0009867007388908579,
      "loss": 1.3204,
      "step": 212
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0009865224352899118,
      "loss": 1.4958,
      "step": 213
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0009863429607008835,
      "loss": 1.3921,
      "step": 214
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.000986162315555742,
      "loss": 1.3797,
      "step": 215
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0009859805002892731,
      "loss": 1.3843,
      "step": 216
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0009857975153390798,
      "loss": 1.3499,
      "step": 217
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0009856133611455802,
      "loss": 1.3883,
      "step": 218
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.000985428038152006,
      "loss": 1.3232,
      "step": 219
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0009852415468044027,
      "loss": 1.2649,
      "step": 220
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0009850538875516275,
      "loss": 1.464,
      "step": 221
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0009848650608453489,
      "loss": 1.4095,
      "step": 222
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0009846750671400446,
      "loss": 1.3725,
      "step": 223
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.000984483906893002,
      "loss": 1.4044,
      "step": 224
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0009842915805643156,
      "loss": 1.4382,
      "step": 225
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0009840980886168865,
      "loss": 1.4694,
      "step": 226
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0009839034315164217,
      "loss": 1.3952,
      "step": 227
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.000983707609731432,
      "loss": 1.4813,
      "step": 228
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0009835106237332319,
      "loss": 1.2885,
      "step": 229
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0009833124739959374,
      "loss": 1.2921,
      "step": 230
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0009831131609964665,
      "loss": 1.3938,
      "step": 231
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0009829126852145356,
      "loss": 1.3763,
      "step": 232
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.000982711047132661,
      "loss": 1.3415,
      "step": 233
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0009825082472361558,
      "loss": 1.3092,
      "step": 234
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0009823042860131294,
      "loss": 1.3952,
      "step": 235
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0009820991639544869,
      "loss": 1.3381,
      "step": 236
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0009818928815539265,
      "loss": 1.3368,
      "step": 237
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0009816854393079402,
      "loss": 1.3213,
      "step": 238
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0009814768377158106,
      "loss": 1.3887,
      "step": 239
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0009812670772796114,
      "loss": 1.3018,
      "step": 240
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0009810561585042049,
      "loss": 1.3719,
      "step": 241
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0009808440818972418,
      "loss": 1.4556,
      "step": 242
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0009806308479691594,
      "loss": 1.3795,
      "step": 243
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0009804164572331804,
      "loss": 1.3805,
      "step": 244
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0009802009102053118,
      "loss": 1.4197,
      "step": 245
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0009799842074043438,
      "loss": 1.3661,
      "step": 246
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.000979766349351848,
      "loss": 1.4526,
      "step": 247
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.000979547336572177,
      "loss": 1.3615,
      "step": 248
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0009793271695924621,
      "loss": 1.4038,
      "step": 249
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0009791058489426133,
      "loss": 1.3945,
      "step": 250
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0009788833751553164,
      "loss": 1.4041,
      "step": 251
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0009786597487660335,
      "loss": 1.4274,
      "step": 252
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0009784349703130008,
      "loss": 1.4079,
      "step": 253
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0009782090403372262,
      "loss": 1.3579,
      "step": 254
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0009779819593824907,
      "loss": 1.3967,
      "step": 255
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0009777537279953446,
      "loss": 1.4134,
      "step": 256
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0009775243467251074,
      "loss": 1.3772,
      "step": 257
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.000977293816123866,
      "loss": 1.3835,
      "step": 258
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0009770621367464738,
      "loss": 1.3109,
      "step": 259
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.000976829309150549,
      "loss": 1.3386,
      "step": 260
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0009765953338964734,
      "loss": 1.3822,
      "step": 261
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0009763602115473914,
      "loss": 1.3712,
      "step": 262
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0009761239426692076,
      "loss": 1.4224,
      "step": 263
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0009758865278305869,
      "loss": 1.3868,
      "step": 264
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0009756479676029515,
      "loss": 1.3193,
      "step": 265
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.000975408262560481,
      "loss": 1.345,
      "step": 266
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0009751674132801106,
      "loss": 1.4077,
      "step": 267
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0009749254203415288,
      "loss": 1.313,
      "step": 268
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0009746822843271771,
      "loss": 1.3724,
      "step": 269
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0009744380058222483,
      "loss": 1.4058,
      "step": 270
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.000974192585414685,
      "loss": 1.4076,
      "step": 271
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0009739460236951777,
      "loss": 1.3888,
      "step": 272
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0009736983212571646,
      "loss": 1.2439,
      "step": 273
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0009734494786968292,
      "loss": 1.306,
      "step": 274
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0009731994966130988,
      "loss": 1.3795,
      "step": 275
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0009729483756076435,
      "loss": 1.4211,
      "step": 276
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0009726961162848748,
      "loss": 1.3797,
      "step": 277
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.000972442719251944,
      "loss": 1.3158,
      "step": 278
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0009721881851187406,
      "loss": 1.2975,
      "step": 279
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0009719325144978906,
      "loss": 1.3552,
      "step": 280
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0009716757080047557,
      "loss": 1.4167,
      "step": 281
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0009714177662574317,
      "loss": 1.3506,
      "step": 282
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0009711586898767462,
      "loss": 1.261,
      "step": 283
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0009708984794862581,
      "loss": 1.3075,
      "step": 284
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0009706371357122559,
      "loss": 1.4306,
      "step": 285
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0009703746591837552,
      "loss": 1.3797,
      "step": 286
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0009701110505324989,
      "loss": 1.3526,
      "step": 287
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0009698463103929542,
      "loss": 1.3251,
      "step": 288
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0009695804394023118,
      "loss": 1.3114,
      "step": 289
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0009693134382004839,
      "loss": 1.308,
      "step": 290
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0009690453074301035,
      "loss": 1.3964,
      "step": 291
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0009687760477365217,
      "loss": 1.3318,
      "step": 292
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0009685056597678075,
      "loss": 1.3355,
      "step": 293
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0009682341441747445,
      "loss": 1.3281,
      "step": 294
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0009679615016108313,
      "loss": 1.3453,
      "step": 295
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0009676877327322783,
      "loss": 1.3556,
      "step": 296
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0009674128381980072,
      "loss": 1.3232,
      "step": 297
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0009671368186696487,
      "loss": 1.355,
      "step": 298
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0009668596748115412,
      "loss": 1.4265,
      "step": 299
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0009665814072907293,
      "loss": 1.3281,
      "step": 300
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0009663020167769622,
      "loss": 1.3375,
      "step": 301
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0009660215039426918,
      "loss": 1.3823,
      "step": 302
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0009657398694630713,
      "loss": 1.2811,
      "step": 303
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0009654571140159533,
      "loss": 1.333,
      "step": 304
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0009651732382818892,
      "loss": 1.2948,
      "step": 305
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0009648882429441257,
      "loss": 1.3909,
      "step": 306
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0009646021286886051,
      "loss": 1.3878,
      "step": 307
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0009643148962039621,
      "loss": 1.2793,
      "step": 308
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0009640265461815234,
      "loss": 1.3911,
      "step": 309
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0009637370793153051,
      "loss": 1.3576,
      "step": 310
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0009634464963020114,
      "loss": 1.3697,
      "step": 311
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.000963154797841033,
      "loss": 1.34,
      "step": 312
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0009628619846344453,
      "loss": 1.3676,
      "step": 313
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0009625680573870066,
      "loss": 1.3792,
      "step": 314
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0009622730168061567,
      "loss": 1.3444,
      "step": 315
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0009619768636020149,
      "loss": 1.3018,
      "step": 316
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0009616795984873785,
      "loss": 1.3287,
      "step": 317
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0009613812221777212,
      "loss": 1.3368,
      "step": 318
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0009610817353911907,
      "loss": 1.3967,
      "step": 319
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.000960781138848608,
      "loss": 1.349,
      "step": 320
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0009604794332734646,
      "loss": 1.4192,
      "step": 321
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0009601766193919217,
      "loss": 1.4963,
      "step": 322
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0009598726979328079,
      "loss": 1.3626,
      "step": 323
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0009595676696276172,
      "loss": 1.3039,
      "step": 324
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0009592615352105083,
      "loss": 1.3544,
      "step": 325
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0009589542954183017,
      "loss": 1.3263,
      "step": 326
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0009586459509904786,
      "loss": 1.3788,
      "step": 327
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0009583365026691784,
      "loss": 1.3109,
      "step": 328
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0009580259511991982,
      "loss": 1.3415,
      "step": 329
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0009577142973279895,
      "loss": 1.329,
      "step": 330
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0009574015418056577,
      "loss": 1.3532,
      "step": 331
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0009570876853849593,
      "loss": 1.3308,
      "step": 332
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0009567727288213005,
      "loss": 1.2253,
      "step": 333
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0009564566728727357,
      "loss": 1.3772,
      "step": 334
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0009561395182999651,
      "loss": 1.3524,
      "step": 335
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.000955821265866333,
      "loss": 1.3131,
      "step": 336
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0009555019163378267,
      "loss": 1.3555,
      "step": 337
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0009551814704830733,
      "loss": 1.418,
      "step": 338
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0009548599290733392,
      "loss": 1.3122,
      "step": 339
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0009545372928825271,
      "loss": 1.3773,
      "step": 340
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0009542135626871752,
      "loss": 1.3254,
      "step": 341
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0009538887392664543,
      "loss": 1.286,
      "step": 342
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0009535628234021669,
      "loss": 1.3976,
      "step": 343
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0009532358158787446,
      "loss": 1.28,
      "step": 344
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0009529077174832466,
      "loss": 1.3623,
      "step": 345
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0009525785290053572,
      "loss": 1.2996,
      "step": 346
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0009522482512373851,
      "loss": 1.3225,
      "step": 347
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0009519168849742603,
      "loss": 1.3192,
      "step": 348
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0009515844310135328,
      "loss": 1.2743,
      "step": 349
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0009512508901553703,
      "loss": 1.3478,
      "step": 350
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.000950916263202557,
      "loss": 1.3602,
      "step": 351
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0009505805509604906,
      "loss": 1.2831,
      "step": 352
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0009502437542371811,
      "loss": 1.3278,
      "step": 353
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0009499058738432491,
      "loss": 1.3458,
      "step": 354
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0009495669105919228,
      "loss": 1.3307,
      "step": 355
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0009492268652990372,
      "loss": 1.2941,
      "step": 356
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0009488857387830314,
      "loss": 1.3377,
      "step": 357
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0009485435318649467,
      "loss": 1.3217,
      "step": 358
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.000948200245368425,
      "loss": 1.382,
      "step": 359
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0009478558801197064,
      "loss": 1.2613,
      "step": 360
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0009475104369476278,
      "loss": 1.2744,
      "step": 361
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0009471639166836199,
      "loss": 1.3121,
      "step": 362
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0009468163201617061,
      "loss": 1.2647,
      "step": 363
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0009464676482185004,
      "loss": 1.4053,
      "step": 364
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.000946117901693205,
      "loss": 1.3187,
      "step": 365
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0009457670814276083,
      "loss": 1.2807,
      "step": 366
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0009454151882660831,
      "loss": 1.2766,
      "step": 367
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0009450622230555847,
      "loss": 1.3021,
      "step": 368
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0009447081866456489,
      "loss": 1.307,
      "step": 369
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0009443530798883885,
      "loss": 1.2801,
      "step": 370
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0009439969036384943,
      "loss": 1.2729,
      "step": 371
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0009436396587532296,
      "loss": 1.2862,
      "step": 372
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0009432813460924307,
      "loss": 1.3125,
      "step": 373
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0009429219665185034,
      "loss": 1.2449,
      "step": 374
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0009425615208964217,
      "loss": 1.1805,
      "step": 375
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0009422000100937253,
      "loss": 1.2768,
      "step": 376
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0009418374349805175,
      "loss": 1.3051,
      "step": 377
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0009414737964294635,
      "loss": 1.2857,
      "step": 378
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0009411090953157879,
      "loss": 1.321,
      "step": 379
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0009407433325172727,
      "loss": 1.2772,
      "step": 380
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0009403765089142553,
      "loss": 1.3138,
      "step": 381
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0009400086253896263,
      "loss": 1.299,
      "step": 382
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0009396396828288272,
      "loss": 1.2155,
      "step": 383
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0009392696821198487,
      "loss": 1.2683,
      "step": 384
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0009388986241532282,
      "loss": 1.3428,
      "step": 385
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0009385265098220477,
      "loss": 1.2841,
      "step": 386
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0009381533400219318,
      "loss": 1.2653,
      "step": 387
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0009377791156510454,
      "loss": 1.2746,
      "step": 388
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0009374038376100917,
      "loss": 1.2859,
      "step": 389
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0009370275068023098,
      "loss": 1.2945,
      "step": 390
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0009366501241334727,
      "loss": 1.2356,
      "step": 391
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0009362716905118851,
      "loss": 1.4164,
      "step": 392
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0009358922068483812,
      "loss": 1.2241,
      "step": 393
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0009355116740563225,
      "loss": 1.3404,
      "step": 394
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0009351300930515953,
      "loss": 1.2394,
      "step": 395
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0009347474647526095,
      "loss": 1.2378,
      "step": 396
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.000934363790080295,
      "loss": 1.313,
      "step": 397
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0009339790699581003,
      "loss": 1.3382,
      "step": 398
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0009335933053119906,
      "loss": 1.3432,
      "step": 399
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0009332064970704445,
      "loss": 1.26,
      "step": 400
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0009328186461644528,
      "loss": 1.2849,
      "step": 401
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0009324297535275155,
      "loss": 1.3198,
      "step": 402
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0009320398200956402,
      "loss": 1.2924,
      "step": 403
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0009316488468073396,
      "loss": 1.3357,
      "step": 404
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0009312568346036287,
      "loss": 1.2649,
      "step": 405
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0009308637844280237,
      "loss": 1.3008,
      "step": 406
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0009304696972265383,
      "loss": 1.2861,
      "step": 407
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0009300745739476828,
      "loss": 1.323,
      "step": 408
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0009296784155424607,
      "loss": 1.3812,
      "step": 409
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0009292812229643674,
      "loss": 1.3196,
      "step": 410
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0009288829971693868,
      "loss": 1.2383,
      "step": 411
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0009284837391159903,
      "loss": 1.2932,
      "step": 412
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0009280834497651332,
      "loss": 1.3129,
      "step": 413
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0009276821300802534,
      "loss": 1.2712,
      "step": 414
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0009272797810272682,
      "loss": 1.3371,
      "step": 415
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0009268764035745727,
      "loss": 1.2658,
      "step": 416
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0009264719986930376,
      "loss": 1.2297,
      "step": 417
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0009260665673560057,
      "loss": 1.3327,
      "step": 418
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0009256601105392907,
      "loss": 1.3486,
      "step": 419
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0009252526292211749,
      "loss": 1.3231,
      "step": 420
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0009248441243824057,
      "loss": 1.2676,
      "step": 421
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0009244345970061943,
      "loss": 1.3881,
      "step": 422
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0009240240480782129,
      "loss": 1.3203,
      "step": 423
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.000923612478586593,
      "loss": 1.307,
      "step": 424
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0009231998895219212,
      "loss": 1.2146,
      "step": 425
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0009227862818772392,
      "loss": 1.305,
      "step": 426
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0009223716566480399,
      "loss": 1.343,
      "step": 427
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0009219560148322654,
      "loss": 1.3237,
      "step": 428
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0009215393574303044,
      "loss": 1.3399,
      "step": 429
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0009211216854449902,
      "loss": 1.3032,
      "step": 430
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0009207029998815979,
      "loss": 1.2397,
      "step": 431
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0009202833017478421,
      "loss": 1.3672,
      "step": 432
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0009198625920538749,
      "loss": 1.3244,
      "step": 433
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0009194408718122824,
      "loss": 1.408,
      "step": 434
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0009190181420380836,
      "loss": 1.3506,
      "step": 435
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0009185944037487271,
      "loss": 1.3678,
      "step": 436
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0009181696579640882,
      "loss": 1.3023,
      "step": 437
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0009177439057064682,
      "loss": 1.2523,
      "step": 438
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0009173171480005901,
      "loss": 1.2705,
      "step": 439
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0009168893858735971,
      "loss": 1.2803,
      "step": 440
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0009164606203550497,
      "loss": 1.2583,
      "step": 441
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0009160308524769239,
      "loss": 1.2564,
      "step": 442
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0009156000832736073,
      "loss": 1.3196,
      "step": 443
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0009151683137818989,
      "loss": 1.3278,
      "step": 444
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0009147355450410037,
      "loss": 1.3261,
      "step": 445
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0009143017780925331,
      "loss": 1.2981,
      "step": 446
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0009138670139805003,
      "loss": 1.2312,
      "step": 447
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0009134312537513187,
      "loss": 1.3267,
      "step": 448
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0009129944984537993,
      "loss": 1.2548,
      "step": 449
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0009125567491391475,
      "loss": 1.3221,
      "step": 450
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0009121180068609624,
      "loss": 1.3266,
      "step": 451
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0009116782726752316,
      "loss": 1.3046,
      "step": 452
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0009112375476403312,
      "loss": 1.2873,
      "step": 453
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0009107958328170213,
      "loss": 1.1994,
      "step": 454
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0009103531292684449,
