{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9965217391304346,
  "global_step": 574,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 2.7777777777777776e-05,
      "loss": 1.7259,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.555555555555555e-05,
      "loss": 1.7375,
      "step": 2
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.333333333333333e-05,
      "loss": 1.6292,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001111111111111111,
      "loss": 1.5607,
      "step": 4
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001388888888888889,
      "loss": 1.4619,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00016666666666666666,
      "loss": 1.4399,
      "step": 6
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019444444444444446,
      "loss": 1.4174,
      "step": 7
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002222222222222222,
      "loss": 1.373,
      "step": 8
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00025,
      "loss": 1.3324,
      "step": 9
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0002777777777777778,
      "loss": 1.3099,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003055555555555556,
      "loss": 1.3236,
      "step": 11
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003333333333333333,
      "loss": 1.2759,
      "step": 12
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003611111111111111,
      "loss": 1.2616,
      "step": 13
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0003888888888888889,
      "loss": 1.2344,
      "step": 14
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0004166666666666667,
      "loss": 1.2341,
      "step": 15
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0004444444444444444,
      "loss": 1.2406,
      "step": 16
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00047222222222222224,
      "loss": 1.1984,
      "step": 17
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0005,
      "loss": 1.2297,
      "step": 18
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0004999960092086724,
      "loss": 1.1762,
      "step": 19
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0004999840369621011,
      "loss": 1.1648,
      "step": 20
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0004999640836425159,
      "loss": 1.2112,
      "step": 21
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.000499936149886953,
      "loss": 1.1805,
      "step": 22
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0004999002365872348,
      "loss": 1.1409,
      "step": 23
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0004998563448899413,
      "loss": 1.1581,
      "step": 24
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.000499804476196373,
      "loss": 1.1575,
      "step": 25
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0004997446321625073,
      "loss": 1.1499,
      "step": 26
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0004996768146989446,
      "loss": 1.1491,
      "step": 27
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0004996010259708473,
      "loss": 1.1246,
      "step": 28
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.000499517268397872,
      "loss": 1.1482,
      "step": 29
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0004994255446540899,
      "loss": 1.1153,
      "step": 30
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0004993258576679043,
      "loss": 1.1439,
      "step": 31
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0004992182106219544,
      "loss": 1.1065,
      "step": 32
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0004991026069530156,
      "loss": 1.0962,
      "step": 33
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0004989790503518888,
      "loss": 1.0437,
      "step": 34
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0004988475447632828,
      "loss": 1.1058,
      "step": 35
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0004987080943856887,
      "loss": 1.1249,
      "step": 36
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0004985607036712452,
      "loss": 1.1146,
      "step": 37
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0004984053773255971,
      "loss": 1.0915,
      "step": 38
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0004982421203077445,
      "loss": 1.1169,
      "step": 39
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0004980709378298851,
      "loss": 1.0296,
      "step": 40
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0004978918353572471,
      "loss": 1.0623,
      "step": 41
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0004977048186079155,
      "loss": 1.0638,
      "step": 42
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0004975098935526487,
      "loss": 1.031,
      "step": 43
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0004973070664146885,
      "loss": 1.0486,
      "step": 44
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0004970963436695612,
      "loss": 1.0436,
      "step": 45
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0004968777320448706,
      "loss": 1.0549,
      "step": 46
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0004966512385200841,
      "loss": 1.0772,
      "step": 47
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0004964168703263086,
      "loss": 0.9709,
      "step": 48
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0004961746349460606,
      "loss": 1.0218,
      "step": 49
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0004959245401130269,
      "loss": 1.0372,
      "step": 50
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0004956665938118179,
      "loss": 1.048,
      "step": 51
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0004954008042777125,
      "loss": 1.0615,
      "step": 52
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0004951271799963951,
      "loss": 1.0119,
      "step": 53
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0004948457297036851,
      "loss": 0.9908,
      "step": 54
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0004945564623852576,
      "loss": 0.9936,
      "step": 55
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0004942593872763566,
      "loss": 1.0219,
      "step": 56
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0004939545138615002,
      "loss": 1.0299,
      "step": 57
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.000493641851874178,
      "loss": 0.9755,
      "step": 58
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0004933214112965398,
      "loss": 0.9702,
      "step": 59
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0004929932023590776,
      "loss": 0.963,
      "step": 60
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0004926572355402983,
      "loss": 0.9493,
      "step": 61
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0004923135215663897,
      "loss": 0.9647,
      "step": 62
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0004919620714108777,
      "loss": 0.936,
      "step": 63
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0004916028962942763,
      "loss": 0.9509,
      "step": 64
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0004912360076837289,
      "loss": 0.9644,
      "step": 65
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0004908614172926426,
      "loss": 0.9685,
      "step": 66
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0004904791370803142,
      "loss": 1.0046,
      "step": 67
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0004900891792515479,
      "loss": 0.9882,
      "step": 68
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0004896915562562665,
      "loss": 0.9508,
      "step": 69
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0004892862807891131,
      "loss": 0.9531,
      "step": 70
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0004888733657890463,
      "loss": 0.9307,
      "step": 71
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0004884528244389269,
      "loss": 0.9529,
      "step": 72
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.000488024670165097,
      "loss": 0.9039,
      "step": 73
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0004875889166369516,