      "loss": 1.3173,
      "step": 455
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0009099094380601244,
      "loss": 1.2981,
      "step": 456
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0009094647602599593,
      "loss": 1.2421,
      "step": 457
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0009090190969382239,
      "loss": 1.2778,
      "step": 458
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0009085724491675642,
      "loss": 1.272,
      "step": 459
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0009081248180229962,
      "loss": 1.3485,
      "step": 460
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.000907676204581902,
      "loss": 1.1852,
      "step": 461
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0009072266099240285,
      "loss": 1.333,
      "step": 462
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0009067760351314837,
      "loss": 1.2193,
      "step": 463
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0009063244812887356,
      "loss": 1.338,
      "step": 464
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0009058719494826074,
      "loss": 1.3509,
      "step": 465
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0009054184408022772,
      "loss": 1.2384,
      "step": 466
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0009049639563392736,
      "loss": 1.3114,
      "step": 467
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0009045084971874737,
      "loss": 1.3142,
      "step": 468
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0009040520644431014,
      "loss": 1.2841,
      "step": 469
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0009035946592047228,
      "loss": 1.2831,
      "step": 470
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0009031362825732456,
      "loss": 1.2314,
      "step": 471
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0009026769356519148,
      "loss": 1.3252,
      "step": 472
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.000902216619546311,
      "loss": 1.2777,
      "step": 473
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0009017553353643477,
      "loss": 1.2873,
      "step": 474
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0009012930842162679,
      "loss": 1.2769,
      "step": 475
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0009008298672146425,
      "loss": 1.3632,
      "step": 476
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0009003656854743667,
      "loss": 1.2767,
      "step": 477
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.000899900540112658,
      "loss": 1.3192,
      "step": 478
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0008994344322490527,
      "loss": 1.2775,
      "step": 479
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0008989673630054042,
      "loss": 1.2542,
      "step": 480
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0008984993335058798,
      "loss": 1.2689,
      "step": 481
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0008980303448769574,
      "loss": 1.3299,
      "step": 482
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0008975603982474239,
      "loss": 1.3523,
      "step": 483
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0008970894947483719,
      "loss": 1.3149,
      "step": 484
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0008966176355131972,
      "loss": 1.3776,
      "step": 485
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0008961448216775954,
      "loss": 1.2507,
      "step": 486
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0008956710543795601,
      "loss": 1.2891,
      "step": 487
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0008951963347593796,
      "loss": 1.3054,
      "step": 488
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0008947206639596346,
      "loss": 1.2771,
      "step": 489
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0008942440431251946,
      "loss": 1.1999,
      "step": 490
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0008937664734032158,
      "loss": 1.287,
      "step": 491
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0008932879559431391,
      "loss": 1.3068,
      "step": 492
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.000892808491896685,
      "loss": 1.2622,
      "step": 493
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0008923280824178536,
      "loss": 1.2931,
      "step": 494
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0008918467286629199,
      "loss": 1.2669,
      "step": 495
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0008913644317904317,
      "loss": 1.317,
      "step": 496
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0008908811929612066,
      "loss": 1.3092,
      "step": 497
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0008903970133383297,
      "loss": 1.3269,
      "step": 498
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0008899118940871502,
      "loss": 1.3153,
      "step": 499
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0008894258363752789,
      "loss": 1.282,
      "step": 500
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0008889388413725857,
      "loss": 1.336,
      "step": 501
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0008884509102511955,
      "loss": 1.288,
      "step": 502
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0008879620441854873,
      "loss": 1.2335,
      "step": 503
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0008874722443520898,
      "loss": 1.2456,
      "step": 504
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0008869815119298795,
      "loss": 1.2871,
      "step": 505
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.000886489848099977,
      "loss": 1.3806,
      "step": 506
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0008859972540457451,
      "loss": 1.3176,
      "step": 507
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0008855037309527853,
      "loss": 1.2371,
      "step": 508
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0008850092800089355,
      "loss": 1.2672,
      "step": 509
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0008845139024042663,
      "loss": 1.2842,
      "step": 510
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0008840175993310791,
      "loss": 1.2968,
      "step": 511
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0008835203719839022,
      "loss": 1.2343,
      "step": 512
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.000883022221559489,
      "loss": 1.3354,
      "step": 513
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0008825231492568146,
      "loss": 1.2891,
      "step": 514
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0008820231562770726,
      "loss": 1.3196,
      "step": 515
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0008815222438236725,
      "loss": 1.2956,
      "step": 516
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0008810204131022376,
      "loss": 1.2482,
      "step": 517
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0008805176653206003,
      "loss": 1.2699,
      "step": 518
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0008800140016888008,
      "loss": 1.3441,
      "step": 519
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0008795094234190836,
      "loss": 1.2961,
      "step": 520
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0008790039317258948,
      "loss": 1.2937,
      "step": 521
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0008784975278258782,
      "loss": 1.217,
      "step": 522
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.000877990212937874,
      "loss": 1.2368,
      "step": 523
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0008774819882829143,
      "loss": 1.2271,
      "step": 524
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0008769728550842218,
      "loss": 1.1387,
      "step": 525
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0008764628145672048,
      "loss": 1.2991,
      "step": 526
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0008759518679594563,
      "loss": 1.2625,
      "step": 527
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0008754400164907496,
      "loss": 1.2937,
      "step": 528
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0008749272613930362,
      "loss": 1.2513,
      "step": 529
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0008744136039004422,
      "loss": 1.3616,
      "step": 530
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.000873899045249266,
      "loss": 1.2922,
      "step": 531
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0008733835866779745,
      "loss": 1.1738,
      "step": 532
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0008728672294272009,
      "loss": 1.3082,
      "step": 533
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0008723499747397414,
      "loss": 1.2865,
      "step": 534
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0008718318238605518,
      "loss": 1.215,
      "step": 535
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0008713127780367457,
      "loss": 1.2626,
      "step": 536
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0008707928385175898,
      "loss": 1.2648,
      "step": 537
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0008702720065545023,
      "loss": 1.1852,
      "step": 538
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0008697502834010493,
      "loss": 1.2334,
      "step": 539
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0008692276703129421,
      "loss": 1.237,
      "step": 540
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.000868704168548033,
      "loss": 1.3011,
      "step": 541
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0008681797793663146,
      "loss": 1.3535,
      "step": 542
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0008676545040299144,
      "loss": 1.1985,
      "step": 543
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0008671283438030929,
      "loss": 1.3086,
      "step": 544
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0008666012999522408,
      "loss": 1.3061,
      "step": 545
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0008660733737458752,
      "loss": 1.2511,
      "step": 546
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0008655445664546369,
      "loss": 1.3166,
      "step": 547
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0008650148793512873,
      "loss": 1.2567,
      "step": 548
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0008644843137107057,
      "loss": 1.2109,
      "step": 549
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0008639528708098859,
      "loss": 1.3037,
      "step": 550
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0008634205519279326,
      "loss": 1.2747,
      "step": 551
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0008628873583460592,
      "loss": 1.2499,
      "step": 552
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0008623532913475847,
      "loss": 1.2836,
      "step": 553
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0008618183522179295,
      "loss": 1.2993,
      "step": 554
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.000861282542244614,
      "loss": 1.2431,
      "step": 555
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0008607458627172541,
      "loss": 1.3345,
      "step": 556
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0008602083149275583,
      "loss": 1.2962,
      "step": 557
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0008596699001693256,
      "loss": 1.2109,
      "step": 558
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0008591306197384411,
      "loss": 1.2513,
      "step": 559
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0008585904749328736,
      "loss": 1.2311,
      "step": 560
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0008580494670526724,
      "loss": 1.2921,
      "step": 561
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0008575075973999642,
      "loss": 1.2222,
      "step": 562
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0008569648672789496,
      "loss": 1.255,
      "step": 563
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0008564212779959002,
      "loss": 1.3199,
      "step": 564
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0008558768308591558,
      "loss": 1.2826,
      "step": 565
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0008553315271791205,
      "loss": 1.3044,
      "step": 566
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0008547853682682604,
      "loss": 1.3547,
      "step": 567
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0008542383554411,
      "loss": 1.1997,
      "step": 568
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0008536904900142185,
      "loss": 1.2085,
      "step": 569
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0008531417733062477,
      "loss": 1.2278,
      "step": 570
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0008525922066378682,
      "loss": 1.2821,
      "step": 571
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0008520417913318064,
      "loss": 1.309,
      "step": 572
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.000851490528712831,
      "loss": 1.2609,
      "step": 573
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0008509384201077501,
      "loss": 1.278,
      "step": 574
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0008503854668454083,
      "loss": 1.2606,
      "step": 575
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0008498316702566827,
      "loss": 1.2915,
      "step": 576
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0008492770316744807,
      "loss": 1.2124,
      "step": 577
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0008487215524337357,
      "loss": 1.2577,
      "step": 578
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0008481652338714048,
      "loss": 1.3336,
      "step": 579
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0008476080773264649,
      "loss": 1.2826,
      "step": 580
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0008470500841399103,
      "loss": 1.2565,
      "step": 581
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0008464912556547486,
      "loss": 1.2932,
      "step": 582
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0008459315932159979,
      "loss": 1.2023,
      "step": 583
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0008453710981706837,
      "loss": 1.2711,
      "step": 584
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0008448097718678349,
      "loss": 1.2913,
      "step": 585
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0008442476156584819,
      "loss": 1.2211,
      "step": 586
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0008436846308956518,
      "loss": 1.278,
      "step": 587
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0008431208189343669,
      "loss": 1.2365,
      "step": 588
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0008425561811316391,
      "loss": 1.2593,
      "step": 589
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0008419907188464691,
      "loss": 1.2333,
      "step": 590
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0008414244334398418,
      "loss": 1.2009,
      "step": 591
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0008408573262747225,
      "loss": 1.3409,
      "step": 592
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0008402893987160552,
      "loss": 1.2753,
      "step": 593
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0008397206521307583,
      "loss": 1.2734,
      "step": 594
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0008391510878877211,
      "loss": 1.222,
      "step": 595
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0008385807073578014,
      "loss": 1.3198,
      "step": 596
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0008380095119138209,
      "loss": 1.3423,
      "step": 597
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0008374375029305639,
      "loss": 1.3286,
      "step": 598
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0008368646817847714,
      "loss": 1.2325,
      "step": 599
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0008362910498551402,
      "loss": 1.2494,
      "step": 600
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0008357166085223178,
      "loss": 1.3176,
      "step": 601
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0008351413591689006,
      "loss": 1.254,
      "step": 602
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0008345653031794292,
      "loss": 1.2058,
      "step": 603
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0008339884419403856,
      "loss": 1.2507,
      "step": 604
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0008334107768401902,
      "loss": 1.2402,
      "step": 605
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0008328323092691984,
      "loss": 1.2913,
      "step": 606
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0008322530406196963,
      "loss": 1.3409,
      "step": 607
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0008316729722858987,
      "loss": 1.2669,
      "step": 608
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0008310921056639451,
      "loss": 1.2624,
      "step": 609
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0008305104421518958,
      "loss": 1.2816,
      "step": 610
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0008299279831497298,
      "loss": 1.3373,
      "step": 611
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0008293447300593402,
      "loss": 1.2558,
      "step": 612
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0008287606842845319,
      "loss": 1.3055,
      "step": 613
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.000828175847231017,
      "loss": 1.2143,
      "step": 614
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0008275902203064126,
      "loss": 1.2676,
      "step": 615
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0008270038049202367,
      "loss": 1.2161,
      "step": 616
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0008264166024839051,
      "loss": 1.2389,
      "step": 617
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0008258286144107276,
      "loss": 1.2129,
      "step": 618
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0008252398421159055,
      "loss": 1.2601,
      "step": 619
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0008246502870165273,
      "loss": 1.3371,
      "step": 620
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0008240599505315655,
      "loss": 1.251,
      "step": 621
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0008234688340818733,
      "loss": 1.2202,
      "step": 622
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0008228769390901811,
      "loss": 1.2857,
      "step": 623
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0008222842669810935,
      "loss": 1.2479,
      "step": 624
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0008216908191810852,
      "loss": 1.2333,
      "step": 625
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0008210965971184979,
      "loss": 1.2842,
      "step": 626
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0008205016022235367,
      "loss": 1.2927,
      "step": 627
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0008199058359282674,
      "loss": 1.3098,
      "step": 628
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0008193092996666117,
      "loss": 1.2432,
      "step": 629
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0008187119948743449,
      "loss": 1.3965,
      "step": 630
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0008181139229890919,
      "loss": 1.2213,
      "step": 631
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0008175150854503239,
      "loss": 1.2632,
      "step": 632
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0008169154836993551,
      "loss": 1.2207,
      "step": 633
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.000816315119179339,
      "loss": 1.3216,
      "step": 634
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0008157139933352644,
      "loss": 1.2689,
      "step": 635
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0008151121076139533,
      "loss": 1.3096,
      "step": 636
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0008145094634640562,
      "loss": 1.3367,
      "step": 637
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.0008139060623360494,
      "loss": 1.2335,
      "step": 638
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0008133019056822303,
      "loss": 1.2841,
      "step": 639
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0008126969949567156,
      "loss": 1.2068,
      "step": 640
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0008120913316154365,
      "loss": 1.1934,
      "step": 641
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0008114849171161361,
      "loss": 1.2504,
      "step": 642
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0008108777529183644,
      "loss": 1.2598,
      "step": 643
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0008102698404834772,
      "loss": 1.216,
      "step": 644
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0008096611812746301,
      "loss": 1.1882,
      "step": 645
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0008090517767567764,
      "loss": 1.3034,
      "step": 646
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0008084416283966636,
      "loss": 1.2061,
      "step": 647
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0008078307376628291,
      "loss": 1.2224,
      "step": 648
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0008072191060255973,
      "loss": 1.2573,
      "step": 649
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0008066067349570756,
      "loss": 1.2226,
      "step": 650
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0008059936259311513,
      "loss": 1.278,
      "step": 651
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0008053797804234882,
      "loss": 1.2811,
      "step": 652
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0008047651999115216,
      "loss": 1.2231,
      "step": 653
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0008041498858744571,
      "loss": 1.286,
      "step": 654
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0008035338397932652,
      "loss": 1.2627,
      "step": 655
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0008029170631506785,
      "loss": 1.2633,
      "step": 656
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0008022995574311875,
      "loss": 1.2814,
      "step": 657
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0008016813241210379,
      "loss": 1.2839,
      "step": 658
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0008010623647082268,
      "loss": 1.2662,
      "step": 659
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0008004426806824984,
      "loss": 1.2516,
      "step": 660
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0007998222735353411,
      "loss": 1.23,
      "step": 661
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0007992011447599841,
      "loss": 1.1748,
      "step": 662
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0007985792958513931,
      "loss": 1.279,
      "step": 663
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.000797956728306267,
      "loss": 1.2593,
      "step": 664
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0007973334436230346,
      "loss": 1.2644,
      "step": 665
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0007967094433018508,
      "loss": 1.214,
      "step": 666
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0007960847288445924,
      "loss": 1.2344,
      "step": 667
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0007954593017548556,
      "loss": 1.229,
      "step": 668
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0007948331635379516,
      "loss": 1.2538,
      "step": 669
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0007942063157009031,
      "loss": 1.3289,
      "step": 670
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.000793578759752441,
      "loss": 1.2659,
      "step": 671
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0007929504972030003,
      "loss": 1.2832,
      "step": 672
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0007923215295647166,
      "loss": 1.3282,
      "step": 673
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0007916918583514227,
      "loss": 1.199,
      "step": 674
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0007910614850786447,
      "loss": 1.2298,
      "step": 675
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0007904304112635988,
      "loss": 1.1962,
      "step": 676
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0007897986384251868,
      "loss": 1.2191,
      "step": 677
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0007891661680839932,
      "loss": 1.1527,
      "step": 678
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0007885330017622815,
      "loss": 1.2055,
      "step": 679
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0007878991409839898,
      "loss": 1.3235,
      "step": 680
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0007872645872747282,
      "loss": 1.1858,
      "step": 681
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.000786629342161774,
      "loss": 1.2867,
      "step": 682
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0007859934071740692,
      "loss": 1.3074,
      "step": 683
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.000785356783842216,
      "loss": 1.3113,
      "step": 684
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0007847194736984729,
      "loss": 1.3093,
      "step": 685
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0007840814782767525,
      "loss": 1.1601,
      "step": 686
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0007834427991126155,
      "loss": 1.2467,
      "step": 687
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0007828034377432694,
      "loss": 1.2161,
      "step": 688
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0007821633957075627,
      "loss": 1.2403,
      "step": 689
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0007815226745459831,
      "loss": 1.3021,
      "step": 690
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.000780881275800652,
      "loss": 1.2711,
      "step": 691
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0007802392010153223,
      "loss": 1.2456,
      "step": 692
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0007795964517353734,
      "loss": 1.291,
      "step": 693
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0007789530295078089,
      "loss": 1.2443,
      "step": 694
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0007783089358812513,
      "loss": 1.1918,
      "step": 695
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0007776641724059396,
      "loss": 1.1849,
      "step": 696
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0007770187406337248,
      "loss": 1.1555,
      "step": 697