      "loss": 0.8599,
      "step": 74
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.000487145577766502,
      "loss": 0.9116,
      "step": 75
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00048669466770793134,
      "loss": 0.8865,
      "step": 76
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0004862362008571434,
      "loss": 0.9277,
      "step": 77
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00048577019185130227,
      "loss": 0.9473,
      "step": 78
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0004852966555683657,
      "loss": 0.8967,
      "step": 79
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00048481560712660955,
      "loss": 0.9528,
      "step": 80
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00048432706188414543,
      "loss": 0.9099,
      "step": 81
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00048383103543843034,
      "loss": 0.8784,
      "step": 82
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00048332754362576835,
      "loss": 0.9174,
      "step": 83
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0004828166025208058,
      "loss": 0.8529,
      "step": 84
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00048229822843601726,
      "loss": 0.8678,
      "step": 85
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0004817724379211851,
      "loss": 0.8857,
      "step": 86
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00048123924776287115,
      "loss": 0.8852,
      "step": 87
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00048069867498388064,
      "loss": 0.9003,
      "step": 88
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.000480150736842719,
      "loss": 0.8946,
      "step": 89
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0004795954508330403,
      "loss": 0.8796,
      "step": 90
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00047903283468308936,
      "loss": 0.8279,
      "step": 91
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00047846290635513534,
      "loss": 0.8816,
      "step": 92
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0004778856840448985,
      "loss": 0.8773,
      "step": 93
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0004773011861809693,
      "loss": 0.813,
      "step": 94
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0004767094314242195,
      "loss": 0.8728,
      "step": 95
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0004761104386672074,
      "loss": 0.8441,
      "step": 96
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00047550422703357355,
      "loss": 0.8635,
      "step": 97
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0004748908158774312,
      "loss": 0.8549,
      "step": 98
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00047427022478274764,
      "loss": 0.8721,
      "step": 99
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0004736424735627193,
      "loss": 0.8716,
      "step": 100
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0004730075822591392,
      "loss": 0.8211,
      "step": 101
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.000472365571141757,
      "loss": 0.8409,
      "step": 102
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.000471716460707632,
      "loss": 0.7906,
      "step": 103
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0004710602716804784,
      "loss": 0.8018,
      "step": 104
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00047039702501000405,
      "loss": 0.8292,
      "step": 105
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0004697267418712415,
      "loss": 0.8371,
      "step": 106
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0004690494436638718,
      "loss": 0.7689,
      "step": 107
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0004683651520115414,
      "loss": 0.7916,
      "step": 108
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00046767388876117196,
      "loss": 0.8327,
      "step": 109
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0004669756759822624,
      "loss": 0.773,
      "step": 110
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.000466270535966185,
      "loss": 0.774,
      "step": 111
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00046555849122547266,
      "loss": 0.7407,
      "step": 112
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00046483956449310155,
      "loss": 0.7295,
      "step": 113
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00046411377872176404,
      "loss": 0.7941,
      "step": 114
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00046338115708313666,
      "loss": 0.7871,
      "step": 115
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00046264172296714,
      "loss": 0.7379,
      "step": 116
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0004618954999811923,
      "loss": 0.8152,
      "step": 117
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0004611425119494551,
      "loss": 0.7843,
      "step": 118
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00046038278291207334,
      "loss": 0.7352,
      "step": 119
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0004596163371244076,
      "loss": 0.6882,
      "step": 120
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0004588431990562593,
      "loss": 0.7295,
      "step": 121
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0004580633933910901,
      "loss": 0.7648,
      "step": 122
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0004572769450252335,
      "loss": 0.6952,
      "step": 123
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00045648387906709997,
      "loss": 0.7591,
      "step": 124
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00045568422083637554,
      "loss": 0.7524,
      "step": 125
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0004548779958632133,
      "loss": 0.7553,
      "step": 126
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00045406522988741826,
      "loss": 0.6946,
      "step": 127
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0004532459488576258,
      "loss": 0.7236,
      "step": 128
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0004524201789304727,
      "loss": 0.6709,
      "step": 129
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00045158794646976285,
      "loss": 0.6892,
      "step": 130
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0004507492780456249,
      "loss": 0.6584,
      "step": 131
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0004499042004336642,
      "loss": 0.7718,
      "step": 132
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0004490527406141081,
      "loss": 0.769,
      "step": 133
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00044819492577094424,
      "loss": 0.7023,
      "step": 134
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0004473307832910529,
      "loss": 0.6831,
      "step": 135
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00044646034076333257,
      "loss": 0.7355,
      "step": 136
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00044558362597781927,
      "loss": 0.7315,
      "step": 137
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00044470066692479905,
      "loss": 0.6711,
      "step": 138
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0004438114917939145,
      "loss": 0.6975,
      "step": 139
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00044291612897326497,
      "loss": 0.6449,
      "step": 140
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0004420146070484997,
      "loss": 0.7436,
      "step": 141
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00044110695480190597,
      "loss": 0.66,
      "step": 142
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00044019320121148933,
      "loss": 0.6256,
      "step": 143
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.000439273375450049,
      "loss": 0.5968,
      "step": 144
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0004383475068842464,
      "loss": 0.6513,
      "step": 145
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0004374156250736675,
      "loss": 0.6522,
      "step": 146