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0007763726421180664,
      "loss": 1.1984,
      "step": 698
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0007757258784140287,
      "loss": 1.2279,
      "step": 699
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.000775078451078277,
      "loss": 1.194,
      "step": 700
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0007744303616690738,
      "loss": 1.1813,
      "step": 701
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0007737816117462751,
      "loss": 1.1434,
      "step": 702
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.000773132202871327,
      "loss": 1.2536,
      "step": 703
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0007724821366072612,
      "loss": 1.2158,
      "step": 704
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0007718314145186917,
      "loss": 1.1762,
      "step": 705
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0007711800381718111,
      "loss": 1.2076,
      "step": 706
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0007705280091343867,
      "loss": 1.1205,
      "step": 707
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0007698753289757565,
      "loss": 1.2154,
      "step": 708
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0007692219992668261,
      "loss": 1.2317,
      "step": 709
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0007685680215800639,
      "loss": 1.1582,
      "step": 710
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0007679133974894983,
      "loss": 1.2153,
      "step": 711
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0007672581285707136,
      "loss": 1.1358,
      "step": 712
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0007666022164008457,
      "loss": 1.1708,
      "step": 713
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0007659456625585789,
      "loss": 1.2703,
      "step": 714
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0007652884686241419,
      "loss": 1.2445,
      "step": 715
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0007646306361793042,
      "loss": 1.1885,
      "step": 716
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0007639721668073718,
      "loss": 1.314,
      "step": 717
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0007633130620931837,
      "loss": 1.1903,
      "step": 718
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0007626533236231084,
      "loss": 1.2797,
      "step": 719
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0007619929529850397,
      "loss": 1.2206,
      "step": 720
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0007613319517683924,
      "loss": 1.1575,
      "step": 721
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.0007606703215640994,
      "loss": 1.1287,
      "step": 722
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0007600080639646077,
      "loss": 1.2098,
      "step": 723
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0007593451805638742,
      "loss": 1.1805,
      "step": 724
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0007586816729573617,
      "loss": 1.2756,
      "step": 725
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0007580175427420357,
      "loss": 1.2654,
      "step": 726
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0007573527915163602,
      "loss": 1.168,
      "step": 727
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0007566874208802939,
      "loss": 1.1808,
      "step": 728
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0007560214324352858,
      "loss": 1.2017,
      "step": 729
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.0007553548277842727,
      "loss": 1.1901,
      "step": 730
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.0007546876085316742,
      "loss": 1.1695,
      "step": 731
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.0007540197762833889,
      "loss": 1.2689,
      "step": 732
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0007533513326467912,
      "loss": 1.1781,
      "step": 733
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0007526822792307266,
      "loss": 1.2157,
      "step": 734
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0007520126176455083,
      "loss": 1.2479,
      "step": 735
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0007513423495029139,
      "loss": 1.2777,
      "step": 736
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00075067147641618,
      "loss": 1.1869,
      "step": 737
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00075,
      "loss": 1.2998,
      "step": 738
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.0007493279218705189,
      "loss": 1.1676,
      "step": 739
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00074865524364533,
      "loss": 1.1974,
      "step": 740
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.0007479819669434711,
      "loss": 1.1823,
      "step": 741
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.0007473080933854204,
      "loss": 1.1944,
      "step": 742
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.0007466336245930927,
      "loss": 1.2239,
      "step": 743
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.0007459585621898353,
      "loss": 1.1953,
      "step": 744
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.0007452829078004245,
      "loss": 1.1727,
      "step": 745
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.0007446066630510609,
      "loss": 1.1727,
      "step": 746
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0007439298295693664,
      "loss": 1.1814,
      "step": 747
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0007432524089843799,
      "loss": 1.2628,
      "step": 748
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0007425744029265535,
      "loss": 1.1427,
      "step": 749
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0007418958130277482,
      "loss": 1.2016,
      "step": 750
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00074121664092123,
      "loss": 1.1896,
      "step": 751
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0007405368882416667,
      "loss": 1.2848,
      "step": 752
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0007398565566251232,
      "loss": 1.2178,
      "step": 753
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.0007391756477090576,
      "loss": 1.1474,
      "step": 754
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.000738494163132318,
      "loss": 1.2614,
      "step": 755
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.0007378121045351377,
      "loss": 1.2518,
      "step": 756
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.0007371294735591314,
      "loss": 1.2274,
      "step": 757
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.0007364462718472918,
      "loss": 1.1583,
      "step": 758
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.0007357625010439852,
      "loss": 1.1966,
      "step": 759
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.0007350781627949475,
      "loss": 1.2213,
      "step": 760
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0007343932587472803,
      "loss": 1.157,
      "step": 761
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0007337077905494472,
      "loss": 1.2594,
      "step": 762
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0007330217598512695,
      "loss": 1.15,
      "step": 763
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0007323351683039223,
      "loss": 1.124,
      "step": 764
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0007316480175599309,
      "loss": 1.2586,
      "step": 765
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.000730960309273166,
      "loss": 1.1638,
      "step": 766
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0007302720450988407,
      "loss": 1.2584,
      "step": 767
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0007295832266935059,
      "loss": 1.2063,
      "step": 768
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.000728893855715046,
      "loss": 1.1813,
      "step": 769
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0007282039338226763,
      "loss": 1.2691,
      "step": 770
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0007275134626769369,
      "loss": 1.2323,
      "step": 771
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0007268224439396909,
      "loss": 1.2579,
      "step": 772
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0007261308792741188,
      "loss": 1.1843,
      "step": 773
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0007254387703447154,
      "loss": 1.1835,
      "step": 774
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0007247461188172852,
      "loss": 1.2046,
      "step": 775
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0007240529263589387,
      "loss": 1.2086,
      "step": 776
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0007233591946380884,
      "loss": 1.1944,
      "step": 777
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0007226649253244447,
      "loss": 1.2069,
      "step": 778
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.000721970120089012,
      "loss": 1.1781,
      "step": 779
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0007212747806040845,
      "loss": 1.166,
      "step": 780
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0007205789085432422,
      "loss": 1.21,
      "step": 781
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.000719882505581347,
      "loss": 1.2072,
      "step": 782
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.0007191855733945387,
      "loss": 1.1883,
      "step": 783
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.0007184881136602308,
      "loss": 1.1572,
      "step": 784
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0007177901280571063,
      "loss": 1.2424,
      "step": 785
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.000717091618265114,
      "loss": 1.1676,
      "step": 786
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0007163925859654649,
      "loss": 1.1507,
      "step": 787
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0007156930328406268,
      "loss": 1.2171,
      "step": 788
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0007149929605743214,
      "loss": 1.2667,
      "step": 789
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0007142923708515199,
      "loss": 1.2108,
      "step": 790
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0007135912653584389,
      "loss": 1.2379,
      "step": 791
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0007128896457825364,
      "loss": 1.2768,
      "step": 792
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0007121875138125077,
      "loss": 1.2527,
      "step": 793
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0007114848711382815,
      "loss": 1.1553,
      "step": 794
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0007107817194510156,
      "loss": 1.134,
      "step": 795
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0007100780604430928,
      "loss": 1.1948,
      "step": 796
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.0007093738958081168,
      "loss": 1.1314,
      "step": 797
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.000708669227240909,
      "loss": 1.2322,
      "step": 798
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0007079640564375028,
      "loss": 1.1523,
      "step": 799
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0007072583850951411,
      "loss": 1.2105,
      "step": 800
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.000706552214912271,
      "loss": 1.1566,
      "step": 801
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0007058455475885405,
      "loss": 1.2123,
      "step": 802
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0007051383848247942,
      "loss": 1.2024,
      "step": 803
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.000704430728323069,
      "loss": 1.1633,
      "step": 804
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0007037225797865902,
      "loss": 1.1965,
      "step": 805
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0007030139409197676,
      "loss": 1.1422,
      "step": 806
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0007023048134281906,
      "loss": 1.1593,
      "step": 807
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0007015951990186251,
      "loss": 1.2575,
      "step": 808
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.000700885099399009,
      "loss": 1.2021,
      "step": 809
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0007001745162784476,
      "loss": 1.1476,
      "step": 810
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0006994634513672101,
      "loss": 1.0767,
      "step": 811
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0006987519063767256,
      "loss": 1.2395,
      "step": 812
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.0006980398830195785,
      "loss": 1.1671,
      "step": 813
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.0006973273830095042,
      "loss": 1.1685,
      "step": 814
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.0006966144080613858,
      "loss": 1.2131,
      "step": 815
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.0006959009598912492,
      "loss": 1.1373,
      "step": 816
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.0006951870402162597,
      "loss": 1.1954,
      "step": 817
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.0006944726507547168,
      "loss": 1.1569,
      "step": 818
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.0006937577932260515,
      "loss": 1.2331,
      "step": 819
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0006930424693508206,
      "loss": 1.1988,
      "step": 820
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0006923266808507037,
      "loss": 1.2639,
      "step": 821
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0006916104294484987,
      "loss": 1.1916,
      "step": 822
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0006908937168681175,
      "loss": 1.1541,
      "step": 823
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0006901765448345822,
      "loss": 1.1777,
      "step": 824
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0006894589150740207,
      "loss": 1.2299,
      "step": 825
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0006887408293136621,
      "loss": 1.1368,
      "step": 826
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.000688022289281834,
      "loss": 1.264,
      "step": 827
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.0006873032967079561,
      "loss": 1.1442,
      "step": 828
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.0006865838533225383,
      "loss": 1.2316,
      "step": 829
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0006858639608571753,
      "loss": 1.2249,
      "step": 830
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0006851436210445427,
      "loss": 1.1791,
      "step": 831
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0006844228356183924,
      "loss": 1.2359,
      "step": 832
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0006837016063135491,
      "loss": 1.2519,
      "step": 833
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.000682979934865906,
      "loss": 1.2299,
      "step": 834
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.0006822578230124206,
      "loss": 1.1473,
      "step": 835
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.0006815352724911095,
      "loss": 1.2009,
      "step": 836
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.000680812285041046,
      "loss": 1.2532,
      "step": 837
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.0006800888624023553,
      "loss": 1.2071,
      "step": 838
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.0006793650063162086,
      "loss": 1.1213,
      "step": 839
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.000678640718524822,
      "loss": 1.171,
      "step": 840
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.0006779160007714494,
      "loss": 1.2456,
      "step": 841
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.0006771908548003803,
      "loss": 1.1979,
      "step": 842
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.0006764652823569344,
      "loss": 1.2318,
      "step": 843
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0006757392851874584,
      "loss": 1.2371,
      "step": 844
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0006750128650393206,
      "loss": 1.2724,
      "step": 845
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0006742860236609076,
      "loss": 1.3096,
      "step": 846
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0006735587628016203,
      "loss": 1.2087,
      "step": 847
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0006728310842118683,
      "loss": 1.188,
      "step": 848
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0006721029896430677,
      "loss": 1.1758,
      "step": 849
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0006713744808476349,
      "loss": 1.1643,
      "step": 850
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.0006706455595789836,
      "loss": 1.1233,
      "step": 851
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.0006699162275915207,
      "loss": 1.2158,
      "step": 852
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.0006691864866406407,
      "loss": 1.147,
      "step": 853
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.0006684563384827232,
      "loss": 1.1856,
      "step": 854
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0006677257848751275,
      "loss": 1.2566,
      "step": 855
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0006669948275761892,
      "loss": 1.1634,
      "step": 856
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0006662634683452149,
      "loss": 1.1817,
      "step": 857
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0006655317089424791,
      "loss": 1.1942,
      "step": 858
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0006647995511292191,
      "loss": 1.2102,
      "step": 859
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0006640669966676316,
      "loss": 1.2346,
      "step": 860
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0006633340473208672,
      "loss": 1.2372,
      "step": 861
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0006626007048530276,
      "loss": 1.2162,
      "step": 862
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0006618669710291606,
      "loss": 1.215,
      "step": 863
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0006611328476152556,
      "loss": 1.1908,
      "step": 864
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00066039833637824,
      "loss": 1.2289,
      "step": 865
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.0006596634390859745,
      "loss": 1.1446,
      "step": 866
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.000658928157507249,
      "loss": 1.1242,
      "step": 867
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.0006581924934117783,
      "loss": 1.1447,
      "step": 868
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.000657456448570198,
      "loss": 1.2178,
      "step": 869
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.0006567200247540598,
      "loss": 1.29,
      "step": 870
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.0006559832237358279,
      "loss": 1.2205,
      "step": 871
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0006552460472888739,
      "loss": 1.1383,
      "step": 872
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0006545084971874737,
      "loss": 1.1509,
      "step": 873
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0006537705752068019,
      "loss": 1.1813,
      "step": 874
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0006530322831229284,
      "loss": 1.2158,
      "step": 875
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0006522936227128139,
      "loss": 1.2429,
      "step": 876
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0006515545957543056,
      "loss": 1.2227,
      "step": 877
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0006508152040261328,
      "loss": 1.1857,
      "step": 878
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0006500754493079029,
      "loss": 1.1798,
      "step": 879
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0006493353333800969,
      "loss": 1.2326,
      "step": 880
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0006485948580240653,
      "loss": 1.2326,
      "step": 881
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0006478540250220234,
      "loss": 1.1962,
      "step": 882
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.0006471128361570475,
      "loss": 1.2355,
      "step": 883
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.0006463712932130708,
      "loss": 1.1608,
      "step": 884
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.0006456293979748778,
      "loss": 1.1631,
      "step": 885
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0006448871522281016,
      "loss": 1.2149,
      "step": 886
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0006441445577592186,
      "loss": 1.1043,
      "step": 887
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0006434016163555452,
      "loss": 1.2877,
      "step": 888
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0006426583298052318,
      "loss": 1.212,
      "step": 889
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.0006419146998972601,
      "loss": 1.087,
      "step": 890
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.0006411707284214383,
      "loss": 1.1614,
      "step": 891
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.0006404264171683964,
      "loss": 1.2826,
      "step": 892
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.0006396817679295822,
      "loss": 1.1767,
      "step": 893
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.0006389367824972574,
      "loss": 1.1043,
      "step": 894
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.0006381914626644924,
      "loss": 1.1907,
      "step": 895
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.0006374458102251628,
      "loss": 1.2641,
      "step": 896
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.0006366998269739441,
      "loss": 1.214,
      "step": 897
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.0006359535147063091,
      "loss": 1.1894,
      "step": 898
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.0006352068752185214,
      "loss": 1.1588,
      "step": 899
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0006344599103076329,
      "loss": 1.1395,
      "step": 900
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0006337126217714784,
      "loss": 1.2779,
      "step": 901
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0006329650114086717,
      "loss": 1.1011,
      "step": 902
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0006322170810186012,
      "loss": 1.3069,
      "step": 903
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.0006314688324014255,
      "loss": 1.2329,
      "step": 904
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.0006307202673580694,
      "loss": 1.1944,
      "step": 905
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.0006299713876902188,
      "loss": 1.2774,
      "step": 906
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.0006292221952003172,
      "loss": 1.1748,
      "step": 907
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.000628472691691561,
      "loss": 1.1275,
      "step": 908
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.0006277228789678953,
      "loss": 1.235,
      "step": 909
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.000626972758834009,
      "loss": 1.2509,
      "step": 910
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.0006262223330953311,
      "loss": 1.1254,
      "step": 911
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.0006254716035580263,
      "loss": 1.1664,
      "step": 912
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.0006247205720289907,
      "loss": 1.1702,
      "step": 913
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0006239692403158465,
      "loss": 1.1921,
      "step": 914
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0006232176102269389,
      "loss": 1.1434,
      "step": 915
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0006224656835713313,
      "loss": 1.1756,
      "step": 916
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0006217134621588009,
      "loss": 1.1517,
      "step": 917
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0006209609477998338,
      "loss": 1.1592,
      "step": 918
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0006202081423056221,
      "loss": 1.2127,
      "step": 919
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.000619455047488058,
      "loss": 1.2089,
      "step": 920
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.0006187016651597299,
      "loss": 1.1625,
      "step": 921
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.0006179479971339186,
      "loss": 1.2292,
      "step": 922
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.0006171940452245924,
      "loss": 1.2075,
      "step": 923
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0006164398112464029,
      "loss": 1.1358,
      "step": 924
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0006156852970146807,
      "loss": 1.1885,
      "step": 925
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0006149305043454306,
      "loss": 1.1671,
      "step": 926
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0006141754350553279,
      "loss": 1.2438,
      "step": 927
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0006134200909617135,
      "loss": 1.192,
      "step": 928
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0006126644738825896,
      "loss": 1.2067,
      "step": 929
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0006119085856366158,
      "loss": 1.145,
      "step": 930
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0006111524280431039,
      "loss": 1.1635,
      "step": 931
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0006103960029220144,
      "loss": 1.2305,
      "step": 932
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0006096393120939516,
      "loss": 1.1915,
      "step": 933
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0006088823573801591,
      "loss": 1.1781,
      "step": 934
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0006081251406025161,
      "loss": 1.1884,
      "step": 935
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0006073676635835317,
      "loss": 1.1184,
      "step": 936
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0006066099281463422,
      "loss": 1.2154,
      "step": 937
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.0006058519361147054,
      "loss": 1.2657,
      "step": 938
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.000605093689312997,
      "loss": 1.1737,
      "step": 939
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.0006043351895662058,