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.000436477759769879,
      "loss": 0.6148,
      "step": 147
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0004355339409154788,
      "loss": 0.6192,
      "step": 148
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0004345841986431396,
      "loss": 0.6312,
      "step": 149
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00043362856327464717,
      "loss": 0.6204,
      "step": 150
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00043266706531993227,
      "loss": 0.6445,
      "step": 151
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0004316997354760965,
      "loss": 0.5941,
      "step": 152
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00043072660462643224,
      "loss": 0.5731,
      "step": 153
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0004297477038394368,
      "loss": 0.6015,
      "step": 154
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0004287630643678203,
      "loss": 0.6448,
      "step": 155
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.000427772717647508,
      "loss": 0.6373,
      "step": 156
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00042677669529663686,
      "loss": 0.5865,
      "step": 157
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00042577502911454566,
      "loss": 0.5902,
      "step": 158
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00042476775108076013,
      "loss": 0.5728,
      "step": 159
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00042375489335397183,
      "loss": 0.6146,
      "step": 160
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00042273648827101135,
      "loss": 0.5693,
      "step": 161
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00042171256834581613,
      "loss": 0.5781,
      "step": 162
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0004206831662683922,
      "loss": 0.6095,
      "step": 163
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00041964831490377065,
      "loss": 0.6044,
      "step": 164
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0004186080472909582,
      "loss": 0.5669,
      "step": 165
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0004175623966418827,
      "loss": 0.566,
      "step": 166
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0004165113963403325,
      "loss": 0.5583,
      "step": 167
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0004154550799408906,
      "loss": 0.5875,
      "step": 168
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0004143934811678637,
      "loss": 0.5918,
      "step": 169
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00041332663391420514,
      "loss": 0.513,
      "step": 170
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0004122545722404332,
      "loss": 0.5142,
      "step": 171
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0004111773303735432,
      "loss": 0.5566,
      "step": 172
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0004100949427059151,
      "loss": 0.5472,
      "step": 173
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0004090074437942155,
      "loss": 0.5576,
      "step": 174
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0004079148683582943,
      "loss": 0.497,
      "step": 175
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.000406817251280076,
      "loss": 0.5864,
      "step": 176
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00040571462760244625,
      "loss": 0.583,
      "step": 177
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00040460703252813326,
      "loss": 0.5339,
      "step": 178
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00040349450141858353,
      "loss": 0.5486,
      "step": 179
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00040237706979283307,
      "loss": 0.5515,
      "step": 180
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0004012547733263734,
      "loss": 0.53,
      "step": 181
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00040012764785001264,
      "loss": 0.5704,
      "step": 182
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00039899572934873134,
      "loss": 0.528,
      "step": 183
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00039785905396053377,
      "loss": 0.511,
      "step": 184
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00039671765797529425,
      "loss": 0.5064,
      "step": 185
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00039557157783359835,
      "loss": 0.5058,
      "step": 186
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0003944208501255796,
      "loss": 0.4964,
      "step": 187
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00039326551158975126,
      "loss": 0.5229,
      "step": 188
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0003921055991118334,
      "loss": 0.5182,
      "step": 189
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0003909411497235752,
      "loss": 0.5174,
      "step": 190
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00038977220060157293,
      "loss": 0.4709,
      "step": 191
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0003885987890660827,
      "loss": 0.4543,
      "step": 192
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0003874209525798292,
      "loss": 0.5248,
      "step": 193
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00038623872874680944,
      "loss": 0.5144,
      "step": 194
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0003850521553110923,
      "loss": 0.4847,
      "step": 195
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0003838612701556138,
      "loss": 0.4903,
      "step": 196
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00038266611130096707,
      "loss": 0.4763,
      "step": 197
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00038146671690418864,
      "loss": 0.4949,
      "step": 198
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0003802631252575409,
      "loss": 0.4662,
      "step": 199
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00037905537478728844,
      "loss": 0.45,
      "step": 200
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0003778435040524721,
      "loss": 0.483,
      "step": 201
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0003766275517436779,
      "loss": 0.4372,
      "step": 202
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.000375407556681801,
      "loss": 0.4271,
      "step": 203
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00037418355781680706,
      "loss": 0.4382,
      "step": 204
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0003729555942264887,
      "loss": 0.4246,
      "step": 205
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0003717237051152175,
      "loss": 0.46,
      "step": 206
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0003704879298126924,
      "loss": 0.4396,
      "step": 207
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00036924830777268426,
      "loss": 0.4577,
      "step": 208
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0003680048785717763,
      "loss": 0.421,
      "step": 209
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0003667576819081002,
      "loss": 0.4514,
      "step": 210
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00036550675760006903,
      "loss": 0.4422,
      "step": 211
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0003642521455851058,
      "loss": 0.4365,
      "step": 212
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0003629938859183686,
      "loss": 0.4277,
      "step": 213
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00036173201877147133,
      "loss": 0.3857,
      "step": 214
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00036046658443120193,
      "loss": 0.4216,
      "step": 215
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0003591976232982355,
      "loss": 0.4145,
      "step": 216
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00035792517588584467,
      "loss": 0.4466,
      "step": 217
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00035664928281860627,
      "loss": 0.433,
      "step": 218
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00035536998483110416,
      "loss": 0.4262,
      "step": 219
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0003540873227666288,
      "loss": 0.3954,
      "step": 220
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0003528013375758734,
      "loss": 0.3893,
      "step": 221
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0003515120703156264,
      "loss": 0.3572,
      "step": 222
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00035021956214746045,
      "loss": 0.3879,
      "step": 223
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00034892385433641873,
      "loss": 0.3537,
      "step": 224
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00034762498824969724,
      "loss": 0.4048,
      "step": 225
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00034632300535532415,
      "loss": 0.3706,
      "step": 226
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00034501794722083555,
      "loss": 0.3797,
      "step": 227
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0003437098555119493,
      "loss": 0.3766,
      "step": 228
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0003423987719912334,
      "loss": 0.3821,
      "step": 229
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00034108473851677406,
      "loss": 0.3544,
      "step": 230
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0003397677970408384,
      "loss": 0.3566,
      "step": 231
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0003384479896085353,
      "loss": 0.334,
      "step": 232
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0003371253583564732,
      "loss": 0.3199,
      "step": 233
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0003357999455114148,
      "loss": 0.3188,
      "step": 234
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00033447179338892886,
      "loss": 0.3352,
      "step": 235
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.000333140944392039,
      "loss": 0.3162,
      "step": 236
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00033180744100987043,
      "loss": 0.3035,
      "step": 237
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.000330471325816293,
      "loss": 0.3624,
      "step": 238
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00032913264146856196,
      "loss": 0.3712,
      "step": 239
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0003277914307059566,
      "loss": 0.3265,
      "step": 240
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0003264477363484151,
      "loss": 0.3465,
      "step": 241
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00032510160129516774,
      "loss": 0.3062,
      "step": 242
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0003237530685233673,
      "loss": 0.3065,
      "step": 243
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0003224021810867168,
      "loss": 0.3435,
      "step": 244
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00032104898211409545,
      "loss": 0.2982,
      "step": 245
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00031969351480818075,
      "loss": 0.2766,
      "step": 246
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0003183358224440703,
      "loss": 0.3304,
      "step": 247
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00031697594836789923,
      "loss": 0.3243,
      "step": 248
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0003156139359954569,
      "loss": 0.2979,
      "step": 249
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00031424982881080065,
      "loss": 0.2779,
      "step": 250
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0003128836703648676,
      "loss": 0.3393,
      "step": 251
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0003115155042740838,
      "loss": 0.2951,
      "step": 252
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0003101453742189722,
      "loss": 0.2794,
      "step": 253
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.000308773323942758,
      "loss": 0.2846,
      "step": 254
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.000307399397249972,
      "loss": 0.2736,
      "step": 255
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00030602363800505194,
      "loss": 0.2854,
      "step": 256
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0003046460901309424,
      "loss": 0.29,
      "step": 257
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00030326679760769225,
      "loss": 0.2803,
      "step": 258
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00030188580447105054,
      "loss": 0.2655,
      "step": 259
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.0003005031548110607,
      "loss": 0.2737,
      "step": 260
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002991188927706531,
      "loss": 0.2738,
      "step": 261
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00029773306254423515,
      "loss": 0.2817,
      "step": 262
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00029634570837628086,
      "loss": 0.2428,
      "step": 263
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00029495687455991816,
      "loss": 0.2829,
      "step": 264
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0002935666054355146,
      "loss": 0.2698,
      "step": 265
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002921749453892618,
      "loss": 0.2363,
      "step": 266
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00029078193885175875,
      "loss": 0.2332,
      "step": 267
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0002893876302965925,
      "loss": 0.2423,
      "step": 268
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002879920642389189,
      "loss": 0.2341,
      "step": 269
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0002865952852340417,
      "loss": 0.1993,
      "step": 270
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00028519733787598885,
      "loss": 0.2245,
      "step": 271
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00028379826679609,
      "loss": 0.2516,
      "step": 272
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00028239811666155103,
      "loss": 0.2319,
      "step": 273
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00028099693217402806,
      "loss": 0.2414,
      "step": 274
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00027959475806820026,
      "loss": 0.2062,
      "step": 275
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00027819163911034175,
      "loss": 0.2216,
      "step": 276
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002767876200968923,
      "loss": 0.1958,
      "step": 277
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00027538274585302703,
      "loss": 0.2216,
      "step": 278
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002739770612312256,
      "loss": 0.1941,
      "step": 279
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00027257061110984006,
      "loss": 0.1983,
      "step": 280
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00027116344039166193,
      "loss": 0.2329,
      "step": 281
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00026975559400248874,
      "loss": 0.1976,
      "step": 282
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002683471168896899,
      "loss": 0.1861,
      "step": 283
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0002669380540207712,
      "loss": 0.1945,
      "step": 284
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00026552845038193976,
      "loss": 0.2051,
      "step": 285
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00026411835097666747,
      "loss": 0.1895,
      "step": 286
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0002627078008242541,
      "loss": 0.1803,
      "step": 287
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00026129684495839015,
      "loss": 0.1429,
      "step": 288
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.000259885528425719,
      "loss": 0.1173,
      "step": 289
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.000258473896284399,
      "loss": 0.1239,
      "step": 290
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002570619936026647,
      "loss": 0.1269,