      "loss": 1.1415,
      "step": 940
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.0006035764386999291,
      "loss": 1.2142,
      "step": 941
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0006028174385403693,
      "loss": 1.1979,
      "step": 942
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0006020581909143279,
      "loss": 1.2643,
      "step": 943
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0006012986976492025,
      "loss": 1.1504,
      "step": 944
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.0006005389605729824,
      "loss": 1.211,
      "step": 945
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.0005997789815142427,
      "loss": 1.2437,
      "step": 946
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.0005990187623021415,
      "loss": 1.2712,
      "step": 947
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.0005982583047664151,
      "loss": 1.1865,
      "step": 948
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.000597497610737373,
      "loss": 1.2424,
      "step": 949
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.0005967366820458939,
      "loss": 1.1715,
      "step": 950
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.0005959755205234218,
      "loss": 1.2058,
      "step": 951
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.0005952141280019605,
      "loss": 1.1844,
      "step": 952
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.0005944525063140702,
      "loss": 1.2242,
      "step": 953
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.0005936906572928624,
      "loss": 1.1708,
      "step": 954
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.0005929285827719958,
      "loss": 1.215,
      "step": 955
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.0005921662845856718,
      "loss": 1.1317,
      "step": 956
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.0005914037645686307,
      "loss": 1.1797,
      "step": 957
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.0005906410245561459,
      "loss": 1.2035,
      "step": 958
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.0005898780663840205,
      "loss": 1.1764,
      "step": 959
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.0005891148918885833,
      "loss": 1.1791,
      "step": 960
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.0005883515029066829,
      "loss": 1.1703,
      "step": 961
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.0005875879012756844,
      "loss": 1.1648,
      "step": 962
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.0005868240888334653,
      "loss": 1.097,
      "step": 963
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.0005860600674184095,
      "loss": 1.1576,
      "step": 964
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.0005852958388694048,
      "loss": 1.1577,
      "step": 965
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.000584531405025837,
      "loss": 1.1263,
      "step": 966
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.0005837667677275863,
      "loss": 1.1775,
      "step": 967
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.0005830019288150222,
      "loss": 1.2317,
      "step": 968
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.0005822368901289993,
      "loss": 1.1992,
      "step": 969
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.0005814716535108544,
      "loss": 1.1569,
      "step": 970
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.000580706220802399,
      "loss": 1.1517,
      "step": 971
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.0005799405938459175,
      "loss": 1.1357,
      "step": 972
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0005791747744841615,
      "loss": 1.1624,
      "step": 973
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0005784087645603459,
      "loss": 1.1524,
      "step": 974
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0005776425659181438,
      "loss": 1.1156,
      "step": 975
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0005768761804016832,
      "loss": 1.1819,
      "step": 976
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.0005761096098555416,
      "loss": 1.1656,
      "step": 977
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.0005753428561247416,
      "loss": 1.2386,
      "step": 978
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.0005745759210547469,
      "loss": 1.0956,
      "step": 979
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.0005738088064914576,
      "loss": 1.2076,
      "step": 980
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.0005730415142812059,
      "loss": 1.1079,
      "step": 981
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.0005722740462707514,
      "loss": 1.1055,
      "step": 982
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.0005715064043072771,
      "loss": 1.1439,
      "step": 983
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0005707385902383845,
      "loss": 1.0889,
      "step": 984
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0005699706059120893,
      "loss": 1.2103,
      "step": 985
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0005692024531768169,
      "loss": 1.1752,
      "step": 986
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.0005684341338813985,
      "loss": 1.1571,
      "step": 987
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.0005676656498750656,
      "loss": 1.2765,
      "step": 988
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.0005668970030074465,
      "loss": 1.1133,
      "step": 989
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.0005661281951285613,
      "loss": 1.2127,
      "step": 990
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.0005653592280888177,
      "loss": 1.15,
      "step": 991
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.0005645901037390067,
      "loss": 1.1815,
      "step": 992
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.0005638208239302974,
      "loss": 1.2119,
      "step": 993
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0005630513905142333,
      "loss": 1.2573,
      "step": 994
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0005622818053427282,
      "loss": 1.1782,
      "step": 995
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0005615120702680604,
      "loss": 1.0716,
      "step": 996
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.000560742187142869,
      "loss": 1.2212,
      "step": 997
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0005599721578201499,
      "loss": 1.2399,
      "step": 998
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0005592019841532506,
      "loss": 1.1585,
      "step": 999
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0005584316679958658,
      "loss": 1.1637,
      "step": 1000
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.0005576612112020338,
      "loss": 1.1817,
      "step": 1001
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.0005568906156261309,
      "loss": 1.1983,
      "step": 1002
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.0005561198831228676,
      "loss": 1.1674,
      "step": 1003
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.0005553490155472834,
      "loss": 1.1178,
      "step": 1004
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.000554578014754744,
      "loss": 1.151,
      "step": 1005
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.0005538068826009349,
      "loss": 1.173,
      "step": 1006
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.0005530356209418578,
      "loss": 1.1019,
      "step": 1007
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0005522642316338268,
      "loss": 1.1251,
      "step": 1008
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0005514927165334622,
      "loss": 1.2816,
      "step": 1009
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.000550721077497688,
      "loss": 1.2117,
      "step": 1010
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.0005499493163837257,
      "loss": 1.1782,
      "step": 1011
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.0005491774350490912,
      "loss": 1.2102,
      "step": 1012
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.0005484054353515896,
      "loss": 1.1782,
      "step": 1013
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.0005476333191493108,
      "loss": 1.1476,
      "step": 1014
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.0005468610883006252,
      "loss": 1.1584,
      "step": 1015
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.000546088744664179,
      "loss": 1.1905,
      "step": 1016
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.0005453162900988901,
      "loss": 1.2215,
      "step": 1017
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0005445437264639432,
      "loss": 1.1278,
      "step": 1018
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0005437710556187854,
      "loss": 1.1409,
      "step": 1019
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0005429982794231221,
      "loss": 1.1426,
      "step": 1020
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0005422253997369123,
      "loss": 1.218,
      "step": 1021
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0005414524184203638,
      "loss": 1.1697,
      "step": 1022
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0005406793373339292,
      "loss": 1.1493,
      "step": 1023
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0005399061583383013,
      "loss": 1.1395,
      "step": 1024
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.000539132883294408,
      "loss": 1.2382,
      "step": 1025
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.0005383595140634093,
      "loss": 1.1458,
      "step": 1026
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.0005375860525066912,
      "loss": 1.1779,
      "step": 1027
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.0005368125004858624,
      "loss": 1.1446,
      "step": 1028
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0005360388598627487,
      "loss": 1.2227,
      "step": 1029
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0005352651324993897,
      "loss": 1.1725,
      "step": 1030
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0005344913202580335,
      "loss": 1.1976,
      "step": 1031
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0005337174250011327,
      "loss": 1.0558,
      "step": 1032
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0005329434485913392,
      "loss": 1.1104,
      "step": 1033
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0005321693928915011,
      "loss": 1.1424,
      "step": 1034
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0005313952597646568,
      "loss": 1.2022,
      "step": 1035
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0005306210510740306,
      "loss": 1.2174,
      "step": 1036
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0005298467686830297,
      "loss": 1.2138,
      "step": 1037
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0005290724144552379,
      "loss": 1.1793,
      "step": 1038
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.0005282979902544123,
      "loss": 1.1718,
      "step": 1039
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.0005275234979444781,
      "loss": 1.1515,
      "step": 1040
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.0005267489393895247,
      "loss": 1.1323,
      "step": 1041
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.0005259743164538008,
      "loss": 1.1724,
      "step": 1042
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.00052519963100171,
      "loss": 1.1684,
      "step": 1043
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0005244248848978067,
      "loss": 1.1773,
      "step": 1044
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0005236500800067908,
      "loss": 1.0887,
      "step": 1045
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.0005228752181935042,
      "loss": 1.0698,
      "step": 1046
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.0005221003013229253,
      "loss": 1.0683,
      "step": 1047
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.0005213253312601653,
      "loss": 1.1105,
      "step": 1048
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.0005205503098704635,
      "loss": 1.1323,
      "step": 1049
    },
    {
      "epoch": 3.02,
      "learning_rate": 0.0005197752390191827,
      "loss": 1.1165,
      "step": 1050
    },
    {
      "epoch": 3.02,
      "learning_rate": 0.0005190001205718041,
      "loss": 1.0818,
      "step": 1051
    },
    {
      "epoch": 3.02,
      "learning_rate": 0.0005182249563939246,
      "loss": 1.1274,
      "step": 1052
    },
    {
      "epoch": 3.03,
      "learning_rate": 0.0005174497483512506,
      "loss": 1.1528,
      "step": 1053
    },
    {
      "epoch": 3.03,
      "learning_rate": 0.0005166744983095936,
      "loss": 1.1359,
      "step": 1054
    },
    {
      "epoch": 3.03,
      "learning_rate": 0.000515899208134867,
      "loss": 1.1533,
      "step": 1055
    },
    {
      "epoch": 3.03,
      "learning_rate": 0.0005151238796930804,
      "loss": 1.1139,
      "step": 1056
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.0005143485148503354,
      "loss": 1.076,
      "step": 1057
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.0005135731154728214,
      "loss": 1.0574,
      "step": 1058
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.0005127976834268111,
      "loss": 1.0484,
      "step": 1059
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.0005120222205786556,
      "loss": 1.0885,
      "step": 1060
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.0005112467287947797,
      "loss": 1.1351,
      "step": 1061
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.0005104712099416785,
      "loss": 1.0984,
      "step": 1062
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.0005096956658859122,
      "loss": 1.0416,
      "step": 1063
    },
    {
      "epoch": 3.06,
      "learning_rate": 0.0005089200984941014,
      "loss": 1.0229,
      "step": 1064
    },
    {
      "epoch": 3.06,
      "learning_rate": 0.0005081445096329229,
      "loss": 1.1459,
      "step": 1065
    },
    {
      "epoch": 3.06,
      "learning_rate": 0.0005073689011691054,
      "loss": 1.0758,
      "step": 1066
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0005065932749694246,
      "loss": 1.138,
      "step": 1067
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0005058176329006986,
      "loss": 1.1594,
      "step": 1068
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0005050419768297843,
      "loss": 1.1017,
      "step": 1069
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0005042663086235718,
      "loss": 1.1152,
      "step": 1070
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.0005034906301489807,
      "loss": 1.0816,
      "step": 1071
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.0005027149432729554,
      "loss": 1.1526,
      "step": 1072
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.0005019392498624602,
      "loss": 1.1559,
      "step": 1073
    },
    {
      "epoch": 3.09,
      "learning_rate": 0.0005011635517844752,
      "loss": 1.1036,
      "step": 1074
    },
    {
      "epoch": 3.09,
      "learning_rate": 0.000500387850905992,
      "loss": 1.0871,
      "step": 1075
    },
    {
      "epoch": 3.09,
      "learning_rate": 0.0004996121490940083,
      "loss": 1.0594,
      "step": 1076
    },
    {
      "epoch": 3.09,
      "learning_rate": 0.0004988364482155249,
      "loss": 1.1318,
      "step": 1077
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.0004980607501375399,
      "loss": 1.1014,
      "step": 1078
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.0004972850567270446,
      "loss": 1.1629,
      "step": 1079
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.0004965093698510193,
      "loss": 1.0418,
      "step": 1080
    },
    {
      "epoch": 3.11,
      "learning_rate": 0.0004957336913764283,
      "loss": 1.0961,
      "step": 1081
    },
    {
      "epoch": 3.11,
      "learning_rate": 0.0004949580231702158,
      "loss": 1.148,
      "step": 1082
    },
    {
      "epoch": 3.11,
      "learning_rate": 0.0004941823670993016,
      "loss": 1.0372,
      "step": 1083
    },
    {
      "epoch": 3.11,
      "learning_rate": 0.0004934067250305757,
      "loss": 1.1152,
      "step": 1084
    },
    {
      "epoch": 3.12,
      "learning_rate": 0.0004926310988308946,
      "loss": 1.0894,
      "step": 1085
    },
    {
      "epoch": 3.12,
      "learning_rate": 0.0004918554903670772,
      "loss": 1.1416,
      "step": 1086
    },
    {
      "epoch": 3.12,
      "learning_rate": 0.0004910799015058986,
      "loss": 1.1093,
      "step": 1087
    },
    {
      "epoch": 3.13,
      "learning_rate": 0.0004903043341140879,
      "loss": 1.0887,
      "step": 1088
    },
    {
      "epoch": 3.13,
      "learning_rate": 0.0004895287900583215,
      "loss": 1.1441,
      "step": 1089
    },
    {
      "epoch": 3.13,
      "learning_rate": 0.0004887532712052206,
      "loss": 1.1349,
      "step": 1090
    },
    {
      "epoch": 3.14,
      "learning_rate": 0.0004879777794213447,
      "loss": 1.1123,
      "step": 1091
    },
    {
      "epoch": 3.14,
      "learning_rate": 0.000487202316573189,
      "loss": 1.1615,
      "step": 1092
    },
    {
      "epoch": 3.14,
      "learning_rate": 0.0004864268845271786,
      "loss": 1.0598,
      "step": 1093
    },
    {
      "epoch": 3.14,
      "learning_rate": 0.00048565148514966467,
      "loss": 1.1224,
      "step": 1094
    },
    {
      "epoch": 3.15,
      "learning_rate": 0.00048487612030691967,
      "loss": 1.0923,
      "step": 1095
    },
    {
      "epoch": 3.15,
      "learning_rate": 0.00048410079186513293,
      "loss": 1.1343,
      "step": 1096
    },
    {
      "epoch": 3.15,
      "learning_rate": 0.00048332550169040656,
      "loss": 1.0975,
      "step": 1097
    },
    {
      "epoch": 3.16,
      "learning_rate": 0.0004825502516487497,
      "loss": 1.121,
      "step": 1098
    },
    {
      "epoch": 3.16,
      "learning_rate": 0.00048177504360607544,
      "loss": 1.0991,
      "step": 1099
    },
    {
      "epoch": 3.16,
      "learning_rate": 0.000480999879428196,
      "loss": 1.1709,
      "step": 1100
    },
    {
      "epoch": 3.16,
      "learning_rate": 0.0004802247609808175,
      "loss": 1.0845,
      "step": 1101
    },
    {
      "epoch": 3.17,
      "learning_rate": 0.0004794496901295365,
      "loss": 1.0718,
      "step": 1102
    },
    {
      "epoch": 3.17,
      "learning_rate": 0.0004786746687398346,
      "loss": 1.0746,
      "step": 1103
    },
    {
      "epoch": 3.17,
      "learning_rate": 0.0004778996986770747,
      "loss": 1.1033,
      "step": 1104
    },
    {
      "epoch": 3.18,
      "learning_rate": 0.000477124781806496,
      "loss": 1.0757,
      "step": 1105
    },
    {
      "epoch": 3.18,
      "learning_rate": 0.0004763499199932093,
      "loss": 1.1045,
      "step": 1106
    },
    {
      "epoch": 3.18,
      "learning_rate": 0.00047557511510219335,
      "loss": 1.1483,
      "step": 1107
    },
    {
      "epoch": 3.18,
      "learning_rate": 0.00047480036899829003,
      "loss": 1.0756,
      "step": 1108
    },
    {
      "epoch": 3.19,
      "learning_rate": 0.0004740256835461993,
      "loss": 1.0617,
      "step": 1109
    },
    {
      "epoch": 3.19,
      "learning_rate": 0.0004732510606104754,
      "loss": 1.1516,
      "step": 1110
    },
    {
      "epoch": 3.19,
      "learning_rate": 0.00047247650205552193,
      "loss": 1.1113,
      "step": 1111
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.0004717020097455879,
      "loss": 1.079,
      "step": 1112
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.0004709275855447621,
      "loss": 1.088,
      "step": 1113
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.0004701532313169703,
      "loss": 1.1353,
      "step": 1114
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.0004693789489259694,
      "loss": 1.0283,
      "step": 1115
    },
    {
      "epoch": 3.21,
      "learning_rate": 0.0004686047402353433,
      "loss": 1.1078,
      "step": 1116
    },
    {
      "epoch": 3.21,
      "learning_rate": 0.00046783060710849886,
      "loss": 1.0536,
      "step": 1117
    },
    {
      "epoch": 3.21,
      "learning_rate": 0.00046705655140866074,
      "loss": 1.0711,
      "step": 1118
    },
    {
      "epoch": 3.22,
      "learning_rate": 0.0004662825749988675,
      "loss": 1.153,
      "step": 1119
    },
    {
      "epoch": 3.22,
      "learning_rate": 0.0004655086797419666,
      "loss": 1.1002,
      "step": 1120
    },
    {
      "epoch": 3.22,
      "learning_rate": 0.00046473486750061037,
      "loss": 1.0747,
      "step": 1121
    },
    {
      "epoch": 3.22,
      "learning_rate": 0.00046396114013725134,
      "loss": 1.0924,
      "step": 1122
    },
    {
      "epoch": 3.23,
      "learning_rate": 0.0004631874995141376,
      "loss": 1.1483,
      "step": 1123
    },
    {
      "epoch": 3.23,
      "learning_rate": 0.0004624139474933087,
      "loss": 1.0951,
      "step": 1124
    },
    {
      "epoch": 3.23,
      "learning_rate": 0.0004616404859365907,
      "loss": 1.0672,
      "step": 1125
    },
    {
      "epoch": 3.24,
      "learning_rate": 0.00046086711670559213,
      "loss": 1.1794,
      "step": 1126
    },
    {
      "epoch": 3.24,
      "learning_rate": 0.000460093841661699,
      "loss": 1.0677,
      "step": 1127
    },
    {
      "epoch": 3.24,
      "learning_rate": 0.00045932066266607093,
      "loss": 1.1208,
      "step": 1128
    },
    {
      "epoch": 3.24,
      "learning_rate": 0.0004585475815796363,
      "loss": 1.0255,
      "step": 1129
    },
    {
      "epoch": 3.25,
      "learning_rate": 0.00045777460026308776,
      "loss": 1.0785,
      "step": 1130
    },
    {
      "epoch": 3.25,
      "learning_rate": 0.0004570017205768779,
      "loss": 1.1453,
      "step": 1131
    },
    {
      "epoch": 3.25,
      "learning_rate": 0.00045622894438121465,
      "loss": 1.0464,
      "step": 1132
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.000455456273536057,
      "loss": 1.1709,
      "step": 1133
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.00045468370990111,
      "loss": 1.0298,
      "step": 1134
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.00045391125533582105,
      "loss": 1.1343,
      "step": 1135
    },
    {
      "epoch": 3.26,
      "learning_rate": 0.00045313891169937486,
      "loss": 1.118,
      "step": 1136
    },
    {
      "epoch": 3.27,
      "learning_rate": 0.00045236668085068925,
      "loss": 1.1113,
      "step": 1137
    },
    {
      "epoch": 3.27,
      "learning_rate": 0.0004515945646484105,
      "loss": 1.122,
      "step": 1138
    },
    {
      "epoch": 3.27,
      "learning_rate": 0.0004508225649509089,
      "loss": 1.1064,
      "step": 1139
    },
    {
      "epoch": 3.28,
      "learning_rate": 0.00045005068361627455,
      "loss": 1.082,
      "step": 1140
    },
    {
      "epoch": 3.28,
      "learning_rate": 0.0004492789225023123,
      "loss": 1.0921,
      "step": 1141
    },
    {
      "epoch": 3.28,
      "learning_rate": 0.0004485072834665379,
      "loss": 1.1317,
      "step": 1142
    },
    {
      "epoch": 3.28,
      "learning_rate": 0.00044773576836617336,
      "loss": 1.2059,
      "step": 1143
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.0004469643790581422,
      "loss": 1.0632,
      "step": 1144
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.0004461931173990652,
      "loss": 1.1203,
      "step": 1145
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.000445421985245256,
      "loss": 1.0624,
      "step": 1146
    },
    {
      "epoch": 3.3,
      "learning_rate": 0.0004446509844527165,
      "loss": 1.0436,
      "step": 1147
    },
    {
      "epoch": 3.3,
      "learning_rate": 0.0004438801168771327,
      "loss": 1.1602,
      "step": 1148
    },
    {
      "epoch": 3.3,
      "learning_rate": 0.00044310938437386923,
      "loss": 1.1196,
      "step": 1149
    },
    {
      "epoch": 3.3,
      "learning_rate": 0.00044233878879796625,
      "loss": 1.108,
      "step": 1150
    },
    {
      "epoch": 3.31,
      "learning_rate": 0.00044156833200413424,
      "loss": 1.0733,
      "step": 1151
    },
    {
      "epoch": 3.31,
      "learning_rate": 0.0004407980158467495,
      "loss": 1.1218,
      "step": 1152
    },
    {
      "epoch": 3.31,
      "learning_rate": 0.0004400278421798501,
      "loss": 1.1814,
      "step": 1153
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.00043925781285713097,
      "loss": 1.1814,
      "step": 1154
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.0004384879297319398,
      "loss": 1.1227,
      "step": 1155
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.0004377181946572719,
      "loss": 1.1042,
      "step": 1156
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.00043694860948576664,
      "loss": 1.1486,
      "step": 1157
    },
    {
      "epoch": 3.33,
      "learning_rate": 0.0004361791760697027,
      "loss": 1.0406,
      "step": 1158
    },
    {
      "epoch": 3.33,
      "learning_rate": 0.0004354098962609934,
      "loss": 1.1463,
      "step": 1159
    },
    {
      "epoch": 3.33,
      "learning_rate": 0.0004346407719111823,
      "loss": 1.1041,
      "step": 1160
    },
    {
      "epoch": 3.34,
      "learning_rate": 0.0004338718048714387,
      "loss": 1.0816,
      "step": 1161
    },
    {
      "epoch": 3.34,
      "learning_rate": 0.0004331029969925537,
      "loss": 1.1427,
      "step": 1162
    },
    {
      "epoch": 3.34,
      "learning_rate": 0.0004323343501249346,
      "loss": 1.0796,
      "step": 1163
    },
    {
      "epoch": 3.34,
      "learning_rate": 0.0004315658661186016,
      "loss": 1.0871,
      "step": 1164
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.00043079754682318305,
      "loss": 1.1895,
      "step": 1165
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.00043002939408791075,
      "loss": 1.1064,
      "step": 1166
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.00042926140976161554,
      "loss": 1.1318,
      "step": 1167
    },
    {
      "epoch": 3.36,
      "learning_rate": 0.0004284935956927229,
      "loss": 1.0531,
      "step": 1168
    },
    {
      "epoch": 3.36,
      "learning_rate": 0.0004277259537292487,
      "loss": 1.0917,
      "step": 1169
    },
    {
      "epoch": 3.36,
      "learning_rate": 0.00042695848571879425,
      "loss": 1.1799,
      "step": 1170
    },
    {
      "epoch": 3.36,
      "learning_rate": 0.0004261911935085425,
      "loss": 1.1172,
      "step": 1171
    },
    {
      "epoch": 3.37,
      "learning_rate": 0.0004254240789452532,
      "loss": 1.1582,
      "step": 1172
    },
    {
      "epoch": 3.37,
      "learning_rate": 0.00042465714387525844,
      "loss": 1.1074,
      "step": 1173
    },
    {
      "epoch": 3.37,
      "learning_rate": 0.0004238903901444584,
      "loss": 1.0945,
      "step": 1174
    },
    {
      "epoch": 3.38,
      "learning_rate": 0.0004231238195983167,
      "loss": 1.1258,
      "step": 1175
    },