      "step": 291
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00025564986545738766,
      "loss": 0.1301,
      "step": 292
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002542375569326382,
      "loss": 0.1173,
      "step": 293
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00025282511311824495,
      "loss": 0.1221,
      "step": 294
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002514125791083563,
      "loss": 0.1281,
      "step": 295
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00025,
      "loss": 0.1055,
      "step": 296
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0002485874208916438,
      "loss": 0.1035,
      "step": 297
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002471748868817551,
      "loss": 0.1182,
      "step": 298
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00024576244306736194,
      "loss": 0.1214,
      "step": 299
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0002443501345426124,
      "loss": 0.101,
      "step": 300
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00024293800639733535,
      "loss": 0.101,
      "step": 301
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00024152610371560093,
      "loss": 0.1127,
      "step": 302
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00024011447157428102,
      "loss": 0.1029,
      "step": 303
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00023870315504160994,
      "loss": 0.1028,
      "step": 304
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00023729219917574594,
      "loss": 0.0947,
      "step": 305
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00023588164902333257,
      "loss": 0.1069,
      "step": 306
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00023447154961806028,
      "loss": 0.0971,
      "step": 307
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00023306194597922885,
      "loss": 0.0993,
      "step": 308
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00023165288311031025,
      "loss": 0.0991,
      "step": 309
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00023024440599751133,
      "loss": 0.1146,
      "step": 310
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0002288365596083381,
      "loss": 0.1039,
      "step": 311
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002274293888901599,
      "loss": 0.1065,
      "step": 312
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0002260229387687744,
      "loss": 0.0932,
      "step": 313
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.000224617254146973,
      "loss": 0.0963,
      "step": 314
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00022321237990310778,
      "loss": 0.0967,
      "step": 315
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00022180836088965831,
      "loss": 0.0892,
      "step": 316
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00022040524193179978,
      "loss": 0.0837,
      "step": 317
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.000219003067825972,
      "loss": 0.0962,
      "step": 318
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00021760188333844904,
      "loss": 0.1008,
      "step": 319
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00021620173320391006,
      "loss": 0.0994,
      "step": 320
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00021480266212401113,
      "loss": 0.0884,
      "step": 321
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00021340471476595833,
      "loss": 0.0943,
      "step": 322
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00021200793576108104,
      "loss": 0.1097,
      "step": 323
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00021061236970340756,
      "loss": 0.0851,
      "step": 324
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002092180611482413,
      "loss": 0.0954,
      "step": 325
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0002078250546107382,
      "loss": 0.0929,
      "step": 326
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00020643339456448545,
      "loss": 0.0921,
      "step": 327
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002050431254400819,
      "loss": 0.0921,
      "step": 328
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002036542916237192,
      "loss": 0.0885,
      "step": 329
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00020226693745576492,
      "loss": 0.0872,
      "step": 330
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00020088110722934688,
      "loss": 0.075,
      "step": 331
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00019949684518893925,
      "loss": 0.084,
      "step": 332
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00019811419552894953,
      "loss": 0.0871,
      "step": 333
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0001967332023923078,
      "loss": 0.0894,
      "step": 334
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00019535390986905762,
      "loss": 0.0946,
      "step": 335
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0001939763619949481,
      "loss": 0.0919,
      "step": 336
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.000192600602750028,
      "loss": 0.0836,
      "step": 337
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.000191226676057242,
      "loss": 0.0848,
      "step": 338
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00018985462578102787,
      "loss": 0.0794,
      "step": 339
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00018848449572591628,
      "loss": 0.0675,
      "step": 340
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00018711632963513237,
      "loss": 0.0796,
      "step": 341
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00018575017118919928,
      "loss": 0.0711,
      "step": 342
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0001843860640045431,
      "loss": 0.0821,
      "step": 343
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00018302405163210078,
      "loss": 0.0822,
      "step": 344
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00018166417755592973,
      "loss": 0.0681,
      "step": 345
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00018030648519181924,
      "loss": 0.0778,
      "step": 346
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001789510178859046,
      "loss": 0.075,
      "step": 347
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001775978189132832,
      "loss": 0.0721,
      "step": 348
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001762469314766328,
      "loss": 0.0811,
      "step": 349
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00017489839870483235,
      "loss": 0.0726,
      "step": 350
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00017355226365158488,
      "loss": 0.0801,
      "step": 351
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0001722085692940434,
      "loss": 0.0685,
      "step": 352
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00017086735853143802,
      "loss": 0.066,
      "step": 353
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00016952867418370708,
      "loss": 0.0669,
      "step": 354
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001681925589901296,
      "loss": 0.062,
      "step": 355
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.000166859055607961,
      "loss": 0.067,
      "step": 356
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00016552820661107118,
      "loss": 0.0721,
      "step": 357
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00016420005448858523,
      "loss": 0.0581,
      "step": 358
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00016287464164352683,
      "loss": 0.0698,
      "step": 359
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001615520103914648,
      "loss": 0.0725,
      "step": 360
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0001602322029591616,
      "loss": 0.0736,
      "step": 361
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00015891526148322595,
      "loss": 0.0628,
      "step": 362
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0001576012280087666,
      "loss": 0.0677,
      "step": 363