    {
      "epoch": 3.38,
      "learning_rate": 0.0004223574340818563,
      "loss": 1.0785,
      "step": 1176
    },
    {
      "epoch": 3.38,
      "learning_rate": 0.00042159123543965434,
      "loss": 1.0812,
      "step": 1177
    },
    {
      "epoch": 3.39,
      "learning_rate": 0.00042082522551583864,
      "loss": 1.1946,
      "step": 1178
    },
    {
      "epoch": 3.39,
      "learning_rate": 0.0004200594061540826,
      "loss": 1.067,
      "step": 1179
    },
    {
      "epoch": 3.39,
      "learning_rate": 0.000419293779197601,
      "loss": 1.1152,
      "step": 1180
    },
    {
      "epoch": 3.39,
      "learning_rate": 0.00041852834648914563,
      "loss": 1.1436,
      "step": 1181
    },
    {
      "epoch": 3.4,
      "learning_rate": 0.0004177631098710005,
      "loss": 1.1282,
      "step": 1182
    },
    {
      "epoch": 3.4,
      "learning_rate": 0.0004169980711849781,
      "loss": 1.1321,
      "step": 1183
    },
    {
      "epoch": 3.4,
      "learning_rate": 0.0004162332322724139,
      "loss": 1.0552,
      "step": 1184
    },
    {
      "epoch": 3.41,
      "learning_rate": 0.000415468594974163,
      "loss": 1.1434,
      "step": 1185
    },
    {
      "epoch": 3.41,
      "learning_rate": 0.0004147041611305952,
      "loss": 0.9777,
      "step": 1186
    },
    {
      "epoch": 3.41,
      "learning_rate": 0.0004139399325815905,
      "loss": 1.1401,
      "step": 1187
    },
    {
      "epoch": 3.41,
      "learning_rate": 0.00041317591116653486,
      "loss": 1.1323,
      "step": 1188
    },
    {
      "epoch": 3.42,
      "learning_rate": 0.00041241209872431564,
      "loss": 1.139,
      "step": 1189
    },
    {
      "epoch": 3.42,
      "learning_rate": 0.0004116484970933174,
      "loss": 1.0547,
      "step": 1190
    },
    {
      "epoch": 3.42,
      "learning_rate": 0.0004108851081114169,
      "loss": 1.1146,
      "step": 1191
    },
    {
      "epoch": 3.43,
      "learning_rate": 0.0004101219336159795,
      "loss": 1.1124,
      "step": 1192
    },
    {
      "epoch": 3.43,
      "learning_rate": 0.00040935897544385424,
      "loss": 1.0973,
      "step": 1193
    },
    {
      "epoch": 3.43,
      "learning_rate": 0.0004085962354313694,
      "loss": 1.1036,
      "step": 1194
    },
    {
      "epoch": 3.43,
      "learning_rate": 0.0004078337154143281,
      "loss": 1.0389,
      "step": 1195
    },
    {
      "epoch": 3.44,
      "learning_rate": 0.00040707141722800427,
      "loss": 1.0622,
      "step": 1196
    },
    {
      "epoch": 3.44,
      "learning_rate": 0.0004063093427071376,
      "loss": 1.1476,
      "step": 1197
    },
    {
      "epoch": 3.44,
      "learning_rate": 0.00040554749368592994,
      "loss": 1.0923,
      "step": 1198
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.0004047858719980396,
      "loss": 1.0991,
      "step": 1199
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.0004040244794765783,
      "loss": 1.1234,
      "step": 1200
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.00040326331795410614,
      "loss": 1.1391,
      "step": 1201
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.0004025023892626272,
      "loss": 1.0998,
      "step": 1202
    },
    {
      "epoch": 3.46,
      "learning_rate": 0.0004017416952335849,
      "loss": 1.1592,
      "step": 1203
    },
    {
      "epoch": 3.46,
      "learning_rate": 0.00040098123769785843,
      "loss": 1.1339,
      "step": 1204
    },
    {
      "epoch": 3.46,
      "learning_rate": 0.00040022101848575755,
      "loss": 1.1047,
      "step": 1205
    },
    {
      "epoch": 3.47,
      "learning_rate": 0.0003994610394270178,
      "loss": 1.1565,
      "step": 1206
    },
    {
      "epoch": 3.47,
      "learning_rate": 0.0003987013023507975,
      "loss": 1.011,
      "step": 1207
    },
    {
      "epoch": 3.47,
      "learning_rate": 0.00039794180908567223,
      "loss": 1.1354,
      "step": 1208
    },
    {
      "epoch": 3.47,
      "learning_rate": 0.00039718256145963073,
      "loss": 1.1952,
      "step": 1209
    },
    {
      "epoch": 3.48,
      "learning_rate": 0.00039642356130007074,
      "loss": 1.0661,
      "step": 1210
    },
    {
      "epoch": 3.48,
      "learning_rate": 0.00039566481043379413,
      "loss": 1.1195,
      "step": 1211
    },
    {
      "epoch": 3.48,
      "learning_rate": 0.00039490631068700313,
      "loss": 1.0968,
      "step": 1212
    },
    {
      "epoch": 3.49,
      "learning_rate": 0.00039414806388529477,
      "loss": 1.1565,
      "step": 1213
    },
    {
      "epoch": 3.49,
      "learning_rate": 0.0003933900718536579,
      "loss": 1.0669,
      "step": 1214
    },
    {
      "epoch": 3.49,
      "learning_rate": 0.0003926323364164684,
      "loss": 1.139,
      "step": 1215
    },
    {
      "epoch": 3.49,
      "learning_rate": 0.000391874859397484,
      "loss": 1.1439,
      "step": 1216
    },
    {
      "epoch": 3.5,
      "learning_rate": 0.00039111764261984077,
      "loss": 1.1568,
      "step": 1217
    },
    {
      "epoch": 3.5,
      "learning_rate": 0.0003903606879060483,
      "loss": 1.1454,
      "step": 1218
    },
    {
      "epoch": 3.5,
      "learning_rate": 0.0003896039970779857,
      "loss": 1.1343,
      "step": 1219
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.0003888475719568961,
      "loss": 1.0951,
      "step": 1220
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.00038809141436338434,
      "loss": 1.0398,
      "step": 1221
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.00038733552611741053,
      "loss": 1.1971,
      "step": 1222
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.0003865799090382866,
      "loss": 1.1419,
      "step": 1223
    },
    {
      "epoch": 3.52,
      "learning_rate": 0.0003858245649446721,
      "loss": 1.0533,
      "step": 1224
    },
    {
      "epoch": 3.52,
      "learning_rate": 0.0003850694956545694,
      "loss": 1.1525,
      "step": 1225
    },
    {
      "epoch": 3.52,
      "learning_rate": 0.0003843147029853194,
      "loss": 1.0634,
      "step": 1226
    },
    {
      "epoch": 3.53,
      "learning_rate": 0.0003835601887535971,
      "loss": 1.1188,
      "step": 1227
    },
    {
      "epoch": 3.53,
      "learning_rate": 0.0003828059547754077,
      "loss": 1.2265,
      "step": 1228
    },
    {
      "epoch": 3.53,
      "learning_rate": 0.0003820520028660815,
      "loss": 1.1111,
      "step": 1229
    },
    {
      "epoch": 3.53,
      "learning_rate": 0.0003812983348402703,
      "loss": 1.0358,
      "step": 1230
    },
    {
      "epoch": 3.54,
      "learning_rate": 0.00038054495251194217,
      "loss": 1.1176,
      "step": 1231
    },
    {
      "epoch": 3.54,
      "learning_rate": 0.0003797918576943779,
      "loss": 1.0404,
      "step": 1232
    },
    {
      "epoch": 3.54,
      "learning_rate": 0.0003790390522001662,
      "loss": 1.1569,
      "step": 1233
    },
    {
      "epoch": 3.55,
      "learning_rate": 0.0003782865378411993,
      "loss": 1.1015,
      "step": 1234
    },
    {
      "epoch": 3.55,
      "learning_rate": 0.00037753431642866876,
      "loss": 1.0963,
      "step": 1235
    },
    {
      "epoch": 3.55,
      "learning_rate": 0.0003767823897730612,
      "loss": 1.0757,
      "step": 1236
    },
    {
      "epoch": 3.55,
      "learning_rate": 0.0003760307596841536,
      "loss": 1.1504,
      "step": 1237
    },
    {
      "epoch": 3.56,
      "learning_rate": 0.0003752794279710094,
      "loss": 1.0463,
      "step": 1238
    },
    {
      "epoch": 3.56,
      "learning_rate": 0.00037452839644197354,
      "loss": 1.0775,
      "step": 1239
    },
    {
      "epoch": 3.56,
      "learning_rate": 0.00037377766690466885,
      "loss": 1.0404,
      "step": 1240
    },
    {
      "epoch": 3.57,
      "learning_rate": 0.0003730272411659912,
      "loss": 1.1212,
      "step": 1241
    },
    {
      "epoch": 3.57,
      "learning_rate": 0.0003722771210321048,
      "loss": 1.1282,
      "step": 1242
    },
    {
      "epoch": 3.57,
      "learning_rate": 0.000371527308308439,
      "loss": 1.1237,
      "step": 1243
    },
    {
      "epoch": 3.57,
      "learning_rate": 0.0003707778047996828,
      "loss": 1.0828,
      "step": 1244
    },
    {
      "epoch": 3.58,
      "learning_rate": 0.00037002861230978133,
      "loss": 1.1194,
      "step": 1245
    },
    {
      "epoch": 3.58,
      "learning_rate": 0.0003692797326419307,
      "loss": 1.1095,
      "step": 1246
    },
    {
      "epoch": 3.58,
      "learning_rate": 0.0003685311675985745,
      "loss": 1.0495,
      "step": 1247
    },
    {
      "epoch": 3.59,
      "learning_rate": 0.00036778291898139905,
      "loss": 1.1702,
      "step": 1248
    },
    {
      "epoch": 3.59,
      "learning_rate": 0.00036703498859132845,
      "loss": 1.1353,
      "step": 1249
    },
    {
      "epoch": 3.59,
      "learning_rate": 0.00036628737822852177,
      "loss": 1.1516,
      "step": 1250
    },
    {
      "epoch": 3.59,
      "learning_rate": 0.00036554008969236717,
      "loss": 1.1132,
      "step": 1251
    },
    {
      "epoch": 3.6,
      "learning_rate": 0.00036479312478147864,
      "loss": 1.0936,
      "step": 1252
    },
    {
      "epoch": 3.6,
      "learning_rate": 0.0003640464852936909,
      "loss": 1.0472,
      "step": 1253
    },
    {
      "epoch": 3.6,
      "learning_rate": 0.00036330017302605577,
      "loss": 1.1039,
      "step": 1254
    },
    {
      "epoch": 3.61,
      "learning_rate": 0.00036255418977483746,
      "loss": 1.0149,
      "step": 1255
    },
    {
      "epoch": 3.61,
      "learning_rate": 0.00036180853733550766,
      "loss": 1.1908,
      "step": 1256
    },
    {
      "epoch": 3.61,
      "learning_rate": 0.0003610632175027427,
      "loss": 1.0795,
      "step": 1257
    },
    {
      "epoch": 3.61,
      "learning_rate": 0.0003603182320704179,
      "loss": 1.0151,
      "step": 1258
    },
    {
      "epoch": 3.62,
      "learning_rate": 0.0003595735828316037,
      "loss": 1.1572,
      "step": 1259
    },
    {
      "epoch": 3.62,
      "learning_rate": 0.0003588292715785617,
      "loss": 1.0668,
      "step": 1260
    },
    {
      "epoch": 3.62,
      "learning_rate": 0.00035808530010273987,
      "loss": 1.0474,
      "step": 1261
    },
    {
      "epoch": 3.63,
      "learning_rate": 0.00035734167019476845,
      "loss": 1.1472,
      "step": 1262
    },
    {
      "epoch": 3.63,
      "learning_rate": 0.00035659838364445503,
      "loss": 1.0584,
      "step": 1263
    },
    {
      "epoch": 3.63,
      "learning_rate": 0.0003558554422407814,
      "loss": 1.1284,
      "step": 1264
    },
    {
      "epoch": 3.64,
      "learning_rate": 0.0003551128477718985,
      "loss": 1.1419,
      "step": 1265
    },
    {
      "epoch": 3.64,
      "learning_rate": 0.00035437060202512226,
      "loss": 1.0989,
      "step": 1266
    },
    {
      "epoch": 3.64,
      "learning_rate": 0.00035362870678692926,
      "loss": 1.1245,
      "step": 1267
    },
    {
      "epoch": 3.64,
      "learning_rate": 0.00035288716384295236,
      "loss": 1.1798,
      "step": 1268
    },
    {
      "epoch": 3.65,
      "learning_rate": 0.0003521459749779768,
      "loss": 1.0601,
      "step": 1269
    },
    {
      "epoch": 3.65,
      "learning_rate": 0.0003514051419759349,
      "loss": 1.1217,
      "step": 1270
    },
    {
      "epoch": 3.65,
      "learning_rate": 0.00035066466661990315,
      "loss": 1.0551,
      "step": 1271
    },
    {
      "epoch": 3.66,
      "learning_rate": 0.00034992455069209717,
      "loss": 1.1149,
      "step": 1272
    },
    {
      "epoch": 3.66,
      "learning_rate": 0.00034918479597386727,
      "loss": 1.0224,
      "step": 1273
    },
    {
      "epoch": 3.66,
      "learning_rate": 0.0003484454042456945,
      "loss": 1.138,
      "step": 1274
    },
    {
      "epoch": 3.66,
      "learning_rate": 0.00034770637728718607,
      "loss": 1.1843,
      "step": 1275
    },
    {
      "epoch": 3.67,
      "learning_rate": 0.00034696771687707176,
      "loss": 1.1707,
      "step": 1276
    },
    {
      "epoch": 3.67,
      "learning_rate": 0.0003462294247931982,
      "loss": 1.1132,
      "step": 1277
    },
    {
      "epoch": 3.67,
      "learning_rate": 0.00034549150281252633,
      "loss": 1.0982,
      "step": 1278
    },
    {
      "epoch": 3.68,
      "learning_rate": 0.0003447539527111261,
      "loss": 1.0505,
      "step": 1279
    },
    {
      "epoch": 3.68,
      "learning_rate": 0.00034401677626417224,
      "loss": 1.1529,
      "step": 1280
    },
    {
      "epoch": 3.68,
      "learning_rate": 0.00034327997524594026,
      "loss": 1.1424,
      "step": 1281
    },
    {
      "epoch": 3.68,
      "learning_rate": 0.0003425435514298021,
      "loss": 1.1188,
      "step": 1282
    },
    {
      "epoch": 3.69,
      "learning_rate": 0.0003418075065882217,
      "loss": 1.0826,
      "step": 1283
    },
    {
      "epoch": 3.69,
      "learning_rate": 0.00034107184249275116,
      "loss": 1.1197,
      "step": 1284
    },
    {
      "epoch": 3.69,
      "learning_rate": 0.0003403365609140256,
      "loss": 1.2033,
      "step": 1285
    },
    {
      "epoch": 3.7,
      "learning_rate": 0.0003396016636217601,
      "loss": 1.0914,
      "step": 1286
    },
    {
      "epoch": 3.7,
      "learning_rate": 0.00033886715238474455,
      "loss": 1.1586,
      "step": 1287
    },
    {
      "epoch": 3.7,
      "learning_rate": 0.0003381330289708395,
      "loss": 1.1275,
      "step": 1288
    },
    {
      "epoch": 3.7,
      "learning_rate": 0.00033739929514697245,
      "loss": 1.0731,
      "step": 1289
    },
    {
      "epoch": 3.71,
      "learning_rate": 0.0003366659526791329,
      "loss": 1.028,
      "step": 1290
    },
    {
      "epoch": 3.71,
      "learning_rate": 0.0003359330033323686,
      "loss": 1.0471,
      "step": 1291
    },
    {
      "epoch": 3.71,
      "learning_rate": 0.00033520044887078095,
      "loss": 1.0809,
      "step": 1292
    },
    {
      "epoch": 3.72,
      "learning_rate": 0.000334468291057521,
      "loss": 1.0576,
      "step": 1293
    },
    {
      "epoch": 3.72,
      "learning_rate": 0.0003337365316547852,
      "loss": 1.1566,
      "step": 1294
    },
    {
      "epoch": 3.72,
      "learning_rate": 0.0003330051724238109,
      "loss": 1.0915,
      "step": 1295
    },
    {
      "epoch": 3.72,
      "learning_rate": 0.00033227421512487253,
      "loss": 1.1704,
      "step": 1296
    },
    {
      "epoch": 3.73,
      "learning_rate": 0.0003315436615172769,
      "loss": 1.1026,
      "step": 1297
    },
    {
      "epoch": 3.73,
      "learning_rate": 0.0003308135133593595,
      "loss": 1.112,
      "step": 1298
    },
    {
      "epoch": 3.73,
      "learning_rate": 0.0003300837724084795,
      "loss": 1.1145,
      "step": 1299
    },
    {
      "epoch": 3.74,
      "learning_rate": 0.0003293544404210164,
      "loss": 1.0967,
      "step": 1300
    },
    {
      "epoch": 3.74,
      "learning_rate": 0.0003286255191523652,
      "loss": 1.0699,
      "step": 1301
    },
    {
      "epoch": 3.74,
      "learning_rate": 0.00032789701035693244,
      "loss": 1.0608,
      "step": 1302
    },
    {
      "epoch": 3.74,
      "learning_rate": 0.00032716891578813166,
      "loss": 1.1206,
      "step": 1303
    },
    {
      "epoch": 3.75,
      "learning_rate": 0.0003264412371983797,
      "loss": 1.1646,
      "step": 1304
    },
    {
      "epoch": 3.75,
      "learning_rate": 0.0003257139763390925,
      "loss": 1.0123,
      "step": 1305
    },
    {
      "epoch": 3.75,
      "learning_rate": 0.0003249871349606796,
      "loss": 1.1629,
      "step": 1306
    },
    {
      "epoch": 3.76,
      "learning_rate": 0.0003242607148125418,
      "loss": 1.0411,
      "step": 1307
    },
    {
      "epoch": 3.76,
      "learning_rate": 0.00032353471764306563,
      "loss": 1.0418,
      "step": 1308
    },
    {
      "epoch": 3.76,
      "learning_rate": 0.0003228091451996198,
      "loss": 1.0995,
      "step": 1309
    },
    {
      "epoch": 3.76,
      "learning_rate": 0.0003220839992285505,
      "loss": 1.0431,
      "step": 1310
    },
    {
      "epoch": 3.77,
      "learning_rate": 0.00032135928147517803,
      "loss": 1.0853,
      "step": 1311
    },
    {
      "epoch": 3.77,
      "learning_rate": 0.0003206349936837915,
      "loss": 1.0508,
      "step": 1312
    },
    {
      "epoch": 3.77,
      "learning_rate": 0.0003199111375976449,
      "loss": 1.041,
      "step": 1313
    },
    {
      "epoch": 3.78,
      "learning_rate": 0.00031918771495895393,
      "loss": 1.0864,
      "step": 1314
    },
    {
      "epoch": 3.78,
      "learning_rate": 0.00031846472750889066,
      "loss": 1.1069,
      "step": 1315
    },
    {
      "epoch": 3.78,
      "learning_rate": 0.0003177421769875796,
      "loss": 1.1347,
      "step": 1316
    },
    {
      "epoch": 3.78,
      "learning_rate": 0.0003170200651340939,
      "loss": 1.1424,
      "step": 1317
    },
    {
      "epoch": 3.79,
      "learning_rate": 0.00031629839368645086,
      "loss": 1.0853,
      "step": 1318
    },
    {
      "epoch": 3.79,
      "learning_rate": 0.00031557716438160787,
      "loss": 1.105,
      "step": 1319
    },
    {
      "epoch": 3.79,
      "learning_rate": 0.0003148563789554575,
      "loss": 1.133,
      "step": 1320
    },
    {
      "epoch": 3.8,
      "learning_rate": 0.00031413603914282474,
      "loss": 1.0474,
      "step": 1321
    },
    {
      "epoch": 3.8,
      "learning_rate": 0.00031341614667746164,
      "loss": 1.099,
      "step": 1322
    },
    {
      "epoch": 3.8,
      "learning_rate": 0.00031269670329204396,
      "loss": 1.0156,
      "step": 1323
    },
    {
      "epoch": 3.8,
      "learning_rate": 0.0003119777107181662,
      "loss": 1.157,
      "step": 1324
    },
    {
      "epoch": 3.81,
      "learning_rate": 0.0003112591706863378,
      "loss": 1.0669,
      "step": 1325
    },
    {
      "epoch": 3.81,
      "learning_rate": 0.00031054108492597956,
      "loss": 1.1242,
      "step": 1326
    },
    {
      "epoch": 3.81,
      "learning_rate": 0.00030982345516541776,
      "loss": 1.1134,
      "step": 1327
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.0003091062831318825,
      "loss": 1.0876,
      "step": 1328
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.00030838957055150134,
      "loss": 1.1316,
      "step": 1329
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.00030767331914929635,
      "loss": 1.126,
      "step": 1330
    },
    {
      "epoch": 3.82,
      "learning_rate": 0.0003069575306491794,
      "loss": 1.0188,
      "step": 1331
    },
    {
      "epoch": 3.83,
      "learning_rate": 0.0003062422067739485,
      "loss": 1.0437,
      "step": 1332
    },
    {
      "epoch": 3.83,
      "learning_rate": 0.000305527349245283,
      "loss": 1.0687,
      "step": 1333
    },
    {
      "epoch": 3.83,
      "learning_rate": 0.00030481295978374034,
      "loss": 1.1534,
      "step": 1334
    },
    {
      "epoch": 3.84,
      "learning_rate": 0.0003040990401087508,
      "loss": 1.1119,
      "step": 1335
    },
    {
      "epoch": 3.84,
      "learning_rate": 0.0003033855919386143,
      "loss": 1.1918,
      "step": 1336
    },
    {
      "epoch": 3.84,
      "learning_rate": 0.0003026726169904959,
      "loss": 1.0773,
      "step": 1337
    },
    {
      "epoch": 3.84,
      "learning_rate": 0.0003019601169804216,
      "loss": 1.1186,
      "step": 1338
    },
    {
      "epoch": 3.85,
      "learning_rate": 0.0003012480936232744,
      "loss": 1.0725,
      "step": 1339
    },
    {
      "epoch": 3.85,
      "learning_rate": 0.00030053654863278987,
      "loss": 1.1453,
      "step": 1340
    },
    {
      "epoch": 3.85,
      "learning_rate": 0.0002998254837215526,
      "loss": 1.1075,
      "step": 1341
    },
    {
      "epoch": 3.86,
      "learning_rate": 0.00029911490060099114,
      "loss": 1.0844,
      "step": 1342
    },
    {
      "epoch": 3.86,
      "learning_rate": 0.000298404800981375,
      "loss": 1.1322,
      "step": 1343
    },
    {
      "epoch": 3.86,
      "learning_rate": 0.0002976951865718095,
      "loss": 1.0411,
      "step": 1344
    },
    {
      "epoch": 3.86,
      "learning_rate": 0.0002969860590802326,
      "loss": 1.0456,
      "step": 1345
    },
    {
      "epoch": 3.87,
      "learning_rate": 0.0002962774202134098,
      "loss": 1.1246,
      "step": 1346
    },
    {
      "epoch": 3.87,
      "learning_rate": 0.00029556927167693104,
      "loss": 1.1682,
      "step": 1347
    },
    {
      "epoch": 3.87,
      "learning_rate": 0.00029486161517520596,
      "loss": 1.0113,
      "step": 1348
    },
    {
      "epoch": 3.88,
      "learning_rate": 0.00029415445241145967,
      "loss": 1.0967,
      "step": 1349
    },
    {
      "epoch": 3.88,
      "learning_rate": 0.00029344778508772916,
      "loss": 1.1221,
      "step": 1350
    },
    {
      "epoch": 3.88,
      "learning_rate": 0.0002927416149048591,
      "loss": 1.1205,
      "step": 1351
    },
    {
      "epoch": 3.89,
      "learning_rate": 0.00029203594356249724,
      "loss": 1.1395,
      "step": 1352
    },
    {
      "epoch": 3.89,
      "learning_rate": 0.0002913307727590911,
      "loss": 1.1085,
      "step": 1353
    },
    {
      "epoch": 3.89,
      "learning_rate": 0.0002906261041918831,
      "loss": 1.1068,
      "step": 1354
    },
    {
      "epoch": 3.89,
      "learning_rate": 0.0002899219395569075,
      "loss": 1.0502,
      "step": 1355
    },
    {
      "epoch": 3.9,
      "learning_rate": 0.0002892182805489846,
      "loss": 1.0593,
      "step": 1356
    },
    {
      "epoch": 3.9,
      "learning_rate": 0.0002885151288617185,
      "loss": 1.063,
      "step": 1357
    },
    {
      "epoch": 3.9,
      "learning_rate": 0.0002878124861874923,
      "loss": 1.076,
      "step": 1358
    },
    {
      "epoch": 3.91,
      "learning_rate": 0.00028711035421746366,
      "loss": 1.1588,
      "step": 1359
    },
    {
      "epoch": 3.91,
      "learning_rate": 0.0002864087346415613,
      "loss": 1.09,
      "step": 1360
    },
    {
      "epoch": 3.91,
      "learning_rate": 0.0002857076291484801,
      "loss": 1.1083,
      "step": 1361
    },
    {
      "epoch": 3.91,
      "learning_rate": 0.00028500703942567875,
      "loss": 1.0771,
      "step": 1362
    },
    {
      "epoch": 3.92,
      "learning_rate": 0.0002843069671593734,
      "loss": 1.0822,
      "step": 1363
    },
    {
      "epoch": 3.92,
      "learning_rate": 0.0002836074140345352,
      "loss": 1.0481,
      "step": 1364
    },
    {
      "epoch": 3.92,
      "learning_rate": 0.00028290838173488595,
      "loss": 1.1627,
      "step": 1365
    },
    {
      "epoch": 3.93,
      "learning_rate": 0.00028220987194289376,
      "loss": 1.1189,
      "step": 1366
    },
    {
      "epoch": 3.93,
      "learning_rate": 0.0002815118863397694,
      "loss": 1.1086,
      "step": 1367
    },
    {
      "epoch": 3.93,
      "learning_rate": 0.00028081442660546124,
      "loss": 1.0687,
      "step": 1368
    },
    {
      "epoch": 3.93,
      "learning_rate": 0.000280117494418653,
      "loss": 1.0745,
      "step": 1369
    },
    {
      "epoch": 3.94,
      "learning_rate": 0.000279421091456758,
      "loss": 1.058,
      "step": 1370
    },
    {
      "epoch": 3.94,
      "learning_rate": 0.00027872521939591556,
      "loss": 1.0708,
      "step": 1371
    },
    {
      "epoch": 3.94,
      "learning_rate": 0.00027802987991098815,
      "loss": 1.1077,
      "step": 1372
    },
    {
      "epoch": 3.95,
      "learning_rate": 0.0002773350746755553,
      "loss": 1.0548,
      "step": 1373
    },
    {
      "epoch": 3.95,
      "learning_rate": 0.0002766408053619118,
      "loss": 1.1224,
      "step": 1374
    },
    {
      "epoch": 3.95,
      "learning_rate": 0.00027594707364106137,
      "loss": 1.0293,
      "step": 1375
    },
    {
      "epoch": 3.95,
      "learning_rate": 0.00027525388118271495,
      "loss": 1.1254,
      "step": 1376
    },
    {
      "epoch": 3.96,
      "learning_rate": 0.00027456122965528474,
      "loss": 1.0609,
      "step": 1377
    },
    {
      "epoch": 3.96,
      "learning_rate": 0.0002738691207258812,
      "loss": 1.0853,
      "step": 1378
    },
    {
      "epoch": 3.96,
      "learning_rate": 0.0002731775560603092,
      "loss": 1.0399,
      "step": 1379
    },
    {
      "epoch": 3.97,
      "learning_rate": 0.00027248653732306317,
      "loss": 1.101,
      "step": 1380
    },
    {
      "epoch": 3.97,
      "learning_rate": 0.000271796066177324,
      "loss": 1.0538,
      "step": 1381
    },
    {
      "epoch": 3.97,
      "learning_rate": 0.00027110614428495396,
      "loss": 1.1322,
      "step": 1382
    },
    {
      "epoch": 3.97,
      "learning_rate": 0.0002704167733064941,
      "loss": 1.1962,
      "step": 1383
    },
    {
      "epoch": 3.98,
      "learning_rate": 0.00026972795490115943,
      "loss": 1.077,
      "step": 1384
    },
    {
      "epoch": 3.98,
      "learning_rate": 0.00026903969072683396,
      "loss": 1.0896,
      "step": 1385
    },
    {
      "epoch": 3.98,
      "learning_rate": 0.00026835198244006924,
      "loss": 1.0934,
      "step": 1386
    },
    {
      "epoch": 3.99,
      "learning_rate": 0.0002676648316960777,
      "loss": 1.0652,
      "step": 1387
    },
    {
      "epoch": 3.99,
      "learning_rate": 0.0002669782401487307,
      "loss": 1.0697,
      "step": 1388
    },
    {
      "epoch": 3.99,
      "learning_rate": 0.0002662922094505529,
      "loss": 1.0753,
      "step": 1389
    },
    {
      "epoch": 3.99,
      "learning_rate": 0.0002656067412527197,
      "loss": 1.0786,
      "step": 1390
    },
    {
      "epoch": 4.0,
      "learning_rate": 0.0002649218372050528,
      "loss": 1.037,
      "step": 1391
    },
    {
      "epoch": 4.0,
      "learning_rate": 0.0002642374989560149,
      "loss": 0.9392,
      "step": 1392
    },
    {
      "epoch": 4.0,
      "learning_rate": 0.00026355372815270835,
      "loss": 0.9085,
      "step": 1393
    },
    {
      "epoch": 4.01,
      "learning_rate": 0.00026287052644086866,
      "loss": 1.068,
      "step": 1394
    },
    {
      "epoch": 4.01,
      "learning_rate": 0.00026218789546486235,
      "loss": 0.9927,
      "step": 1395
    },
    {
      "epoch": 4.01,
      "learning_rate": 0.000261505836867682,
      "loss": 1.0365,
      "step": 1396
    },
    {
      "epoch": 4.01,
      "learning_rate": 0.00026082435229094224,
      "loss": 0.9705,
      "step": 1397
    },
    {
      "epoch": 4.02,
      "learning_rate": 0.0002601434433748771,
      "loss": 1.0147,
      "step": 1398
    },
    {
      "epoch": 4.02,
      "learning_rate": 0.00025946311175833336,
      "loss": 1.0635,
      "step": 1399
    },
    {
      "epoch": 4.02,
      "learning_rate": 0.0002587833590787699,
      "loss": 0.9752,
      "step": 1400
    },
    {
      "epoch": 4.03,
      "learning_rate": 0.0002581041869722519,
      "loss": 1.0546,
      "step": 1401
    },
    {
      "epoch": 4.03,
      "learning_rate": 0.0002574255970734464,
      "loss": 0.9612,
      "step": 1402
    },
    {
      "epoch": 4.03,
      "learning_rate": 0.00025674759101562006,
      "loss": 1.0275,
      "step": 1403
    },
    {
      "epoch": 4.03,
      "learning_rate": 0.0002560701704306336,
      "loss": 0.9698,