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0001562901444880508,
      "loss": 0.0587,
      "step": 364
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00015498205277916444,
      "loss": 0.0541,
      "step": 365
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00015367699464467595,
      "loss": 0.0632,
      "step": 366
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0001523750117503028,
      "loss": 0.064,
      "step": 367
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00015107614566358136,
      "loss": 0.0608,
      "step": 368
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00014978043785253965,
      "loss": 0.0656,
      "step": 369
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00014848792968437375,
      "loss": 0.061,
      "step": 370
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00014719866242412662,
      "loss": 0.0575,
      "step": 371
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0001459126772333712,
      "loss": 0.0672,
      "step": 372
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00014463001516889595,
      "loss": 0.0557,
      "step": 373
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00014335071718139377,
      "loss": 0.0524,
      "step": 374
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00014207482411415532,
      "loss": 0.0711,
      "step": 375
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00014080237670176455,
      "loss": 0.0467,
      "step": 376
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00013953341556879808,
      "loss": 0.0643,
      "step": 377
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00013826798122852868,
      "loss": 0.0605,
      "step": 378
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00013700611408163157,
      "loss": 0.0541,
      "step": 379
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00013574785441489428,
      "loss": 0.0499,
      "step": 380
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00013449324239993093,
      "loss": 0.0463,
      "step": 381
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00013324231809189985,
      "loss": 0.0614,
      "step": 382
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00013199512142822372,
      "loss": 0.0521,
      "step": 383
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00013075169222731572,
      "loss": 0.0468,
      "step": 384
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0001295120701873077,
      "loss": 0.0535,
      "step": 385
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00012827629488478253,
      "loss": 0.0534,
      "step": 386
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001270444057735113,
      "loss": 0.0571,
      "step": 387
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.000125816442183193,
      "loss": 0.0482,
      "step": 388
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001245924433181991,
      "loss": 0.048,
      "step": 389
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00012337244825632215,
      "loss": 0.0468,
      "step": 390
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0001221564959475278,
      "loss": 0.052,
      "step": 391
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00012094462521271154,
      "loss": 0.0465,
      "step": 392
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0001197368747424592,
      "loss": 0.0423,
      "step": 393
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00011853328309581137,
      "loss": 0.0463,
      "step": 394
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.000117333888699033,
      "loss": 0.0517,
      "step": 395
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011613872984438628,
      "loss": 0.0568,
      "step": 396
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011494784468890771,
      "loss": 0.0376,
      "step": 397
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011376127125319063,
      "loss": 0.0339,
      "step": 398
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011257904742017081,
      "loss": 0.0395,
      "step": 399
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011140121093391736,
      "loss": 0.0425,
      "step": 400
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011022779939842702,
      "loss": 0.0376,
      "step": 401
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00010905885027642484,
      "loss": 0.0422,
      "step": 402
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00010789440088816665,
      "loss": 0.0501,
      "step": 403
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00010673448841024874,
      "loss": 0.0471,
      "step": 404
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00010557914987442046,
      "loss": 0.0546,
      "step": 405
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00010442842216640166,
      "loss": 0.0387,
      "step": 406
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00010328234202470574,
      "loss": 0.0341,
      "step": 407
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00010214094603946631,
      "loss": 0.0446,
      "step": 408
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00010100427065126874,
      "loss": 0.0313,
      "step": 409
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.987235214998742e-05,
      "loss": 0.0315,
      "step": 410
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.87452266736266e-05,
      "loss": 0.0414,
      "step": 411
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.762293020716694e-05,
      "loss": 0.0325,
      "step": 412
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.650549858141647e-05,
      "loss": 0.0363,
      "step": 413
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.539296747186679e-05,
      "loss": 0.0387,
      "step": 414
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.42853723975538e-05,
      "loss": 0.0301,
      "step": 415
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.318274871992408e-05,
      "loss": 0.0286,
      "step": 416
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.208513164170579e-05,
      "loss": 0.0303,
      "step": 417
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.09925562057845e-05,
      "loss": 0.0354,
      "step": 418
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.990505729408494e-05,
      "loss": 0.031,
      "step": 419
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.882266962645694e-05,
      "loss": 0.0268,
      "step": 420
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.774542775956679e-05,
      "loss": 0.0287,
      "step": 421
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.667336608579487e-05,
      "loss": 0.0325,
      "step": 422
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.560651883213633e-05,
      "loss": 0.0375,
      "step": 423
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.454492005910941e-05,
      "loss": 0.0379,
      "step": 424
    },
    {
      "epoch": 1.48,
      "learning_rate": 8.348860365966759e-05,
      "loss": 0.0266,
      "step": 425
    },
    {
      "epoch": 1.48,
      "learning_rate": 8.243760335811734e-05,
      "loss": 0.0272,
      "step": 426
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.139195270904182e-05,
      "loss": 0.0238,
      "step": 427
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.035168509622948e-05,
      "loss": 0.0283,
      "step": 428
    },
    {
      "epoch": 1.49,
      "learning_rate": 7.931683373160789e-05,
      "loss": 0.0241,
      "step": 429
    },
    {
      "epoch": 1.5,
      "learning_rate": 7.828743165418392e-05,
      "loss": 0.0326,
      "step": 430
    },
    {
      "epoch": 1.5,
      "learning_rate": 7.726351172898869e-05,
      "loss": 0.0251,
      "step": 431
    },
    {
      "epoch": 1.5,
      "learning_rate": 7.624510664602819e-05,
      "loss": 0.0272,
      "step": 432
    },
    {
      "epoch": 1.51,
      "learning_rate": 7.523224891923982e-05,
      "loss": 0.0243,
      "step": 433
    },
    {
      "epoch": 1.51,
      "learning_rate": 7.422497088545435e-05,
      "loss": 0.0319,
      "step": 434
    },
    {
      "epoch": 1.51,
      "learning_rate": 7.322330470336314e-05,
      "loss": 0.0289,
      "step": 435
    },
    {