      "step": 1404
    },
    {
      "epoch": 4.04,
      "learning_rate": 0.0002553933369489395,
      "loss": 0.9866,
      "step": 1405
    },
    {
      "epoch": 4.04,
      "learning_rate": 0.0002547170921995757,
      "loss": 1.0127,
      "step": 1406
    },
    {
      "epoch": 4.04,
      "learning_rate": 0.0002540414378101647,
      "loss": 1.0202,
      "step": 1407
    },
    {
      "epoch": 4.05,
      "learning_rate": 0.00025336637540690737,
      "loss": 1.006,
      "step": 1408
    },
    {
      "epoch": 4.05,
      "learning_rate": 0.00025269190661457956,
      "loss": 1.0233,
      "step": 1409
    },
    {
      "epoch": 4.05,
      "learning_rate": 0.000252018033056529,
      "loss": 1.004,
      "step": 1410
    },
    {
      "epoch": 4.05,
      "learning_rate": 0.00025134475635467,
      "loss": 0.983,
      "step": 1411
    },
    {
      "epoch": 4.06,
      "learning_rate": 0.00025067207812948125,
      "loss": 0.9769,
      "step": 1412
    },
    {
      "epoch": 4.06,
      "learning_rate": 0.0002500000000000001,
      "loss": 1.0267,
      "step": 1413
    },
    {
      "epoch": 4.06,
      "learning_rate": 0.0002493285235838199,
      "loss": 0.9571,
      "step": 1414
    },
    {
      "epoch": 4.07,
      "learning_rate": 0.0002486576504970862,
      "loss": 1.0481,
      "step": 1415
    },
    {
      "epoch": 4.07,
      "learning_rate": 0.0002479873823544916,
      "loss": 1.0873,
      "step": 1416
    },
    {
      "epoch": 4.07,
      "learning_rate": 0.00024731772076927363,
      "loss": 0.9655,
      "step": 1417
    },
    {
      "epoch": 4.07,
      "learning_rate": 0.00024664866735320885,
      "loss": 1.0679,
      "step": 1418
    },
    {
      "epoch": 4.08,
      "learning_rate": 0.0002459802237166111,
      "loss": 1.0472,
      "step": 1419
    },
    {
      "epoch": 4.08,
      "learning_rate": 0.0002453123914683259,
      "loss": 1.0581,
      "step": 1420
    },
    {
      "epoch": 4.08,
      "learning_rate": 0.00024464517221572723,
      "loss": 1.0082,
      "step": 1421
    },
    {
      "epoch": 4.09,
      "learning_rate": 0.0002439785675647143,
      "loss": 1.0309,
      "step": 1422
    },
    {
      "epoch": 4.09,
      "learning_rate": 0.00024331257911970627,
      "loss": 1.0537,
      "step": 1423
    },
    {
      "epoch": 4.09,
      "learning_rate": 0.0002426472084836399,
      "loss": 0.9972,
      "step": 1424
    },
    {
      "epoch": 4.09,
      "learning_rate": 0.00024198245725796426,
      "loss": 1.0592,
      "step": 1425
    },
    {
      "epoch": 4.1,
      "learning_rate": 0.0002413183270426384,
      "loss": 0.9019,
      "step": 1426
    },
    {
      "epoch": 4.1,
      "learning_rate": 0.000240654819436126,
      "loss": 1.0839,
      "step": 1427
    },
    {
      "epoch": 4.1,
      "learning_rate": 0.00023999193603539232,
      "loss": 1.066,
      "step": 1428
    },
    {
      "epoch": 4.11,
      "learning_rate": 0.00023932967843590076,
      "loss": 1.0023,
      "step": 1429
    },
    {
      "epoch": 4.11,
      "learning_rate": 0.00023866804823160777,
      "loss": 1.0005,
      "step": 1430
    },
    {
      "epoch": 4.11,
      "learning_rate": 0.00023800704701496051,
      "loss": 1.0193,
      "step": 1431
    },
    {
      "epoch": 4.11,
      "learning_rate": 0.0002373466763768915,
      "loss": 1.0218,
      "step": 1432
    },
    {
      "epoch": 4.12,
      "learning_rate": 0.00023668693790681634,
      "loss": 1.0485,
      "step": 1433
    },
    {
      "epoch": 4.12,
      "learning_rate": 0.00023602783319262843,
      "loss": 0.9926,
      "step": 1434
    },
    {
      "epoch": 4.12,
      "learning_rate": 0.0002353693638206959,
      "loss": 0.9638,
      "step": 1435
    },
    {
      "epoch": 4.13,
      "learning_rate": 0.00023471153137585821,
      "loss": 1.0285,
      "step": 1436
    },
    {
      "epoch": 4.13,
      "learning_rate": 0.0002340543374414212,
      "loss": 1.0193,
      "step": 1437
    },
    {
      "epoch": 4.13,
      "learning_rate": 0.0002333977835991545,
      "loss": 1.0685,
      "step": 1438
    },
    {
      "epoch": 4.14,
      "learning_rate": 0.00023274187142928644,
      "loss": 0.9989,
      "step": 1439
    },
    {
      "epoch": 4.14,
      "learning_rate": 0.00023208660251050156,
      "loss": 1.0505,
      "step": 1440
    },
    {
      "epoch": 4.14,
      "learning_rate": 0.00023143197841993634,
      "loss": 1.0146,
      "step": 1441
    },
    {
      "epoch": 4.14,
      "learning_rate": 0.00023077800073317413,
      "loss": 0.9759,
      "step": 1442
    },
    {
      "epoch": 4.15,
      "learning_rate": 0.00023012467102424372,
      "loss": 1.0419,
      "step": 1443
    },
    {
      "epoch": 4.15,
      "learning_rate": 0.00022947199086561344,
      "loss": 1.012,
      "step": 1444
    },
    {
      "epoch": 4.15,
      "learning_rate": 0.00022881996182818888,
      "loss": 0.9735,
      "step": 1445
    },
    {
      "epoch": 4.16,
      "learning_rate": 0.00022816858548130837,
      "loss": 0.9502,
      "step": 1446
    },
    {
      "epoch": 4.16,
      "learning_rate": 0.00022751786339273878,
      "loss": 0.9871,
      "step": 1447
    },
    {
      "epoch": 4.16,
      "learning_rate": 0.00022686779712867316,
      "loss": 1.0013,
      "step": 1448
    },
    {
      "epoch": 4.16,
      "learning_rate": 0.00022621838825372491,
      "loss": 0.9927,
      "step": 1449
    },
    {
      "epoch": 4.17,
      "learning_rate": 0.00022556963833092648,
      "loss": 1.0906,
      "step": 1450
    },
    {
      "epoch": 4.17,
      "learning_rate": 0.00022492154892172318,
      "loss": 0.9828,
      "step": 1451
    },
    {
      "epoch": 4.17,
      "learning_rate": 0.00022427412158597133,
      "loss": 1.0019,
      "step": 1452
    },
    {
      "epoch": 4.18,
      "learning_rate": 0.00022362735788193367,
      "loss": 1.0062,
      "step": 1453
    },
    {
      "epoch": 4.18,
      "learning_rate": 0.00022298125936627517,
      "loss": 0.9921,
      "step": 1454
    },
    {
      "epoch": 4.18,
      "learning_rate": 0.00022233582759406063,
      "loss": 0.9761,
      "step": 1455
    },
    {
      "epoch": 4.18,
      "learning_rate": 0.0002216910641187488,
      "loss": 1.0569,
      "step": 1456
    },
    {
      "epoch": 4.19,
      "learning_rate": 0.00022104697049219114,
      "loss": 0.9876,
      "step": 1457
    },
    {
      "epoch": 4.19,
      "learning_rate": 0.00022040354826462666,
      "loss": 1.1193,
      "step": 1458
    },
    {
      "epoch": 4.19,
      "learning_rate": 0.00021976079898467776,
      "loss": 1.037,
      "step": 1459
    },
    {
      "epoch": 4.2,
      "learning_rate": 0.00021911872419934803,
      "loss": 1.0456,
      "step": 1460
    },
    {
      "epoch": 4.2,
      "learning_rate": 0.0002184773254540169,
      "loss": 0.9441,
      "step": 1461
    },
    {
      "epoch": 4.2,
      "learning_rate": 0.00021783660429243747,
      "loss": 1.0454,
      "step": 1462
    },
    {
      "epoch": 4.2,
      "learning_rate": 0.0002171965622567308,
      "loss": 1.0297,
      "step": 1463
    },
    {
      "epoch": 4.21,
      "learning_rate": 0.0002165572008873845,
      "loss": 0.9828,
      "step": 1464
    },
    {
      "epoch": 4.21,
      "learning_rate": 0.00021591852172324772,
      "loss": 0.9823,
      "step": 1465
    },
    {
      "epoch": 4.21,
      "learning_rate": 0.00021528052630152706,
      "loss": 0.9843,
      "step": 1466
    },
    {
      "epoch": 4.22,
      "learning_rate": 0.0002146432161577842,
      "loss": 1.0309,
      "step": 1467
    },
    {
      "epoch": 4.22,
      "learning_rate": 0.00021400659282593083,
      "loss": 1.0149,
      "step": 1468
    },
    {
      "epoch": 4.22,
      "learning_rate": 0.00021337065783822606,
      "loss": 0.9966,
      "step": 1469
    },
    {
      "epoch": 4.22,
      "learning_rate": 0.00021273541272527204,
      "loss": 1.0574,
      "step": 1470
    },
    {
      "epoch": 4.23,
      "learning_rate": 0.00021210085901601024,
      "loss": 0.9858,
      "step": 1471
    },
    {
      "epoch": 4.23,
      "learning_rate": 0.00021146699823771864,
      "loss": 1.0257,
      "step": 1472
    },
    {
      "epoch": 4.23,
      "learning_rate": 0.00021083383191600674,
      "loss": 1.0065,
      "step": 1473
    },
    {
      "epoch": 4.24,
      "learning_rate": 0.00021020136157481328,
      "loss": 0.9876,
      "step": 1474
    },
    {
      "epoch": 4.24,
      "learning_rate": 0.0002095695887364012,
      "loss": 0.9294,
      "step": 1475
    },
    {
      "epoch": 4.24,
      "learning_rate": 0.00020893851492135535,
      "loss": 1.0694,
      "step": 1476
    },
    {
      "epoch": 4.24,
      "learning_rate": 0.00020830814164857754,
      "loss": 1.0201,
      "step": 1477
    },
    {
      "epoch": 4.25,
      "learning_rate": 0.0002076784704352835,
      "loss": 1.0011,
      "step": 1478
    },
    {
      "epoch": 4.25,
      "learning_rate": 0.00020704950279699987,
      "loss": 1.067,
      "step": 1479
    },
    {
      "epoch": 4.25,
      "learning_rate": 0.00020642124024755892,
      "loss": 0.985,
      "step": 1480
    },
    {
      "epoch": 4.26,
      "learning_rate": 0.0002057936842990969,
      "loss": 1.0007,
      "step": 1481
    },
    {
      "epoch": 4.26,
      "learning_rate": 0.00020516683646204837,
      "loss": 0.9961,
      "step": 1482
    },
    {
      "epoch": 4.26,
      "learning_rate": 0.00020454069824514442,
      "loss": 0.9702,
      "step": 1483
    },
    {
      "epoch": 4.26,
      "learning_rate": 0.00020391527115540777,
      "loss": 1.0197,
      "step": 1484
    },
    {
      "epoch": 4.27,
      "learning_rate": 0.00020329055669814934,
      "loss": 1.0332,
      "step": 1485
    },
    {
      "epoch": 4.27,
      "learning_rate": 0.00020266655637696547,
      "loss": 0.9505,
      "step": 1486
    },
    {
      "epoch": 4.27,
      "learning_rate": 0.000202043271693733,
      "loss": 1.028,
      "step": 1487
    },
    {
      "epoch": 4.28,
      "learning_rate": 0.00020142070414860702,
      "loss": 0.9979,
      "step": 1488
    },
    {
      "epoch": 4.28,
      "learning_rate": 0.00020079885524001585,
      "loss": 1.0094,
      "step": 1489
    },
    {
      "epoch": 4.28,
      "learning_rate": 0.00020017772646465875,
      "loss": 1.0179,
      "step": 1490
    },
    {
      "epoch": 4.28,
      "learning_rate": 0.00019955731931750183,
      "loss": 0.9858,
      "step": 1491
    },
    {
      "epoch": 4.29,
      "learning_rate": 0.00019893763529177329,
      "loss": 0.9672,
      "step": 1492
    },
    {
      "epoch": 4.29,
      "learning_rate": 0.00019831867587896218,
      "loss": 1.0064,
      "step": 1493
    },
    {
      "epoch": 4.29,
      "learning_rate": 0.00019770044256881258,
      "loss": 0.9741,
      "step": 1494
    },
    {
      "epoch": 4.3,
      "learning_rate": 0.00019708293684932165,
      "loss": 0.9994,
      "step": 1495
    },
    {
      "epoch": 4.3,
      "learning_rate": 0.00019646616020673475,
      "loss": 0.9838,
      "step": 1496
    },
    {
      "epoch": 4.3,
      "learning_rate": 0.00019585011412554272,
      "loss": 1.0428,
      "step": 1497
    },
    {
      "epoch": 4.3,
      "learning_rate": 0.00019523480008847854,
      "loss": 0.9872,
      "step": 1498
    },
    {
      "epoch": 4.31,
      "learning_rate": 0.00019462021957651206,
      "loss": 1.0393,
      "step": 1499
    },
    {
      "epoch": 4.31,
      "learning_rate": 0.00019400637406884874,
      "loss": 0.9901,
      "step": 1500
    },
    {
      "epoch": 4.31,
      "learning_rate": 0.00019339326504292444,
      "loss": 0.9529,
      "step": 1501
    },
    {
      "epoch": 4.32,
      "learning_rate": 0.00019278089397440268,
      "loss": 1.032,
      "step": 1502
    },
    {
      "epoch": 4.32,
      "learning_rate": 0.00019216926233717085,
      "loss": 1.0156,
      "step": 1503
    },
    {
      "epoch": 4.32,
      "learning_rate": 0.00019155837160333628,
      "loss": 0.9891,
      "step": 1504
    },
    {
      "epoch": 4.32,
      "learning_rate": 0.00019094822324322374,
      "loss": 0.9682,
      "step": 1505
    },
    {
      "epoch": 4.33,
      "learning_rate": 0.00019033881872537007,
      "loss": 1.0437,
      "step": 1506
    },
    {
      "epoch": 4.33,
      "learning_rate": 0.000189730159516523,
      "loss": 1.0728,
      "step": 1507
    },
    {
      "epoch": 4.33,
      "learning_rate": 0.00018912224708163562,
      "loss": 1.136,
      "step": 1508
    },
    {
      "epoch": 4.34,
      "learning_rate": 0.000188515082883864,
      "loss": 0.9912,
      "step": 1509
    },
    {
      "epoch": 4.34,
      "learning_rate": 0.0001879086683845635,
      "loss": 0.9792,
      "step": 1510
    },
    {
      "epoch": 4.34,
      "learning_rate": 0.00018730300504328436,
      "loss": 1.0441,
      "step": 1511
    },
    {
      "epoch": 4.34,
      "learning_rate": 0.0001866980943177699,
      "loss": 0.9867,
      "step": 1512
    },
    {
      "epoch": 4.35,
      "learning_rate": 0.00018609393766395082,
      "loss": 0.9496,
      "step": 1513
    },
    {
      "epoch": 4.35,
      "learning_rate": 0.00018549053653594373,
      "loss": 1.058,
      "step": 1514
    },
    {
      "epoch": 4.35,
      "learning_rate": 0.00018488789238604677,
      "loss": 1.0208,
      "step": 1515
    },
    {
      "epoch": 4.36,
      "learning_rate": 0.0001842860066647356,
      "loss": 1.053,
      "step": 1516
    },
    {
      "epoch": 4.36,
      "learning_rate": 0.0001836848808206612,
      "loss": 1.0398,
      "step": 1517
    },
    {
      "epoch": 4.36,
      "learning_rate": 0.0001830845163006448,
      "loss": 0.9846,
      "step": 1518
    },
    {
      "epoch": 4.36,
      "learning_rate": 0.00018248491454967607,
      "loss": 1.0947,
      "step": 1519
    },
    {
      "epoch": 4.37,
      "learning_rate": 0.00018188607701090826,
      "loss": 0.9599,
      "step": 1520
    },
    {
      "epoch": 4.37,
      "learning_rate": 0.00018128800512565513,
      "loss": 1.0515,
      "step": 1521
    },
    {
      "epoch": 4.37,
      "learning_rate": 0.0001806907003333884,
      "loss": 0.9748,
      "step": 1522
    },
    {
      "epoch": 4.38,
      "learning_rate": 0.00018009416407173257,
      "loss": 1.0615,
      "step": 1523
    },
    {
      "epoch": 4.38,
      "learning_rate": 0.00017949839777646327,
      "loss": 0.9797,
      "step": 1524
    },
    {
      "epoch": 4.38,
      "learning_rate": 0.00017890340288150214,
      "loss": 1.0182,
      "step": 1525
    },
    {
      "epoch": 4.39,
      "learning_rate": 0.00017830918081891488,
      "loss": 0.9973,
      "step": 1526
    },
    {
      "epoch": 4.39,
      "learning_rate": 0.00017771573301890664,
      "loss": 0.9715,
      "step": 1527
    },
    {
      "epoch": 4.39,
      "learning_rate": 0.00017712306090981894,
      "loss": 0.9903,
      "step": 1528
    },
    {
      "epoch": 4.39,
      "learning_rate": 0.0001765311659181269,
      "loss": 1.0584,
      "step": 1529
    },
    {
      "epoch": 4.4,
      "learning_rate": 0.00017594004946843456,
      "loss": 1.0168,
      "step": 1530
    },
    {
      "epoch": 4.4,
      "learning_rate": 0.00017534971298347274,
      "loss": 0.9991,
      "step": 1531
    },
    {
      "epoch": 4.4,
      "learning_rate": 0.00017476015788409438,
      "loss": 1.0644,
      "step": 1532
    },
    {
      "epoch": 4.41,
      "learning_rate": 0.00017417138558927244,
      "loss": 1.0257,
      "step": 1533
    },
    {
      "epoch": 4.41,
      "learning_rate": 0.0001735833975160952,
      "loss": 0.9382,
      "step": 1534
    },
    {
      "epoch": 4.41,
      "learning_rate": 0.00017299619507976343,
      "loss": 1.0251,
      "step": 1535
    },
    {
      "epoch": 4.41,
      "learning_rate": 0.00017240977969358758,
      "loss": 1.0315,
      "step": 1536
    },
    {
      "epoch": 4.42,
      "learning_rate": 0.00017182415276898307,
      "loss": 1.0939,
      "step": 1537
    },
    {
      "epoch": 4.42,
      "learning_rate": 0.00017123931571546826,
      "loss": 1.0648,
      "step": 1538
    },
    {
      "epoch": 4.42,
      "learning_rate": 0.00017065526994065972,
      "loss": 1.0663,
      "step": 1539
    },
    {
      "epoch": 4.43,
      "learning_rate": 0.0001700720168502703,
      "loss": 1.0047,
      "step": 1540
    },
    {
      "epoch": 4.43,
      "learning_rate": 0.00016948955784810437,
      "loss": 0.9987,
      "step": 1541
    },
    {
      "epoch": 4.43,
      "learning_rate": 0.00016890789433605507,
      "loss": 1.005,
      "step": 1542
    },
    {
      "epoch": 4.43,
      "learning_rate": 0.0001683270277141014,
      "loss": 0.984,
      "step": 1543
    },
    {
      "epoch": 4.44,
      "learning_rate": 0.00016774695938030377,
      "loss": 1.0267,
      "step": 1544
    },
    {
      "epoch": 4.44,
      "learning_rate": 0.0001671676907308018,
      "loss": 1.0053,
      "step": 1545
    },
    {
      "epoch": 4.44,
      "learning_rate": 0.00016658922315980972,
      "loss": 0.9758,
      "step": 1546
    },
    {
      "epoch": 4.45,
      "learning_rate": 0.00016601155805961437,
      "loss": 0.9388,
      "step": 1547
    },
    {
      "epoch": 4.45,
      "learning_rate": 0.00016543469682057105,
      "loss": 1.0032,
      "step": 1548
    },
    {
      "epoch": 4.45,
      "learning_rate": 0.00016485864083109946,
      "loss": 1.0916,
      "step": 1549
    },
    {
      "epoch": 4.45,
      "learning_rate": 0.0001642833914776823,
      "loss": 1.0647,
      "step": 1550
    },
    {
      "epoch": 4.46,
      "learning_rate": 0.00016370895014486,
      "loss": 1.0361,
      "step": 1551
    },
    {
      "epoch": 4.46,
      "learning_rate": 0.00016313531821522876,
      "loss": 0.9948,
      "step": 1552
    },
    {
      "epoch": 4.46,
      "learning_rate": 0.00016256249706943628,
      "loss": 1.0164,
      "step": 1553
    },
    {
      "epoch": 4.47,
      "learning_rate": 0.00016199048808617894,
      "loss": 1.0174,
      "step": 1554
    },
    {
      "epoch": 4.47,
      "learning_rate": 0.0001614192926421989,
      "loss": 0.9894,
      "step": 1555
    },
    {
      "epoch": 4.47,
      "learning_rate": 0.00016084891211227896,
      "loss": 1.0224,
      "step": 1556
    },
    {
      "epoch": 4.47,
      "learning_rate": 0.00016027934786924186,
      "loss": 1.0632,
      "step": 1557
    },
    {
      "epoch": 4.48,
      "learning_rate": 0.00015971060128394484,
      "loss": 0.9536,
      "step": 1558
    },
    {
      "epoch": 4.48,
      "learning_rate": 0.00015914267372527753,
      "loss": 0.9469,
      "step": 1559
    },
    {
      "epoch": 4.48,
      "learning_rate": 0.00015857556656015837,
      "loss": 0.9727,
      "step": 1560
    },
    {
      "epoch": 4.49,
      "learning_rate": 0.0001580092811535308,
      "loss": 0.9892,
      "step": 1561
    },
    {
      "epoch": 4.49,
      "learning_rate": 0.0001574438188683609,
      "loss": 1.0007,
      "step": 1562
    },
    {
      "epoch": 4.49,
      "learning_rate": 0.00015687918106563326,
      "loss": 1.0486,
      "step": 1563
    },
    {
      "epoch": 4.49,
      "learning_rate": 0.00015631536910434808,
      "loss": 0.985,
      "step": 1564
    },
    {
      "epoch": 4.5,
      "learning_rate": 0.00015575238434151822,
      "loss": 0.9887,
      "step": 1565
    },
    {
      "epoch": 4.5,
      "learning_rate": 0.0001551902281321651,
      "loss": 1.0012,
      "step": 1566
    },
    {
      "epoch": 4.5,
      "learning_rate": 0.00015462890182931644,
      "loss": 0.9717,
      "step": 1567
    },
    {
      "epoch": 4.51,
      "learning_rate": 0.00015406840678400203,
      "loss": 0.9807,
      "step": 1568
    },
    {
      "epoch": 4.51,
      "learning_rate": 0.0001535087443452514,
      "loss": 1.0023,
      "step": 1569
    },
    {
      "epoch": 4.51,
      "learning_rate": 0.00015294991586008977,
      "loss": 1.0454,
      "step": 1570
    },
    {
      "epoch": 4.51,
      "learning_rate": 0.00015239192267353508,
      "loss": 1.0108,
      "step": 1571
    },
    {
      "epoch": 4.52,
      "learning_rate": 0.00015183476612859538,
      "loss": 1.0085,
      "step": 1572
    },
    {
      "epoch": 4.52,
      "learning_rate": 0.00015127844756626435,
      "loss": 1.1009,
      "step": 1573
    },
    {
      "epoch": 4.52,
      "learning_rate": 0.0001507229683255194,
      "loss": 1.0869,
      "step": 1574
    },
    {
      "epoch": 4.53,
      "learning_rate": 0.00015016832974331724,
      "loss": 1.0008,
      "step": 1575
    },
    {
      "epoch": 4.53,
      "learning_rate": 0.00014961453315459183,
      "loss": 0.9719,
      "step": 1576
    },
    {
      "epoch": 4.53,
      "learning_rate": 0.00014906157989225005,
      "loss": 0.9815,
      "step": 1577
    },
    {
      "epoch": 4.53,
      "learning_rate": 0.00014850947128716913,
      "loss": 1.046,
      "step": 1578
    },
    {
      "epoch": 4.54,
      "learning_rate": 0.00014795820866819376,
      "loss": 0.9528,
      "step": 1579
    },
    {
      "epoch": 4.54,
      "learning_rate": 0.00014740779336213177,
      "loss": 1.0277,
      "step": 1580
    },
    {
      "epoch": 4.54,
      "learning_rate": 0.00014685822669375237,
      "loss": 0.9691,
      "step": 1581
    },
    {
      "epoch": 4.55,
      "learning_rate": 0.00014630950998578156,
      "loss": 0.9996,
      "step": 1582
    },
    {
      "epoch": 4.55,
      "learning_rate": 0.00014576164455890013,
      "loss": 0.9824,
      "step": 1583
    },
    {
      "epoch": 4.55,
      "learning_rate": 0.00014521463173173966,
      "loss": 1.089,
      "step": 1584
    },
    {
      "epoch": 4.55,
      "learning_rate": 0.0001446684728208795,
      "loss": 1.0654,
      "step": 1585
    },
    {
      "epoch": 4.56,
      "learning_rate": 0.00014412316914084439,
      "loss": 0.9783,
      "step": 1586
    },
    {
      "epoch": 4.56,
      "learning_rate": 0.00014357872200409987,
      "loss": 1.0007,
      "step": 1587
    },
    {
      "epoch": 4.56,
      "learning_rate": 0.00014303513272105056,
      "loss": 1.0584,
      "step": 1588
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.00014249240260003575,
      "loss": 0.988,
      "step": 1589
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.00014195053294732758,
      "loss": 0.9916,
      "step": 1590
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.0001414095250671265,
      "loss": 1.0115,
      "step": 1591
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.00014086938026155899,
      "loss": 0.9815,
      "step": 1592
    },
    {
      "epoch": 4.58,
      "learning_rate": 0.00014033009983067452,
      "loss": 1.0083,
      "step": 1593
    },
    {
      "epoch": 4.58,
      "learning_rate": 0.00013979168507244173,
      "loss": 0.9907,
      "step": 1594
    },
    {
      "epoch": 4.58,
      "learning_rate": 0.00013925413728274605,
      "loss": 0.941,
      "step": 1595
    },
    {
      "epoch": 4.59,
      "learning_rate": 0.00013871745775538596,
      "loss": 0.9842,
      "step": 1596
    },
    {
      "epoch": 4.59,
      "learning_rate": 0.00013818164778207039,
      "loss": 1.0284,
      "step": 1597
    },
    {
      "epoch": 4.59,
      "learning_rate": 0.0001376467086524156,
      "loss": 0.9454,
      "step": 1598
    },
    {
      "epoch": 4.59,
      "learning_rate": 0.0001371126416539409,
      "loss": 1.0691,
      "step": 1599
    },
    {
      "epoch": 4.6,
      "learning_rate": 0.00013657944807206763,
      "loss": 0.9755,
      "step": 1600
    },
    {
      "epoch": 4.6,
      "learning_rate": 0.00013604712919011426,
      "loss": 1.0724,
      "step": 1601
    },
    {
      "epoch": 4.6,
      "learning_rate": 0.00013551568628929433,
      "loss": 0.9301,
      "step": 1602
    },
    {
      "epoch": 4.61,
      "learning_rate": 0.0001349851206487127,
      "loss": 0.9956,
      "step": 1603
    },
    {
      "epoch": 4.61,
      "learning_rate": 0.00013445543354536317,
      "loss": 1.094,
      "step": 1604
    },
    {
      "epoch": 4.61,
      "learning_rate": 0.00013392662625412487,
      "loss": 0.975,
      "step": 1605
    },
    {
      "epoch": 4.61,
      "learning_rate": 0.00013339870004775928,
      "loss": 1.0161,
      "step": 1606
    },
    {
      "epoch": 4.62,
      "learning_rate": 0.00013287165619690723,
      "loss": 0.9074,
      "step": 1607
    },
    {
      "epoch": 4.62,
      "learning_rate": 0.00013234549597008571,
      "loss": 1.0114,
      "step": 1608
    },
    {
      "epoch": 4.62,
      "learning_rate": 0.00013182022063368536,
      "loss": 1.0525,
      "step": 1609
    },
    {
      "epoch": 4.63,
      "learning_rate": 0.000131295831451967,
      "loss": 0.9935,
      "step": 1610
    },
    {
      "epoch": 4.63,
      "learning_rate": 0.00013077232968705804,
      "loss": 0.9387,
      "step": 1611
    },
    {
      "epoch": 4.63,
      "learning_rate": 0.00013024971659895068,
      "loss": 1.0169,
      "step": 1612
    },
    {
      "epoch": 4.64,
      "learning_rate": 0.0001297279934454978,
      "loss": 1.0383,
      "step": 1613
    },
    {
      "epoch": 4.64,
      "learning_rate": 0.00012920716148241034,
      "loss": 1.0292,
      "step": 1614
    },
    {
      "epoch": 4.64,
      "learning_rate": 0.00012868722196325439,
      "loss": 0.9401,
      "step": 1615
    },
    {
      "epoch": 4.64,
      "learning_rate": 0.0001281681761394481,
      "loss": 1.0545,
      "step": 1616
    },
    {
      "epoch": 4.65,
      "learning_rate": 0.00012765002526025871,
      "loss": 1.0926,
      "step": 1617
    },
    {
      "epoch": 4.65,
      "learning_rate": 0.0001271327705727991,
      "loss": 1.0878,
      "step": 1618
    },
    {
      "epoch": 4.65,
      "learning_rate": 0.00012661641332202555,
      "loss": 0.9973,
      "step": 1619
    },
    {
      "epoch": 4.66,
      "learning_rate": 0.00012610095475073412,
      "loss": 1.029,
      "step": 1620
    },
    {
      "epoch": 4.66,
      "learning_rate": 0.0001255863960995578,
      "loss": 1.0018,
      "step": 1621
    },
    {
      "epoch": 4.66,
      "learning_rate": 0.0001250727386069639,
      "loss": 0.9682,
      "step": 1622
    },
    {
      "epoch": 4.66,
      "learning_rate": 0.0001245599835092504,
      "loss": 0.9893,
      "step": 1623
    },
    {
      "epoch": 4.67,
      "learning_rate": 0.00012404813204054384,
      "loss": 0.9796,
      "step": 1624
    },
    {
      "epoch": 4.67,
      "learning_rate": 0.00012353718543279518,
      "loss": 1.0412,
      "step": 1625
    },
    {
      "epoch": 4.67,
      "learning_rate": 0.00012302714491577832,
      "loss": 0.9894,
      "step": 1626
    },
    {
      "epoch": 4.68,
      "learning_rate": 0.0001225180117170857,
      "loss": 1.0269,
      "step": 1627
    },
    {
      "epoch": 4.68,
      "learning_rate": 0.00012200978706212606,
      "loss": 1.0356,
      "step": 1628
    },
    {
      "epoch": 4.68,
      "learning_rate": 0.00012150247217412185,
      "loss": 0.9937,
      "step": 1629
    },
    {
      "epoch": 4.68,
      "learning_rate": 0.00012099606827410519,
      "loss": 0.9775,
      "step": 1630
    },
    {
      "epoch": 4.69,
      "learning_rate": 0.00012049057658091628,
      "loss": 0.9847,
      "step": 1631
    },
    {
      "epoch": 4.69,
      "learning_rate": 0.00011998599831119911,