      "epoch": 1.52,
      "learning_rate": 7.222728235249196e-05,
      "loss": 0.0295,
      "step": 436
    },
    {
      "epoch": 1.52,
      "learning_rate": 7.123693563217979e-05,
      "loss": 0.0206,
      "step": 437
    },
    {
      "epoch": 1.52,
      "learning_rate": 7.025229616056325e-05,
      "loss": 0.0285,
      "step": 438
    },
    {
      "epoch": 1.53,
      "learning_rate": 6.927339537356777e-05,
      "loss": 0.0267,
      "step": 439
    },
    {
      "epoch": 1.53,
      "learning_rate": 6.830026452390353e-05,
      "loss": 0.0232,
      "step": 440
    },
    {
      "epoch": 1.53,
      "learning_rate": 6.733293468006773e-05,
      "loss": 0.025,
      "step": 441
    },
    {
      "epoch": 1.54,
      "learning_rate": 6.637143672535281e-05,
      "loss": 0.0219,
      "step": 442
    },
    {
      "epoch": 1.54,
      "learning_rate": 6.541580135686045e-05,
      "loss": 0.0232,
      "step": 443
    },
    {
      "epoch": 1.54,
      "learning_rate": 6.446605908452121e-05,
      "loss": 0.0212,
      "step": 444
    },
    {
      "epoch": 1.55,
      "learning_rate": 6.352224023012096e-05,
      "loss": 0.0268,
      "step": 445
    },
    {
      "epoch": 1.55,
      "learning_rate": 6.258437492633254e-05,
      "loss": 0.0219,
      "step": 446
    },
    {
      "epoch": 1.55,
      "learning_rate": 6.16524931157536e-05,
      "loss": 0.0262,
      "step": 447
    },
    {
      "epoch": 1.56,
      "learning_rate": 6.072662454995101e-05,
      "loss": 0.0256,
      "step": 448
    },
    {
      "epoch": 1.56,
      "learning_rate": 5.980679878851075e-05,
      "loss": 0.0202,
      "step": 449
    },
    {
      "epoch": 1.57,
      "learning_rate": 5.889304519809402e-05,
      "loss": 0.0239,
      "step": 450
    },
    {
      "epoch": 1.57,
      "learning_rate": 5.7985392951500265e-05,
      "loss": 0.0201,
      "step": 451
    },
    {
      "epoch": 1.57,
      "learning_rate": 5.708387102673507e-05,
      "loss": 0.0199,
      "step": 452
    },
    {
      "epoch": 1.58,
      "learning_rate": 5.618850820608548e-05,
      "loss": 0.0196,
      "step": 453
    },
    {
      "epoch": 1.58,
      "learning_rate": 5.529933307520102e-05,
      "loss": 0.0162,
      "step": 454
    },
    {
      "epoch": 1.58,
      "learning_rate": 5.4416374022180776e-05,
      "loss": 0.0211,
      "step": 455
    },
    {
      "epoch": 1.59,
      "learning_rate": 5.353965923666743e-05,
      "loss": 0.0268,
      "step": 456
    },
    {
      "epoch": 1.59,
      "learning_rate": 5.26692167089472e-05,
      "loss": 0.0166,
      "step": 457
    },
    {
      "epoch": 1.59,
      "learning_rate": 5.180507422905584e-05,
      "loss": 0.0229,
      "step": 458
    },
    {
      "epoch": 1.6,
      "learning_rate": 5.094725938589193e-05,
      "loss": 0.0252,
      "step": 459
    },
    {
      "epoch": 1.6,
      "learning_rate": 5.009579956633578e-05,
      "loss": 0.0185,
      "step": 460
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.9250721954375105e-05,
      "loss": 0.0185,
      "step": 461
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.8412053530237145e-05,
      "loss": 0.0231,
      "step": 462
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.757982106952735e-05,
      "loss": 0.0219,
      "step": 463
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.675405114237427e-05,
      "loss": 0.019,
      "step": 464
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.593477011258171e-05,
      "loss": 0.0188,
      "step": 465
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.512200413678672e-05,
      "loss": 0.0135,
      "step": 466
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.4315779163624474e-05,
      "loss": 0.0268,
      "step": 467
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.3516120932900055e-05,
      "loss": 0.0207,
      "step": 468
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.2723054974766585e-05,
      "loss": 0.0154,
      "step": 469
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.1936606608909886e-05,
      "loss": 0.016,
      "step": 470
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.1156800943740744e-05,
      "loss": 0.0156,
      "step": 471
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.0383662875592446e-05,
      "loss": 0.0198,
      "step": 472
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.961721708792662e-05,
      "loss": 0.0157,
      "step": 473
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.88574880505449e-05,
      "loss": 0.0192,
      "step": 474
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.8104500018807805e-05,
      "loss": 0.0158,
      "step": 475
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.7358277032860015e-05,
      "loss": 0.0159,
      "step": 476
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.661884291686338e-05,
      "loss": 0.0179,
      "step": 477
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.5886221278236045e-05,
      "loss": 0.0198,
      "step": 478
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.516043550689851e-05,
      "loss": 0.0187,
      "step": 479
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.444150877452734e-05,
      "loss": 0.0181,
      "step": 480
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.372946403381508e-05,
      "loss": 0.0137,
      "step": 481
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.3024324017737555e-05,
      "loss": 0.0207,
      "step": 482
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.232611123882809e-05,
      "loss": 0.0175,
      "step": 483
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.1634847988458616e-05,
      "loss": 0.0149,
      "step": 484
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.0950556336128255e-05,
      "loss": 0.0132,
      "step": 485
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.0273258128758585e-05,
      "loss": 0.0114,
      "step": 486
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.9602974989996008e-05,
      "loss": 0.0126,
      "step": 487
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.8939728319521658e-05,
      "loss": 0.0186,
      "step": 488
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.82835392923681e-05,
      "loss": 0.0167,
      "step": 489
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.7634428858242992e-05,
      "loss": 0.0169,
      "step": 490
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.699241774086081e-05,
      "loss": 0.0173,
      "step": 491
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.635752643728076e-05,
      "loss": 0.019,
      "step": 492
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.572977521725242e-05,
      "loss": 0.0111,
      "step": 493
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.5109184122568794e-05,
      "loss": 0.0139,
      "step": 494
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.4495772966426466e-05,
      "loss": 0.0134,
      "step": 495
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.3889561332792658e-05,
      "loss": 0.0134,
      "step": 496
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.329056857578049e-05,
      "loss": 0.0147,
      "step": 497
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.26988138190308e-05,
      "loss": 0.0152,
      "step": 498
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.2114315955101494e-05,
      "loss": 0.0115,
      "step": 499
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.153709364486467e-05,
      "loss": 0.0154,
      "step": 500
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.0967165316910674e-05,
      "loss": 0.0142,
      "step": 501
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.040454916695972e-05,
      "loss": 0.0128,
      "step": 502
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.9849263157281054e-05,
      "loss": 0.0133,
      "step": 503
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.9301325016119338e-05,
      "loss": 0.0109,
      "step": 504
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.8760752237128863e-05,
      "loss": 0.0101,
      "step": 505
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.8227562078814902e-05,
      "loss": 0.0126,