      "loss": 0.9692,
      "step": 1632
    },
    {
      "epoch": 4.69,
      "learning_rate": 0.00011948233467939979,
      "loss": 0.9249,
      "step": 1633
    },
    {
      "epoch": 4.7,
      "learning_rate": 0.00011897958689776255,
      "loss": 1.0091,
      "step": 1634
    },
    {
      "epoch": 4.7,
      "learning_rate": 0.00011847775617632744,
      "loss": 1.0236,
      "step": 1635
    },
    {
      "epoch": 4.7,
      "learning_rate": 0.00011797684372292761,
      "loss": 0.9895,
      "step": 1636
    },
    {
      "epoch": 4.7,
      "learning_rate": 0.00011747685074318548,
      "loss": 1.0237,
      "step": 1637
    },
    {
      "epoch": 4.71,
      "learning_rate": 0.00011697777844051105,
      "loss": 1.0812,
      "step": 1638
    },
    {
      "epoch": 4.71,
      "learning_rate": 0.00011647962801609785,
      "loss": 0.9853,
      "step": 1639
    },
    {
      "epoch": 4.71,
      "learning_rate": 0.00011598240066892102,
      "loss": 1.0434,
      "step": 1640
    },
    {
      "epoch": 4.72,
      "learning_rate": 0.00011548609759573375,
      "loss": 1.0248,
      "step": 1641
    },
    {
      "epoch": 4.72,
      "learning_rate": 0.00011499071999106447,
      "loss": 1.0446,
      "step": 1642
    },
    {
      "epoch": 4.72,
      "learning_rate": 0.00011449626904721472,
      "loss": 1.0608,
      "step": 1643
    },
    {
      "epoch": 4.72,
      "learning_rate": 0.00011400274595425497,
      "loss": 1.0188,
      "step": 1644
    },
    {
      "epoch": 4.73,
      "learning_rate": 0.00011351015190002317,
      "loss": 1.0176,
      "step": 1645
    },
    {
      "epoch": 4.73,
      "learning_rate": 0.00011301848807012061,
      "loss": 0.9966,
      "step": 1646
    },
    {
      "epoch": 4.73,
      "learning_rate": 0.00011252775564791024,
      "loss": 1.0143,
      "step": 1647
    },
    {
      "epoch": 4.74,
      "learning_rate": 0.00011203795581451288,
      "loss": 0.9225,
      "step": 1648
    },
    {
      "epoch": 4.74,
      "learning_rate": 0.00011154908974880456,
      "loss": 1.1062,
      "step": 1649
    },
    {
      "epoch": 4.74,
      "learning_rate": 0.00011106115862741456,
      "loss": 1.058,
      "step": 1650
    },
    {
      "epoch": 4.74,
      "learning_rate": 0.00011057416362472111,
      "loss": 1.011,
      "step": 1651
    },
    {
      "epoch": 4.75,
      "learning_rate": 0.00011008810591284996,
      "loss": 0.9559,
      "step": 1652
    },
    {
      "epoch": 4.75,
      "learning_rate": 0.0001096029866616704,
      "loss": 1.0504,
      "step": 1653
    },
    {
      "epoch": 4.75,
      "learning_rate": 0.0001091188070387934,
      "loss": 1.0179,
      "step": 1654
    },
    {
      "epoch": 4.76,
      "learning_rate": 0.00010863556820956838,
      "loss": 0.989,
      "step": 1655
    },
    {
      "epoch": 4.76,
      "learning_rate": 0.00010815327133708014,
      "loss": 1.0057,
      "step": 1656
    },
    {
      "epoch": 4.76,
      "learning_rate": 0.00010767191758214645,
      "loss": 1.0254,
      "step": 1657
    },
    {
      "epoch": 4.76,
      "learning_rate": 0.00010719150810331497,
      "loss": 1.0375,
      "step": 1658
    },
    {
      "epoch": 4.77,
      "learning_rate": 0.00010671204405686108,
      "loss": 0.9745,
      "step": 1659
    },
    {
      "epoch": 4.77,
      "learning_rate": 0.00010623352659678415,
      "loss": 1.0265,
      "step": 1660
    },
    {
      "epoch": 4.77,
      "learning_rate": 0.0001057559568748055,
      "loss": 1.0628,
      "step": 1661
    },
    {
      "epoch": 4.78,
      "learning_rate": 0.00010527933604036549,
      "loss": 1.0573,
      "step": 1662
    },
    {
      "epoch": 4.78,
      "learning_rate": 0.00010480366524062041,
      "loss": 0.9569,
      "step": 1663
    },
    {
      "epoch": 4.78,
      "learning_rate": 0.00010432894562044005,
      "loss": 1.0141,
      "step": 1664
    },
    {
      "epoch": 4.78,
      "learning_rate": 0.00010385517832240471,
      "loss": 0.9644,
      "step": 1665
    },
    {
      "epoch": 4.79,
      "learning_rate": 0.00010338236448680282,
      "loss": 1.0343,
      "step": 1666
    },
    {
      "epoch": 4.79,
      "learning_rate": 0.0001029105052516281,
      "loss": 0.9983,
      "step": 1667
    },
    {
      "epoch": 4.79,
      "learning_rate": 0.00010243960175257604,
      "loss": 0.9353,
      "step": 1668
    },
    {
      "epoch": 4.8,
      "learning_rate": 0.00010196965512304269,
      "loss": 1.0014,
      "step": 1669
    },
    {
      "epoch": 4.8,
      "learning_rate": 0.00010150066649412038,
      "loss": 1.0189,
      "step": 1670
    },
    {
      "epoch": 4.8,
      "learning_rate": 0.00010103263699459569,
      "loss": 0.9749,
      "step": 1671
    },
    {
      "epoch": 4.8,
      "learning_rate": 0.00010056556775094733,
      "loss": 1.02,
      "step": 1672
    },
    {
      "epoch": 4.81,
      "learning_rate": 0.00010009945988734204,
      "loss": 0.9624,
      "step": 1673
    },
    {
      "epoch": 4.81,
      "learning_rate": 9.963431452563332e-05,
      "loss": 0.9888,
      "step": 1674
    },
    {
      "epoch": 4.81,
      "learning_rate": 9.917013278535747e-05,
      "loss": 1.0222,
      "step": 1675
    },
    {
      "epoch": 4.82,
      "learning_rate": 9.870691578373215e-05,
      "loss": 1.0423,
      "step": 1676
    },
    {
      "epoch": 4.82,
      "learning_rate": 9.824466463565245e-05,
      "loss": 0.9822,
      "step": 1677
    },
    {
      "epoch": 4.82,
      "learning_rate": 9.7783380453689e-05,
      "loss": 1.0418,
      "step": 1678
    },
    {
      "epoch": 4.82,
      "learning_rate": 9.73230643480853e-05,
      "loss": 0.9372,
      "step": 1679
    },
    {
      "epoch": 4.83,
      "learning_rate": 9.686371742675442e-05,
      "loss": 0.96,
      "step": 1680
    },
    {
      "epoch": 4.83,
      "learning_rate": 9.640534079527718e-05,
      "loss": 1.0311,
      "step": 1681
    },
    {
      "epoch": 4.83,
      "learning_rate": 9.594793555689868e-05,
      "loss": 0.9328,
      "step": 1682
    },
    {
      "epoch": 4.84,
      "learning_rate": 9.549150281252633e-05,
      "loss": 0.998,
      "step": 1683
    },
    {
      "epoch": 4.84,
      "learning_rate": 9.503604366072666e-05,
      "loss": 1.0283,
      "step": 1684
    },
    {
      "epoch": 4.84,
      "learning_rate": 9.458155919772287e-05,
      "loss": 1.1128,
      "step": 1685
    },
    {
      "epoch": 4.84,
      "learning_rate": 9.412805051739265e-05,
      "loss": 0.9606,
      "step": 1686
    },
    {
      "epoch": 4.85,
      "learning_rate": 9.367551871126445e-05,
      "loss": 0.9828,
      "step": 1687
    },
    {
      "epoch": 4.85,
      "learning_rate": 9.322396486851626e-05,
      "loss": 1.083,
      "step": 1688
    },
    {
      "epoch": 4.85,
      "learning_rate": 9.277339007597158e-05,
      "loss": 0.9512,
      "step": 1689
    },
    {
      "epoch": 4.86,
      "learning_rate": 9.23237954180981e-05,
      "loss": 0.9335,
      "step": 1690
    },
    {
      "epoch": 4.86,
      "learning_rate": 9.187518197700395e-05,
      "loss": 1.0256,
      "step": 1691
    },
    {
      "epoch": 4.86,
      "learning_rate": 9.142755083243575e-05,
      "loss": 1.0245,
      "step": 1692
    },
    {
      "epoch": 4.86,
      "learning_rate": 9.098090306177625e-05,
      "loss": 0.9319,
      "step": 1693
    },
    {
      "epoch": 4.87,
      "learning_rate": 9.053523974004074e-05,
      "loss": 0.9335,
      "step": 1694
    },
    {
      "epoch": 4.87,
      "learning_rate": 9.00905619398757e-05,
      "loss": 0.9753,
      "step": 1695
    },
    {
      "epoch": 4.87,
      "learning_rate": 8.964687073155508e-05,
      "loss": 0.9848,
      "step": 1696
    },
    {
      "epoch": 4.88,
      "learning_rate": 8.920416718297875e-05,
      "loss": 1.0287,
      "step": 1697
    },
    {
      "epoch": 4.88,
      "learning_rate": 8.876245235966885e-05,
      "loss": 1.0156,
      "step": 1698
    },
    {
      "epoch": 4.88,
      "learning_rate": 8.83217273247684e-05,
      "loss": 0.9339,
      "step": 1699
    },
    {
      "epoch": 4.89,
      "learning_rate": 8.788199313903777e-05,
      "loss": 1.0041,
      "step": 1700
    },
    {
      "epoch": 4.89,
      "learning_rate": 8.744325086085247e-05,
      "loss": 0.9781,
      "step": 1701
    },
    {
      "epoch": 4.89,
      "learning_rate": 8.70055015462009e-05,
      "loss": 1.0373,
      "step": 1702
    },
    {
      "epoch": 4.89,
      "learning_rate": 8.656874624868133e-05,
      "loss": 1.0573,
      "step": 1703
    },
    {
      "epoch": 4.9,
      "learning_rate": 8.61329860194997e-05,
      "loss": 1.0222,
      "step": 1704
    },
    {
      "epoch": 4.9,
      "learning_rate": 8.569822190746684e-05,
      "loss": 0.8897,
      "step": 1705
    },
    {
      "epoch": 4.9,
      "learning_rate": 8.526445495899626e-05,
      "loss": 1.0194,
      "step": 1706
    },
    {
      "epoch": 4.91,
      "learning_rate": 8.483168621810133e-05,
      "loss": 1.0474,
      "step": 1707
    },
    {
      "epoch": 4.91,
      "learning_rate": 8.439991672639264e-05,
      "loss": 0.9225,
      "step": 1708
    },
    {
      "epoch": 4.91,
      "learning_rate": 8.396914752307632e-05,
      "loss": 0.9818,
      "step": 1709
    },
    {
      "epoch": 4.91,
      "learning_rate": 8.353937964495028e-05,
      "loss": 1.0363,
      "step": 1710
    },
    {
      "epoch": 4.92,
      "learning_rate": 8.311061412640286e-05,
      "loss": 0.9527,
      "step": 1711
    },
    {
      "epoch": 4.92,
      "learning_rate": 8.268285199940989e-05,
      "loss": 0.9912,
      "step": 1712
    },
    {
      "epoch": 4.92,
      "learning_rate": 8.225609429353187e-05,
      "loss": 0.9387,
      "step": 1713
    },
    {
      "epoch": 4.93,
      "learning_rate": 8.183034203591189e-05,
      "loss": 1.0988,
      "step": 1714
    },
    {
      "epoch": 4.93,
      "learning_rate": 8.140559625127308e-05,
      "loss": 0.9435,
      "step": 1715
    },
    {
      "epoch": 4.93,
      "learning_rate": 8.098185796191632e-05,
      "loss": 1.0873,
      "step": 1716
    },
    {
      "epoch": 4.93,
      "learning_rate": 8.05591281877176e-05,
      "loss": 0.9952,
      "step": 1717
    },
    {
      "epoch": 4.94,
      "learning_rate": 8.013740794612512e-05,
      "loss": 1.0077,
      "step": 1718
    },
    {
      "epoch": 4.94,
      "learning_rate": 7.971669825215788e-05,
      "loss": 1.0262,
      "step": 1719
    },
    {
      "epoch": 4.94,
      "learning_rate": 7.929700011840225e-05,
      "loss": 0.9791,
      "step": 1720
    },
    {
      "epoch": 4.95,
      "learning_rate": 7.887831455501e-05,
      "loss": 0.9691,
      "step": 1721
    },
    {
      "epoch": 4.95,
      "learning_rate": 7.846064256969571e-05,
      "loss": 1.0514,
      "step": 1722
    },
    {
      "epoch": 4.95,
      "learning_rate": 7.804398516773465e-05,
      "loss": 0.9704,
      "step": 1723
    },
    {
      "epoch": 4.95,
      "learning_rate": 7.762834335196011e-05,
      "loss": 1.0219,
      "step": 1724
    },
    {
      "epoch": 4.96,
      "learning_rate": 7.72137181227608e-05,
      "loss": 0.8953,
      "step": 1725
    },
    {
      "epoch": 4.96,
      "learning_rate": 7.680011047807894e-05,
      "loss": 0.9811,
      "step": 1726
    },
    {
      "epoch": 4.96,
      "learning_rate": 7.638752141340727e-05,
      "loss": 1.0684,
      "step": 1727
    },
    {
      "epoch": 4.97,
      "learning_rate": 7.597595192178702e-05,
      "loss": 0.9359,
      "step": 1728
    },
    {
      "epoch": 4.97,
      "learning_rate": 7.556540299380576e-05,
      "loss": 1.0265,
      "step": 1729
    },
    {
      "epoch": 4.97,
      "learning_rate": 7.51558756175943e-05,
      "loss": 0.9795,
      "step": 1730
    },
    {
      "epoch": 4.97,
      "learning_rate": 7.474737077882509e-05,
      "loss": 1.0276,
      "step": 1731
    },
    {
      "epoch": 4.98,
      "learning_rate": 7.433988946070913e-05,
      "loss": 0.9125,
      "step": 1732
    },
    {
      "epoch": 4.98,
      "learning_rate": 7.393343264399439e-05,
      "loss": 1.002,
      "step": 1733
    },
    {
      "epoch": 4.98,
      "learning_rate": 7.352800130696252e-05,
      "loss": 0.9894,
      "step": 1734
    },
    {
      "epoch": 4.99,
      "learning_rate": 7.312359642542726e-05,
      "loss": 1.0343,
      "step": 1735
    },
    {
      "epoch": 4.99,
      "learning_rate": 7.272021897273195e-05,
      "loss": 0.9613,
      "step": 1736
    },
    {
      "epoch": 4.99,
      "learning_rate": 7.23178699197467e-05,
      "loss": 0.9587,
      "step": 1737
    },
    {
      "epoch": 4.99,
      "learning_rate": 7.191655023486682e-05,
      "loss": 1.0677,
      "step": 1738
    },
    {
      "epoch": 5.0,
      "learning_rate": 7.151626088400964e-05,
      "loss": 0.9901,
      "step": 1739
    },
    {
      "epoch": 5.0,
      "learning_rate": 7.111700283061318e-05,
      "loss": 1.0125,
      "step": 1740
    },
    {
      "epoch": 5.0,
      "learning_rate": 7.071877703563279e-05,
      "loss": 0.9592,
      "step": 1741
    },
    {
      "epoch": 5.01,
      "learning_rate": 7.032158445753934e-05,
      "loss": 0.9172,
      "step": 1742
    },
    {
      "epoch": 5.01,
      "learning_rate": 6.992542605231739e-05,
      "loss": 0.9306,
      "step": 1743
    },
    {
      "epoch": 5.01,
      "learning_rate": 6.953030277346179e-05,
      "loss": 0.8882,
      "step": 1744
    },
    {
      "epoch": 5.01,
      "learning_rate": 6.913621557197647e-05,
      "loss": 0.9011,
      "step": 1745
    },
    {
      "epoch": 5.02,
      "learning_rate": 6.874316539637127e-05,
      "loss": 1.036,
      "step": 1746
    },
    {
      "epoch": 5.02,
      "learning_rate": 6.83511531926605e-05,
      "loss": 0.9104,
      "step": 1747
    },
    {
      "epoch": 5.02,
      "learning_rate": 6.796017990435977e-05,
      "loss": 0.9247,
      "step": 1748
    },
    {
      "epoch": 5.03,
      "learning_rate": 6.757024647248456e-05,
      "loss": 0.9631,
      "step": 1749
    },
    {
      "epoch": 5.03,
      "learning_rate": 6.718135383554735e-05,
      "loss": 0.9637,
      "step": 1750
    },
    {
      "epoch": 5.03,
      "learning_rate": 6.679350292955555e-05,
      "loss": 0.943,
      "step": 1751
    },
    {
      "epoch": 5.03,
      "learning_rate": 6.640669468800947e-05,
      "loss": 0.8809,
      "step": 1752
    },
    {
      "epoch": 5.04,
      "learning_rate": 6.602093004189964e-05,
      "loss": 0.8902,
      "step": 1753
    },
    {
      "epoch": 5.04,
      "learning_rate": 6.563620991970509e-05,
      "loss": 0.9265,
      "step": 1754
    },
    {
      "epoch": 5.04,
      "learning_rate": 6.52525352473905e-05,
      "loss": 0.943,
      "step": 1755
    },
    {
      "epoch": 5.05,
      "learning_rate": 6.486990694840466e-05,
      "loss": 0.8971,
      "step": 1756
    },
    {
      "epoch": 5.05,
      "learning_rate": 6.448832594367764e-05,
      "loss": 0.9573,
      "step": 1757
    },
    {
      "epoch": 5.05,
      "learning_rate": 6.410779315161885e-05,
      "loss": 0.9188,
      "step": 1758
    },
    {
      "epoch": 5.05,
      "learning_rate": 6.372830948811498e-05,
      "loss": 0.8975,
      "step": 1759
    },
    {
      "epoch": 5.06,
      "learning_rate": 6.33498758665273e-05,
      "loss": 0.8538,
      "step": 1760
    },
    {
      "epoch": 5.06,
      "learning_rate": 6.297249319769016e-05,
      "loss": 0.9732,
      "step": 1761
    },
    {
      "epoch": 5.06,
      "learning_rate": 6.259616238990828e-05,
      "loss": 0.9005,
      "step": 1762
    },
    {
      "epoch": 5.07,
      "learning_rate": 6.222088434895462e-05,
      "loss": 0.9439,
      "step": 1763
    },
    {
      "epoch": 5.07,
      "learning_rate": 6.184665997806832e-05,
      "loss": 0.816,
      "step": 1764
    },
    {
      "epoch": 5.07,
      "learning_rate": 6.147349017795235e-05,
      "loss": 0.8825,
      "step": 1765
    },
    {
      "epoch": 5.07,
      "learning_rate": 6.110137584677189e-05,
      "loss": 0.957,
      "step": 1766
    },
    {
      "epoch": 5.08,
      "learning_rate": 6.073031788015132e-05,
      "loss": 0.9855,
      "step": 1767
    },
    {
      "epoch": 5.08,
      "learning_rate": 6.0360317171172794e-05,
      "loss": 0.8586,
      "step": 1768
    },
    {
      "epoch": 5.08,
      "learning_rate": 5.99913746103738e-05,
      "loss": 0.9658,
      "step": 1769
    },
    {
      "epoch": 5.09,
      "learning_rate": 5.962349108574478e-05,
      "loss": 0.9692,
      "step": 1770
    },
    {
      "epoch": 5.09,
      "learning_rate": 5.92566674827274e-05,
      "loss": 0.9067,
      "step": 1771
    },
    {
      "epoch": 5.09,
      "learning_rate": 5.8890904684212154e-05,
      "loss": 0.9661,
      "step": 1772
    },
    {
      "epoch": 5.09,
      "learning_rate": 5.852620357053651e-05,
      "loss": 0.9571,
      "step": 1773
    },
    {
      "epoch": 5.1,
      "learning_rate": 5.816256501948258e-05,
      "loss": 0.8985,
      "step": 1774
    },
    {
      "epoch": 5.1,
      "learning_rate": 5.7799989906274776e-05,
      "loss": 0.8924,
      "step": 1775
    },
    {
      "epoch": 5.1,
      "learning_rate": 5.743847910357836e-05,
      "loss": 0.8983,
      "step": 1776
    },
    {
      "epoch": 5.11,
      "learning_rate": 5.7078033481496646e-05,
      "loss": 0.8736,
      "step": 1777
    },
    {
      "epoch": 5.11,
      "learning_rate": 5.671865390756947e-05,
      "loss": 0.9817,
      "step": 1778
    },
    {
      "epoch": 5.11,
      "learning_rate": 5.636034124677042e-05,
      "loss": 0.9348,
      "step": 1779
    },
    {
      "epoch": 5.11,
      "learning_rate": 5.600309636150574e-05,
      "loss": 0.9086,
      "step": 1780
    },
    {
      "epoch": 5.12,
      "learning_rate": 5.5646920111611375e-05,
      "loss": 0.8865,
      "step": 1781
    },
    {
      "epoch": 5.12,
      "learning_rate": 5.529181335435124e-05,
      "loss": 0.9429,
      "step": 1782
    },
    {
      "epoch": 5.12,
      "learning_rate": 5.493777694441521e-05,
      "loss": 0.9246,
      "step": 1783
    },
    {
      "epoch": 5.13,
      "learning_rate": 5.458481173391694e-05,
      "loss": 0.994,
      "step": 1784
    },
    {
      "epoch": 5.13,
      "learning_rate": 5.423291857239177e-05,
      "loss": 0.9531,
      "step": 1785
    },
    {
      "epoch": 5.13,
      "learning_rate": 5.388209830679508e-05,
      "loss": 0.8203,
      "step": 1786
    },
    {
      "epoch": 5.14,
      "learning_rate": 5.353235178149957e-05,
      "loss": 0.9035,
      "step": 1787
    },
    {
      "epoch": 5.14,
      "learning_rate": 5.318367983829392e-05,
      "loss": 0.9811,
      "step": 1788
    },
    {
      "epoch": 5.14,
      "learning_rate": 5.2836083316380176e-05,
      "loss": 0.9064,
      "step": 1789
    },
    {
      "epoch": 5.14,
      "learning_rate": 5.248956305237229e-05,
      "loss": 0.9664,
      "step": 1790
    },
    {
      "epoch": 5.15,
      "learning_rate": 5.214411988029355e-05,
      "loss": 0.9721,
      "step": 1791
    },
    {
      "epoch": 5.15,
      "learning_rate": 5.17997546315751e-05,
      "loss": 0.9666,
      "step": 1792
    },
    {
      "epoch": 5.15,
      "learning_rate": 5.145646813505339e-05,
      "loss": 0.9663,
      "step": 1793
    },
    {
      "epoch": 5.16,
      "learning_rate": 5.111426121696866e-05,
      "loss": 0.9163,
      "step": 1794
    },
    {
      "epoch": 5.16,
      "learning_rate": 5.0773134700962844e-05,
      "loss": 0.9148,
      "step": 1795
    },
    {
      "epoch": 5.16,
      "learning_rate": 5.043308940807717e-05,
      "loss": 0.9644,
      "step": 1796
    },
    {
      "epoch": 5.16,
      "learning_rate": 5.009412615675102e-05,
      "loss": 0.9519,
      "step": 1797
    },
    {
      "epoch": 5.17,
      "learning_rate": 4.97562457628189e-05,
      "loss": 0.94,
      "step": 1798
    },
    {
      "epoch": 5.17,
      "learning_rate": 4.9419449039509564e-05,
      "loss": 0.8746,
      "step": 1799
    },
    {
      "epoch": 5.17,
      "learning_rate": 4.908373679744316e-05,
      "loss": 0.9816,
      "step": 1800
    },
    {
      "epoch": 5.18,
      "learning_rate": 4.874910984462971e-05,
      "loss": 0.8395,
      "step": 1801
    },
    {
      "epoch": 5.18,
      "learning_rate": 4.841556898646732e-05,
      "loss": 0.9007,
      "step": 1802
    },
    {
      "epoch": 5.18,
      "learning_rate": 4.8083115025739754e-05,
      "loss": 0.8468,
      "step": 1803
    },
    {
      "epoch": 5.18,
      "learning_rate": 4.775174876261496e-05,
      "loss": 0.8686,
      "step": 1804
    },
    {
      "epoch": 5.19,
      "learning_rate": 4.7421470994642836e-05,
      "loss": 0.9584,
      "step": 1805
    },
    {
      "epoch": 5.19,
      "learning_rate": 4.709228251675357e-05,
      "loss": 0.8287,
      "step": 1806
    },
    {
      "epoch": 5.19,
      "learning_rate": 4.6764184121255434e-05,
      "loss": 0.8627,
      "step": 1807
    },
    {
      "epoch": 5.2,
      "learning_rate": 4.643717659783309e-05,
      "loss": 0.9579,
      "step": 1808
    },
    {
      "epoch": 5.2,
      "learning_rate": 4.6111260733545715e-05,
      "loss": 0.9129,
      "step": 1809
    },
    {
      "epoch": 5.2,
      "learning_rate": 4.578643731282483e-05,
      "loss": 0.9226,
      "step": 1810
    },
    {
      "epoch": 5.2,
      "learning_rate": 4.546270711747291e-05,
      "loss": 0.9996,
      "step": 1811
    },
    {
      "epoch": 5.21,
      "learning_rate": 4.514007092666084e-05,
      "loss": 0.9303,
      "step": 1812
    },
    {
      "epoch": 5.21,
      "learning_rate": 4.481852951692672e-05,
      "loss": 0.9814,
      "step": 1813
    },
    {
      "epoch": 5.21,
      "learning_rate": 4.4498083662173394e-05,
      "loss": 0.9982,
      "step": 1814
    },
    {
      "epoch": 5.22,
      "learning_rate": 4.417873413366702e-05,
      "loss": 0.9137,
      "step": 1815
    },
    {
      "epoch": 5.22,
      "learning_rate": 4.3860481700035095e-05,
      "loss": 0.8619,
      "step": 1816
    },
    {
      "epoch": 5.22,
      "learning_rate": 4.354332712726439e-05,
      "loss": 0.9781,
      "step": 1817
    },
    {
      "epoch": 5.22,
      "learning_rate": 4.322727117869951e-05,
      "loss": 0.9686,
      "step": 1818
    },
    {
      "epoch": 5.23,
      "learning_rate": 4.2912314615040805e-05,
      "loss": 0.8722,
      "step": 1819
    },
    {
      "epoch": 5.23,
      "learning_rate": 4.2598458194342325e-05,
      "loss": 0.9008,
      "step": 1820
    },
    {
      "epoch": 5.23,
      "learning_rate": 4.228570267201048e-05,
      "loss": 0.9027,
      "step": 1821
    },
    {
      "epoch": 5.24,
      "learning_rate": 4.1974048800801855e-05,
      "loss": 0.8969,
      "step": 1822
    },
    {
      "epoch": 5.24,
      "learning_rate": 4.166349733082153e-05,
      "loss": 1.0464,
      "step": 1823
    },
    {
      "epoch": 5.24,
      "learning_rate": 4.13540490095215e-05,
      "loss": 0.9685,
      "step": 1824
    },
    {
      "epoch": 5.24,
      "learning_rate": 4.104570458169826e-05,
      "loss": 0.9389,
      "step": 1825
    },
    {
      "epoch": 5.25,
      "learning_rate": 4.073846478949167e-05,
      "loss": 0.9697,
      "step": 1826
    },
    {
      "epoch": 5.25,
      "learning_rate": 4.043233037238281e-05,
      "loss": 0.9063,
      "step": 1827
    },
    {
      "epoch": 5.25,
      "learning_rate": 4.012730206719228e-05,
      "loss": 0.9392,
      "step": 1828
    },
    {
      "epoch": 5.26,
      "learning_rate": 3.982338060807833e-05,
      "loss": 0.8842,
      "step": 1829
    },
    {
      "epoch": 5.26,
      "learning_rate": 3.9520566726535364e-05,
      "loss": 0.9002,
      "step": 1830
    },
    {
      "epoch": 5.26,
      "learning_rate": 3.921886115139206e-05,
      "loss": 0.8977,
      "step": 1831
    },
    {
      "epoch": 5.26,
      "learning_rate": 3.8918264608809205e-05,
      "loss": 0.96,
      "step": 1832
    },
    {
      "epoch": 5.27,
      "learning_rate": 3.861877782227885e-05,
      "loss": 0.9282,
      "step": 1833
    },
    {
      "epoch": 5.27,
      "learning_rate": 3.83204015126215e-05,
      "loss": 1.0379,
      "step": 1834
    },
    {
      "epoch": 5.27,
      "learning_rate": 3.8023136397985156e-05,
      "loss": 0.9188,
      "step": 1835
    },
    {
      "epoch": 5.28,
      "learning_rate": 3.772698319384349e-05,
      "loss": 0.9222,
      "step": 1836
    },
    {
      "epoch": 5.28,
      "learning_rate": 3.7431942612993486e-05,
      "loss": 0.9122,
      "step": 1837
    },
    {
      "epoch": 5.28,
      "learning_rate": 3.713801536555483e-05,
      "loss": 0.9586,
      "step": 1838
    },
    {
      "epoch": 5.28,
      "learning_rate": 3.684520215896703e-05,
      "loss": 0.9466,
      "step": 1839
    },
    {
      "epoch": 5.29,
      "learning_rate": 3.65535036979886e-05,
      "loss": 0.9316,
      "step": 1840
    },
    {
      "epoch": 5.29,
      "learning_rate": 3.626292068469489e-05,
      "loss": 0.9265,
      "step": 1841
    },
    {
      "epoch": 5.29,
      "learning_rate": 3.5973453818476555e-05,
      "loss": 0.8604,
      "step": 1842
    },
    {
      "epoch": 5.3,
      "learning_rate": 3.56851037960379e-05,
      "loss": 0.9099,
      "step": 1843
    },
    {
      "epoch": 5.3,
      "learning_rate": 3.5397871311394935e-05,
      "loss": 0.9657,
      "step": 1844
    },
    {
      "epoch": 5.3,
      "learning_rate": 3.5111757055874326e-05,
      "loss": 0.8291,
      "step": 1845
    },
    {
      "epoch": 5.3,
      "learning_rate": 3.482676171811089e-05,
      "loss": 0.9505,
      "step": 1846
    },
    {
      "epoch": 5.31,
      "learning_rate": 3.45428859840467e-05,
      "loss": 0.953,
      "step": 1847
    },
    {
      "epoch": 5.31,
      "learning_rate": 3.426013053692878e-05,
      "loss": 0.8936,
      "step": 1848
    },
    {
      "epoch": 5.31,
      "learning_rate": 3.397849605730829e-05,
      "loss": 0.9272,
      "step": 1849
    },
    {
      "epoch": 5.32,
      "learning_rate": 3.369798322303785e-05,
      "loss": 0.9836,
      "step": 1850
    },
    {
      "epoch": 5.32,
      "learning_rate": 3.3418592709270665e-05,
      "loss": 0.942,
      "step": 1851
    },
    {
      "epoch": 5.32,
      "learning_rate": 3.314032518845889e-05,
      "loss": 0.9049,
      "step": 1852
    },
    {
      "epoch": 5.32,
      "learning_rate": 3.286318133035132e-05,
      "loss": 0.9732,
      "step": 1853
    },
    {
      "epoch": 5.33,
      "learning_rate": 3.258716180199278e-05,
      "loss": 0.8853,
      "step": 1854
    },
    {
      "epoch": 5.33,
      "learning_rate": 3.231226726772163e-05,
      "loss": 0.9191,
      "step": 1855
    },
    {
      "epoch": 5.33,
      "learning_rate": 3.203849838916872e-05,
      "loss": 0.8905,
      "step": 1856
    },
    {
      "epoch": 5.34,
      "learning_rate": 3.176585582525554e-05,
      "loss": 0.9807,
      "step": 1857
    },
    {
      "epoch": 5.34,
      "learning_rate": 3.1494340232192667e-05,
      "loss": 0.8947,
      "step": 1858
    },
    {
      "epoch": 5.34,
      "learning_rate": 3.122395226347835e-05,
      "loss": 0.928,
      "step": 1859
    },
    {
      "epoch": 5.34,
      "learning_rate": 3.095469256989658e-05,
      "loss": 0.9051,
      "step": 1860
    },
    {
      "epoch": 5.35,