      "step": 506
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.7701771563982754e-05,
      "loss": 0.0116,
      "step": 507
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.7183397479194173e-05,
      "loss": 0.0132,
      "step": 508
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.667245637423162e-05,
      "loss": 0.011,
      "step": 509
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.6168964561569717e-05,
      "loss": 0.0117,
      "step": 510
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.5672938115854545e-05,
      "loss": 0.0218,
      "step": 511
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.5184392873390462e-05,
      "loss": 0.0088,
      "step": 512
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.470334443163432e-05,
      "loss": 0.0084,
      "step": 513
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.422980814869773e-05,
      "loss": 0.0179,
      "step": 514
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.3763799142856692e-05,
      "loss": 0.0101,
      "step": 515
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.3305332292068706e-05,
      "loss": 0.0077,
      "step": 516
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.285442223349806e-05,
      "loss": 0.0083,
      "step": 517
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.2411083363048386e-05,
      "loss": 0.0107,
      "step": 518
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.1975329834903015e-05,
      "loss": 0.008,
      "step": 519
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.1547175561073154e-05,
      "loss": 0.0109,
      "step": 520
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.1126634210953751e-05,
      "loss": 0.0117,
      "step": 521
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.071371921088693e-05,
      "loss": 0.0113,
      "step": 522
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.0308443743733548e-05,
      "loss": 0.0088,
      "step": 523
    },
    {
      "epoch": 1.82,
      "learning_rate": 9.910820748452149e-06,
      "loss": 0.0107,
      "step": 524
    },
    {
      "epoch": 1.83,
      "learning_rate": 9.520862919685902e-06,
      "loss": 0.0098,
      "step": 525
    },
    {
      "epoch": 1.83,
      "learning_rate": 9.13858270735743e-06,
      "loss": 0.0106,
      "step": 526
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.763992316271175e-06,
      "loss": 0.0089,
      "step": 527
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.397103705723775e-06,
      "loss": 0.0141,
      "step": 528
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.037928589122306e-06,
      "loss": 0.0099,
      "step": 529
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.686478433610338e-06,
      "loss": 0.0115,
      "step": 530
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.342764459701723e-06,
      "loss": 0.0094,
      "step": 531
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.006797640922435e-06,
      "loss": 0.0131,
      "step": 532
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.678588703460164e-06,
      "loss": 0.0096,
      "step": 533
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.358148125822e-06,
      "loss": 0.0077,
      "step": 534
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.0454861384997565e-06,
      "loss": 0.0105,
      "step": 535
    },
    {
      "epoch": 1.86,
      "learning_rate": 5.740612723643401e-06,
      "loss": 0.0137,
      "step": 536
    },
    {
      "epoch": 1.87,
      "learning_rate": 5.443537614742395e-06,
      "loss": 0.0109,
      "step": 537
    },
    {
      "epoch": 1.87,
      "learning_rate": 5.154270296314878e-06,
      "loss": 0.0095,
      "step": 538
    },
    {
      "epoch": 1.87,
      "learning_rate": 4.872820003604922e-06,
      "loss": 0.0082,
      "step": 539
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.599195722287536e-06,
      "loss": 0.0106,
      "step": 540
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.333406188182093e-06,
      "loss": 0.0115,
      "step": 541
    },
    {
      "epoch": 1.89,
      "learning_rate": 4.075459886973082e-06,
      "loss": 0.0116,
      "step": 542
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.825365053939406e-06,
      "loss": 0.0079,
      "step": 543
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.5831296736914267e-06,
      "loss": 0.0135,
      "step": 544
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.3487614799159185e-06,
      "loss": 0.0084,
      "step": 545
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.1222679551293488e-06,
      "loss": 0.0081,
      "step": 546
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.903656330438903e-06,
      "loss": 0.0095,
      "step": 547
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.69293358531153e-06,
      "loss": 0.0093,
      "step": 548
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.490106447351315e-06,
      "loss": 0.0098,
      "step": 549
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.295181392084511e-06,
      "loss": 0.011,
      "step": 550
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.1081646427528465e-06,
      "loss": 0.0085,
      "step": 551
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.9290621701149314e-06,
      "loss": 0.0092,
      "step": 552
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.7578796922554928e-06,
      "loss": 0.0081,
      "step": 553
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.5946226744029402e-06,
      "loss": 0.009,
      "step": 554
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.43929632875478e-06,
      "loss": 0.0121,
      "step": 555
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.291905614311306e-06,
      "loss": 0.0086,
      "step": 556
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.1524552367171703e-06,
      "loss": 0.008,
      "step": 557
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.0209496481112247e-06,
      "loss": 0.01,
      "step": 558
    },
    {
      "epoch": 1.94,
      "learning_rate": 8.973930469844127e-07,
      "loss": 0.0108,
      "step": 559
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.81789378045572e-07,
      "loss": 0.0091,
      "step": 560
    },
    {
      "epoch": 1.95,
      "learning_rate": 6.741423320957286e-07,
      "loss": 0.0128,
      "step": 561
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.744553459100243e-07,
      "loss": 0.0103,
      "step": 562
    },
    {
      "epoch": 1.96,
      "learning_rate": 4.827316021281114e-07,
      "loss": 0.0099,
      "step": 563
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.9897402915262113e-07,
      "loss": 0.0101,
      "step": 564
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.2318530105546194e-07,
      "loss": 0.0064,
      "step": 565
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.553678374926649e-07,
      "loss": 0.0082,
      "step": 566
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.9552380362697352e-07,
      "loss": 0.0087,
      "step": 567
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.4365511005878795e-07,
      "loss": 0.008,
      "step": 568
    },
    {
      "epoch": 1.98,
      "learning_rate": 9.976341276521362e-08,
      "loss": 0.009,
      "step": 569
    },
    {
      "epoch": 1.98,
      "learning_rate": 6.385011304704813e-08,
      "loss": 0.0094,
      "step": 570
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.5916357484178095e-08,
      "loss": 0.0087,
      "step": 571
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.5963037898913957e-08,
      "loss": 0.0084,
      "step": 572
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.990791327568299e-09,
      "loss": 0.0073,
      "step": 573
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0,
      "loss": 0.0088,
      "step": 574
    },
    {
      "epoch": 2.0,
      "step": 574,
      "total_flos": 547447928258560.0,
      "train_loss": 0.36620661543068905,
      "train_runtime": 20477.0417,
      "train_samples_per_second": 3.593,
      "train_steps_per_second": 0.028
    }
  ],
  "max_steps": 574,
  "num_train_epochs": 2,
  "total_flos": 547447928258560.0,
  "trial_name": null,
  "trial_params": null
}