      "learning_rate": 3.068656179951618e-05,
      "loss": 0.9327,
      "step": 1861
    },
    {
      "epoch": 5.35,
      "learning_rate": 3.04195605976883e-05,
      "loss": 0.9647,
      "step": 1862
    },
    {
      "epoch": 5.35,
      "learning_rate": 3.0153689607045842e-05,
      "loss": 0.9122,
      "step": 1863
    },
    {
      "epoch": 5.36,
      "learning_rate": 2.9888949467501148e-05,
      "loss": 0.951,
      "step": 1864
    },
    {
      "epoch": 5.36,
      "learning_rate": 2.9625340816244796e-05,
      "loss": 0.9223,
      "step": 1865
    },
    {
      "epoch": 5.36,
      "learning_rate": 2.9362864287744263e-05,
      "loss": 0.9684,
      "step": 1866
    },
    {
      "epoch": 5.36,
      "learning_rate": 2.910152051374193e-05,
      "loss": 0.8947,
      "step": 1867
    },
    {
      "epoch": 5.37,
      "learning_rate": 2.884131012325386e-05,
      "loss": 0.9641,
      "step": 1868
    },
    {
      "epoch": 5.37,
      "learning_rate": 2.858223374256841e-05,
      "loss": 0.9394,
      "step": 1869
    },
    {
      "epoch": 5.37,
      "learning_rate": 2.8324291995244334e-05,
      "loss": 0.897,
      "step": 1870
    },
    {
      "epoch": 5.38,
      "learning_rate": 2.806748550210958e-05,
      "loss": 0.9795,
      "step": 1871
    },
    {
      "epoch": 5.38,
      "learning_rate": 2.78118148812595e-05,
      "loss": 0.8935,
      "step": 1872
    },
    {
      "epoch": 5.38,
      "learning_rate": 2.7557280748055968e-05,
      "loss": 0.914,
      "step": 1873
    },
    {
      "epoch": 5.39,
      "learning_rate": 2.73038837151251e-05,
      "loss": 0.9209,
      "step": 1874
    },
    {
      "epoch": 5.39,
      "learning_rate": 2.705162439235648e-05,
      "loss": 0.9892,
      "step": 1875
    },
    {
      "epoch": 5.39,
      "learning_rate": 2.680050338690132e-05,
      "loss": 0.9247,
      "step": 1876
    },
    {
      "epoch": 5.39,
      "learning_rate": 2.6550521303170873e-05,
      "loss": 0.891,
      "step": 1877
    },
    {
      "epoch": 5.4,
      "learning_rate": 2.63016787428354e-05,
      "loss": 0.956,
      "step": 1878
    },
    {
      "epoch": 5.4,
      "learning_rate": 2.605397630482237e-05,
      "loss": 0.9543,
      "step": 1879
    },
    {
      "epoch": 5.4,
      "learning_rate": 2.580741458531516e-05,
      "loss": 0.9271,
      "step": 1880
    },
    {
      "epoch": 5.41,
      "learning_rate": 2.5561994177751735e-05,
      "loss": 1.0192,
      "step": 1881
    },
    {
      "epoch": 5.41,
      "learning_rate": 2.5317715672822926e-05,
      "loss": 0.9655,
      "step": 1882
    },
    {
      "epoch": 5.41,
      "learning_rate": 2.5074579658471265e-05,
      "loss": 0.9577,
      "step": 1883
    },
    {
      "epoch": 5.41,
      "learning_rate": 2.4832586719889417e-05,
      "loss": 0.9575,
      "step": 1884
    },
    {
      "epoch": 5.42,
      "learning_rate": 2.459173743951898e-05,
      "loss": 0.8566,
      "step": 1885
    },
    {
      "epoch": 5.42,
      "learning_rate": 2.4352032397048585e-05,
      "loss": 0.9135,
      "step": 1886
    },
    {
      "epoch": 5.42,
      "learning_rate": 2.4113472169413175e-05,
      "loss": 0.9523,
      "step": 1887
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.3876057330792345e-05,
      "loss": 0.9774,
      "step": 1888
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.3639788452608556e-05,
      "loss": 0.8849,
      "step": 1889
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.340466610352654e-05,
      "loss": 0.9411,
      "step": 1890
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.3170690849451004e-05,
      "loss": 0.9056,
      "step": 1891
    },
    {
      "epoch": 5.44,
      "learning_rate": 2.293786325352626e-05,
      "loss": 0.9487,
      "step": 1892
    },
    {
      "epoch": 5.44,
      "learning_rate": 2.2706183876134045e-05,
      "loss": 0.893,
      "step": 1893
    },
    {
      "epoch": 5.44,
      "learning_rate": 2.2475653274892594e-05,
      "loss": 0.877,
      "step": 1894
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.2246272004655345e-05,
      "loss": 0.9011,
      "step": 1895
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.201804061750917e-05,
      "loss": 0.9258,
      "step": 1896
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.1790959662773723e-05,
      "loss": 0.8806,
      "step": 1897
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.1565029686999303e-05,
      "loss": 0.9161,
      "step": 1898
    },
    {
      "epoch": 5.46,
      "learning_rate": 2.134025123396638e-05,
      "loss": 0.9444,
      "step": 1899
    },
    {
      "epoch": 5.46,
      "learning_rate": 2.111662484468363e-05,
      "loss": 0.9017,
      "step": 1900
    },
    {
      "epoch": 5.46,
      "learning_rate": 2.0894151057386845e-05,
      "loss": 0.927,
      "step": 1901
    },
    {
      "epoch": 5.47,
      "learning_rate": 2.0672830407537924e-05,
      "loss": 1.0153,
      "step": 1902
    },
    {
      "epoch": 5.47,
      "learning_rate": 2.0452663427823094e-05,
      "loss": 0.9957,
      "step": 1903
    },
    {
      "epoch": 5.47,
      "learning_rate": 2.0233650648152026e-05,
      "loss": 0.9319,
      "step": 1904
    },
    {
      "epoch": 5.47,
      "learning_rate": 2.0015792595656222e-05,
      "loss": 1.0197,
      "step": 1905
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.9799089794688196e-05,
      "loss": 0.9034,
      "step": 1906
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.9583542766819616e-05,
      "loss": 0.9518,
      "step": 1907
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.9369152030840554e-05,
      "loss": 0.9657,
      "step": 1908
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.9155918102758185e-05,
      "loss": 0.9141,
      "step": 1909
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.8943841495795088e-05,
      "loss": 0.8918,
      "step": 1910
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.873292272038868e-05,
      "loss": 0.9622,
      "step": 1911
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.8523162284189378e-05,
      "loss": 0.9359,
      "step": 1912
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.8314560692059833e-05,
      "loss": 0.9652,
      "step": 1913
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.810711844607349e-05,
      "loss": 0.9178,
      "step": 1914
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.790083604551318e-05,
      "loss": 0.9333,
      "step": 1915
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.769571398687059e-05,
      "loss": 0.9035,
      "step": 1916
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.7491752763844293e-05,
      "loss": 0.9735,
      "step": 1917
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.728895286733906e-05,
      "loss": 0.9328,
      "step": 1918
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.7087314785464382e-05,
      "loss": 0.95,
      "step": 1919
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.688683900353366e-05,
      "loss": 0.9434,
      "step": 1920
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.668752600406259e-05,
      "loss": 0.9374,
      "step": 1921
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.648937626676822e-05,
      "loss": 0.9725,
      "step": 1922
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.6292390268568102e-05,
      "loss": 0.8637,
      "step": 1923
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.6096568483578377e-05,
      "loss": 0.9359,
      "step": 1924
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.590191138311342e-05,
      "loss": 0.8813,
      "step": 1925
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.5708419435684463e-05,
      "loss": 0.8906,
      "step": 1926
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.5516093106997974e-05,
      "loss": 0.8752,
      "step": 1927
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.5324932859955398e-05,
      "loss": 0.9593,
      "step": 1928
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.5134939154651196e-05,
      "loss": 0.9663,
      "step": 1929
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.4946112448372463e-05,
      "loss": 0.8674,
      "step": 1930
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.4758453195597266e-05,
      "loss": 0.9207,
      "step": 1931
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.4571961847993976e-05,
      "loss": 0.943,
      "step": 1932
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.438663885441982e-05,
      "loss": 0.9549,
      "step": 1933
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.4202484660920057e-05,
      "loss": 0.9355,
      "step": 1934
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.4019499710726914e-05,
      "loss": 0.8933,
      "step": 1935
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.3837684444258092e-05,
      "loss": 0.9496,
      "step": 1936
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.365703929911638e-05,
      "loss": 0.907,
      "step": 1937
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.3477564710088097e-05,
      "loss": 0.9374,
      "step": 1938
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.3299261109142203e-05,
      "loss": 0.9447,
      "step": 1939
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.3122128925429356e-05,
      "loss": 0.8776,
      "step": 1940
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.2946168585280638e-05,
      "loss": 0.9783,
      "step": 1941
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.2771380512206888e-05,
      "loss": 0.9505,
      "step": 1942
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.2597765126897198e-05,
      "loss": 0.9081,
      "step": 1943
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.2425322847218367e-05,
      "loss": 0.9344,
      "step": 1944
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.2254054088213729e-05,
      "loss": 0.9174,
      "step": 1945
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.2083959262101873e-05,
      "loss": 0.9293,
      "step": 1946
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.191503877827621e-05,
      "loss": 0.9089,
      "step": 1947
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.174729304330352e-05,
      "loss": 0.8835,
      "step": 1948
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.1580722460923176e-05,
      "loss": 0.8714,
      "step": 1949
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.141532743204604e-05,
      "loss": 0.9438,
      "step": 1950
    },
    {
      "epoch": 5.61,
      "learning_rate": 1.1251108354753736e-05,
      "loss": 1.0417,
      "step": 1951
    },
    {
      "epoch": 5.61,
      "learning_rate": 1.1088065624297483e-05,
      "loss": 0.9736,
      "step": 1952
    },
    {
      "epoch": 5.61,
      "learning_rate": 1.0926199633097156e-05,
      "loss": 0.8482,
      "step": 1953
    },
    {
      "epoch": 5.61,
      "learning_rate": 1.0765510770740505e-05,
      "loss": 0.8821,
      "step": 1954
    },
    {
      "epoch": 5.62,
      "learning_rate": 1.0605999423981937e-05,
      "loss": 0.91,
      "step": 1955
    },
    {
      "epoch": 5.62,
      "learning_rate": 1.044766597674196e-05,
      "loss": 0.937,
      "step": 1956
    },
    {
      "epoch": 5.62,
      "learning_rate": 1.0290510810105846e-05,
      "loss": 0.8726,
      "step": 1957
    },
    {
      "epoch": 5.63,
      "learning_rate": 1.0134534302323029e-05,
      "loss": 0.8746,
      "step": 1958
    },
    {
      "epoch": 5.63,
      "learning_rate": 9.979736828806096e-06,
      "loss": 1.0505,
      "step": 1959
    },
    {
      "epoch": 5.63,
      "learning_rate": 9.826118762129798e-06,
      "loss": 0.941,
      "step": 1960
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.673680472030322e-06,
      "loss": 0.8703,
      "step": 1961
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.522422325404235e-06,
      "loss": 1.0024,
      "step": 1962
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.372344686307655e-06,
      "loss": 0.9037,
      "step": 1963
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.22344791595553e-06,
      "loss": 0.8514,
      "step": 1964
    },
    {
      "epoch": 5.65,
      "learning_rate": 9.075732372720414e-06,
      "loss": 0.9786,
      "step": 1965
    },
    {
      "epoch": 5.65,
      "learning_rate": 8.929198412131967e-06,
      "loss": 0.9767,
      "step": 1966
    },
    {
      "epoch": 5.65,
      "learning_rate": 8.783846386875959e-06,
      "loss": 0.9577,
      "step": 1967
    },
    {
      "epoch": 5.66,
      "learning_rate": 8.639676646793382e-06,
      "loss": 0.868,
      "step": 1968
    },
    {
      "epoch": 5.66,
      "learning_rate": 8.496689538879553e-06,
      "loss": 1.0617,
      "step": 1969
    },
    {
      "epoch": 5.66,
      "learning_rate": 8.354885407283574e-06,
      "loss": 0.8568,
      "step": 1970
    },
    {
      "epoch": 5.66,
      "learning_rate": 8.214264593307098e-06,
      "loss": 0.9819,
      "step": 1971
    },
    {
      "epoch": 5.67,
      "learning_rate": 8.07482743540372e-06,
      "loss": 0.9332,
      "step": 1972
    },
    {
      "epoch": 5.67,
      "learning_rate": 7.936574269178377e-06,
      "loss": 0.8415,
      "step": 1973
    },
    {
      "epoch": 5.67,
      "learning_rate": 7.799505427386e-06,
      "loss": 0.8843,
      "step": 1974
    },
    {
      "epoch": 5.68,
      "learning_rate": 7.66362123993125e-06,
      "loss": 0.9982,
      "step": 1975
    },
    {
      "epoch": 5.68,
      "learning_rate": 7.528922033867347e-06,
      "loss": 0.9647,
      "step": 1976
    },
    {
      "epoch": 5.68,
      "learning_rate": 7.3954081333955095e-06,
      "loss": 0.9728,
      "step": 1977
    },
    {
      "epoch": 5.68,
      "learning_rate": 7.263079859864297e-06,
      "loss": 0.8579,
      "step": 1978
    },
    {
      "epoch": 5.69,
      "learning_rate": 7.1319375317681625e-06,
      "loss": 0.9804,
      "step": 1979
    },
    {
      "epoch": 5.69,
      "learning_rate": 7.001981464747565e-06,
      "loss": 0.8655,
      "step": 1980
    },
    {
      "epoch": 5.69,
      "learning_rate": 6.873211971587634e-06,
      "loss": 0.8752,
      "step": 1981
    },
    {
      "epoch": 5.7,
      "learning_rate": 6.745629362217731e-06,
      "loss": 0.9783,
      "step": 1982
    },
    {
      "epoch": 5.7,
      "learning_rate": 6.61923394371039e-06,
      "loss": 0.8415,
      "step": 1983
    },
    {
      "epoch": 5.7,
      "learning_rate": 6.494026020280874e-06,
      "loss": 0.9615,
      "step": 1984
    },
    {
      "epoch": 5.7,
      "learning_rate": 6.3700058932862905e-06,
      "loss": 0.8892,
      "step": 1985
    },
    {
      "epoch": 5.71,
      "learning_rate": 6.2471738612247534e-06,
      "loss": 0.8774,
      "step": 1986
    },
    {
      "epoch": 5.71,
      "learning_rate": 6.125530219734998e-06,
      "loss": 0.934,
      "step": 1987
    },
    {
      "epoch": 5.71,
      "learning_rate": 6.005075261595494e-06,
      "loss": 0.9642,
      "step": 1988
    },
    {
      "epoch": 5.72,
      "learning_rate": 5.885809276723608e-06,
      "loss": 0.9142,
      "step": 1989
    },
    {
      "epoch": 5.72,
      "learning_rate": 5.767732552174998e-06,
      "loss": 0.9645,
      "step": 1990
    },
    {
      "epoch": 5.72,
      "learning_rate": 5.650845372142999e-06,
      "loss": 0.9398,
      "step": 1991
    },
    {
      "epoch": 5.72,
      "learning_rate": 5.5351480179580135e-06,
      "loss": 0.9265,
      "step": 1992
    },
    {
      "epoch": 5.73,
      "learning_rate": 5.42064076808646e-06,
      "loss": 0.9346,
      "step": 1993
    },
    {
      "epoch": 5.73,
      "learning_rate": 5.307323898130545e-06,
      "loss": 0.9319,
      "step": 1994
    },
    {
      "epoch": 5.73,
      "learning_rate": 5.195197680827379e-06,
      "loss": 0.8801,
      "step": 1995
    },
    {
      "epoch": 5.74,
      "learning_rate": 5.0842623860482e-06,
      "loss": 0.9238,
      "step": 1996
    },
    {
      "epoch": 5.74,
      "learning_rate": 4.974518280798035e-06,
      "loss": 0.921,
      "step": 1997
    },
    {
      "epoch": 5.74,
      "learning_rate": 4.865965629214819e-06,
      "loss": 0.9408,
      "step": 1998
    },
    {
      "epoch": 5.74,
      "learning_rate": 4.758604692568946e-06,
      "loss": 0.918,
      "step": 1999
    },
    {
      "epoch": 5.75,
      "learning_rate": 4.652435729262272e-06,
      "loss": 1.0032,
      "step": 2000
    },
    {
      "epoch": 5.75,
      "learning_rate": 4.547458994828002e-06,
      "loss": 0.9377,
      "step": 2001
    },
    {
      "epoch": 5.75,
      "learning_rate": 4.443674741929693e-06,
      "loss": 0.8998,
      "step": 2002
    },
    {
      "epoch": 5.76,
      "learning_rate": 4.341083220360864e-06,
      "loss": 0.9254,
      "step": 2003
    },
    {
      "epoch": 5.76,
      "learning_rate": 4.239684677044165e-06,
      "loss": 0.8811,
      "step": 2004
    },
    {
      "epoch": 5.76,
      "learning_rate": 4.1394793560310395e-06,
      "loss": 0.9301,
      "step": 2005
    },
    {
      "epoch": 5.76,
      "learning_rate": 4.04046749850101e-06,
      "loss": 0.891,
      "step": 2006
    },
    {
      "epoch": 5.77,
      "learning_rate": 3.942649342761117e-06,
      "loss": 0.9762,
      "step": 2007
    },
    {
      "epoch": 5.77,
      "learning_rate": 3.846025124245145e-06,
      "loss": 1.0246,
      "step": 2008
    },
    {
      "epoch": 5.77,
      "learning_rate": 3.750595075513563e-06,
      "loss": 0.9927,
      "step": 2009
    },
    {
      "epoch": 5.78,
      "learning_rate": 3.6563594262524203e-06,
      "loss": 0.8961,
      "step": 2010
    },
    {
      "epoch": 5.78,
      "learning_rate": 3.563318403273119e-06,
      "loss": 0.9908,
      "step": 2011
    },
    {
      "epoch": 5.78,
      "learning_rate": 3.471472230511752e-06,
      "loss": 0.8572,
      "step": 2012
    },
    {
      "epoch": 5.78,
      "learning_rate": 3.3808211290284885e-06,
      "loss": 0.9405,
      "step": 2013
    },
    {
      "epoch": 5.79,
      "learning_rate": 3.2913653170073554e-06,
      "loss": 0.9336,
      "step": 2014
    },
    {
      "epoch": 5.79,
      "learning_rate": 3.203105009755236e-06,
      "loss": 0.96,
      "step": 2015
    },
    {
      "epoch": 5.79,
      "learning_rate": 3.1160404197018156e-06,
      "loss": 0.9799,
      "step": 2016
    },
    {
      "epoch": 5.8,
      "learning_rate": 3.0301717563987474e-06,
      "loss": 0.8705,
      "step": 2017
    },
    {
      "epoch": 5.8,
      "learning_rate": 2.9454992265193214e-06,
      "loss": 0.9011,
      "step": 2018
    },
    {
      "epoch": 5.8,
      "learning_rate": 2.8620230338578524e-06,
      "loss": 0.9446,
      "step": 2019
    },
    {
      "epoch": 5.8,
      "learning_rate": 2.7797433793292914e-06,
      "loss": 0.9422,
      "step": 2020
    },
    {
      "epoch": 5.81,
      "learning_rate": 2.6986604609687273e-06,
      "loss": 0.8927,
      "step": 2021
    },
    {
      "epoch": 5.81,
      "learning_rate": 2.6187744739308294e-06,
      "loss": 0.9249,
      "step": 2022
    },
    {
      "epoch": 5.81,
      "learning_rate": 2.5400856104894065e-06,
      "loss": 0.9044,
      "step": 2023
    },
    {
      "epoch": 5.82,
      "learning_rate": 2.4625940600369603e-06,
      "loss": 0.9725,
      "step": 2024
    },
    {
      "epoch": 5.82,
      "learning_rate": 2.386300009084408e-06,
      "loss": 0.9514,
      "step": 2025
    },
    {
      "epoch": 5.82,
      "learning_rate": 2.3112036412602513e-06,
      "loss": 0.914,
      "step": 2026
    },
    {
      "epoch": 5.82,
      "learning_rate": 2.237305137310408e-06,
      "loss": 0.9681,
      "step": 2027
    },
    {
      "epoch": 5.83,
      "learning_rate": 2.1646046750978256e-06,
      "loss": 0.9539,
      "step": 2028
    },
    {
      "epoch": 5.83,
      "learning_rate": 2.093102429601701e-06,
      "loss": 0.9244,
      "step": 2029
    },
    {
      "epoch": 5.83,
      "learning_rate": 2.022798572917539e-06,
      "loss": 0.9184,
      "step": 2030
    },
    {
      "epoch": 5.84,
      "learning_rate": 1.953693274256374e-06,
      "loss": 0.929,
      "step": 2031
    },
    {
      "epoch": 5.84,
      "learning_rate": 1.8857866999444916e-06,
      "loss": 0.9518,
      "step": 2032
    },
    {
      "epoch": 5.84,
      "learning_rate": 1.8190790134231528e-06,
      "loss": 0.9528,
      "step": 2033
    },
    {
      "epoch": 5.84,
      "learning_rate": 1.753570375247815e-06,
      "loss": 0.9379,
      "step": 2034
    },
    {
      "epoch": 5.85,
      "learning_rate": 1.6892609430883e-06,
      "loss": 0.8915,
      "step": 2035
    },
    {
      "epoch": 5.85,
      "learning_rate": 1.6261508717278495e-06,
      "loss": 0.8783,
      "step": 2036
    },
    {
      "epoch": 5.85,
      "learning_rate": 1.5642403130632365e-06,
      "loss": 0.9547,
      "step": 2037
    },
    {
      "epoch": 5.86,
      "learning_rate": 1.503529416103988e-06,
      "loss": 0.9599,
      "step": 2038
    },
    {
      "epoch": 5.86,
      "learning_rate": 1.444018326972385e-06,
      "loss": 0.9067,
      "step": 2039
    },
    {
      "epoch": 5.86,
      "learning_rate": 1.3857071889029072e-06,
      "loss": 0.9541,
      "step": 2040
    },
    {
      "epoch": 5.86,
      "learning_rate": 1.3285961422417892e-06,
      "loss": 0.9173,
      "step": 2041
    },
    {
      "epoch": 5.87,
      "learning_rate": 1.2726853244471316e-06,
      "loss": 0.8999,
      "step": 2042
    },
    {
      "epoch": 5.87,
      "learning_rate": 1.2179748700879012e-06,
      "loss": 0.9359,
      "step": 2043
    },
    {
      "epoch": 5.87,
      "learning_rate": 1.164464910844154e-06,
      "loss": 1.0013,
      "step": 2044
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.1121555755065348e-06,
      "loss": 0.9292,
      "step": 2045
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.061046989976e-06,
      "loss": 0.856,
      "step": 2046
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.011139277263262e-06,
      "loss": 0.9083,
      "step": 2047
    },
    {
      "epoch": 5.89,
      "learning_rate": 9.624325574890126e-07,
      "loss": 0.9101,
      "step": 2048
    },
    {
      "epoch": 5.89,
      "learning_rate": 9.149269478830879e-07,
      "loss": 0.9492,
      "step": 2049
    },
    {
      "epoch": 5.89,
      "learning_rate": 8.686225627845268e-07,
      "loss": 0.9283,
      "step": 2050
    },
    {
      "epoch": 5.89,
      "learning_rate": 8.235195136411799e-07,
      "loss": 0.9059,
      "step": 2051
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.796179090094891e-07,
      "loss": 0.9395,
      "step": 2052
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.369178545542087e-07,
      "loss": 0.8922,
      "step": 2053
    },
    {
      "epoch": 5.9,
      "learning_rate": 6.954194530480185e-07,
      "loss": 0.8971,
      "step": 2054
    },
    {
      "epoch": 5.91,
      "learning_rate": 6.551228043715218e-07,
      "loss": 0.9824,
      "step": 2055
    },
    {
      "epoch": 5.91,
      "learning_rate": 6.160280055128031e-07,
      "loss": 0.9088,
      "step": 2056
    },
    {
      "epoch": 5.91,
      "learning_rate": 5.781351505673715e-07,
      "loss": 0.9579,
      "step": 2057
    },
    {
      "epoch": 5.91,
      "learning_rate": 5.41444330737717e-07,
      "loss": 0.9403,
      "step": 2058
    },
    {
      "epoch": 5.92,
      "learning_rate": 5.059556343333105e-07,
      "loss": 0.9674,
      "step": 2059
    },
    {
      "epoch": 5.92,
      "learning_rate": 4.716691467701595e-07,
      "loss": 0.9178,
      "step": 2060
    },
    {
      "epoch": 5.92,
      "learning_rate": 4.3858495057080837e-07,
      "loss": 0.9729,
      "step": 2061
    },
    {
      "epoch": 5.93,
      "learning_rate": 4.0670312536411624e-07,
      "loss": 0.9259,
      "step": 2062
    },
    {
      "epoch": 5.93,
      "learning_rate": 3.7602374788497927e-07,
      "loss": 0.9497,
      "step": 2063
    },
    {
      "epoch": 5.93,
      "learning_rate": 3.465468919740533e-07,
      "loss": 0.8808,
      "step": 2064
    },
    {
      "epoch": 5.93,
      "learning_rate": 3.182726285778648e-07,
      "loss": 0.9894,
      "step": 2065
    },
    {
      "epoch": 5.94,
      "learning_rate": 2.912010257484221e-07,
      "loss": 0.8658,
      "step": 2066
    },
    {
      "epoch": 5.94,
      "learning_rate": 2.6533214864310486e-07,
      "loss": 0.9026,
      "step": 2067
    },
    {
      "epoch": 5.94,
      "learning_rate": 2.4066605952444145e-07,
      "loss": 0.9133,
      "step": 2068
    },
    {
      "epoch": 5.95,
      "learning_rate": 2.172028177601648e-07,
      "loss": 0.8918,
      "step": 2069
    },
    {
      "epoch": 5.95,
      "learning_rate": 1.9494247982282386e-07,
      "loss": 0.9019,
      "step": 2070
    },
    {
      "epoch": 5.95,
      "learning_rate": 1.7388509928978335e-07,
      "loss": 0.9045,
      "step": 2071
    },
    {
      "epoch": 5.95,
      "learning_rate": 1.540307268430019e-07,
      "loss": 0.9457,
      "step": 2072
    },
    {
      "epoch": 5.96,
      "learning_rate": 1.3537941026914301e-07,
      "loss": 0.9536,
      "step": 2073
    },
    {
      "epoch": 5.96,
      "learning_rate": 1.1793119445918654e-07,
      "loss": 1.0099,
      "step": 2074
    },
    {
      "epoch": 5.96,
      "learning_rate": 1.0168612140831757e-07,
      "loss": 0.9,
      "step": 2075
    },
    {
      "epoch": 5.97,
      "learning_rate": 8.664423021614853e-08,
      "loss": 0.94,
      "step": 2076
    },
    {
      "epoch": 5.97,
      "learning_rate": 7.280555708627511e-08,
      "loss": 0.8817,
      "step": 2077
    },
    {
      "epoch": 5.97,
      "learning_rate": 6.017013532627625e-08,
      "loss": 0.8885,
      "step": 2078
    },
    {
      "epoch": 5.97,
      "learning_rate": 4.873799534788059e-08,
      "loss": 0.8433,
      "step": 2079
    },
    {
      "epoch": 5.98,
      "learning_rate": 3.850916466652255e-08,
      "loss": 1.0004,
      "step": 2080
    },
    {
      "epoch": 5.98,
      "learning_rate": 2.948366790145318e-08,
      "loss": 0.9297,
      "step": 2081
    },
    {
      "epoch": 5.98,
      "learning_rate": 2.1661526775795802e-08,
      "loss": 1.0617,
      "step": 2082
    },
    {
      "epoch": 5.99,
      "learning_rate": 1.504276011621286e-08,
      "loss": 0.8841,
      "step": 2083
    },
    {
      "epoch": 5.99,
      "learning_rate": 9.627383853128002e-09,
      "loss": 0.8912,
      "step": 2084
    },
    {
      "epoch": 5.99,
      "learning_rate": 5.415411020615046e-09,
      "loss": 0.9817,
      "step": 2085
    },
    {
      "epoch": 5.99,
      "learning_rate": 2.406851756231454e-09,
      "loss": 0.939,
      "step": 2086
    },
    {
      "epoch": 6.0,
      "learning_rate": 6.017133011293474e-10,
      "loss": 1.0128,
      "step": 2087
    },
    {
      "epoch": 6.0,
      "learning_rate": 0.0,
      "loss": 0.8203,
      "step": 2088
    },
    {
      "epoch": 6.0,
      "step": 2088,
      "total_flos": 6.471549767129235e+18,
      "train_loss": 1.2130510043138745,
      "train_runtime": 15384.7999,
      "train_samples_per_second": 17.345,
      "train_steps_per_second": 0.136
    }
  ],
  "max_steps": 2088,
  "num_train_epochs": 6,
  "total_flos": 6.471549767129235e+18,
  "trial_name": null,
  "trial_params": null
}