{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 600,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 5.555555555555556e-06,
      "loss": 1.3296,
      "step": 1
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 1.0179,
      "step": 2
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.3208,
      "step": 3
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 0.8602,
      "step": 4
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.8253,
      "step": 5
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 1.4955,
      "step": 6
    },
    {
      "epoch": 0.07,
      "learning_rate": 3.888888888888889e-05,
      "loss": 1.356,
      "step": 7
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 1.3023,
      "step": 8
    },
    {
      "epoch": 0.09,
      "learning_rate": 5e-05,
      "loss": 1.1931,
      "step": 9
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.555555555555556e-05,
      "loss": 1.331,
      "step": 10
    },
    {
      "epoch": 0.11,
      "learning_rate": 6.111111111111112e-05,
      "loss": 1.5246,
      "step": 11
    },
    {
      "epoch": 0.12,
      "learning_rate": 6.666666666666667e-05,
      "loss": 1.543,
      "step": 12
    },
    {
      "epoch": 0.13,
      "learning_rate": 7.222222222222222e-05,
      "loss": 1.6207,
      "step": 13
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.777777777777778e-05,
      "loss": 1.0532,
      "step": 14
    },
    {
      "epoch": 0.15,
      "learning_rate": 8.333333333333334e-05,
      "loss": 0.738,
      "step": 15
    },
    {
      "epoch": 0.16,
      "learning_rate": 8.888888888888889e-05,
      "loss": 1.4879,
      "step": 16
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.444444444444444e-05,
      "loss": 0.8541,
      "step": 17
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0001,
      "loss": 0.3011,
      "step": 18
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.999927156177032e-05,
      "loss": 0.5649,
      "step": 19
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.999708626830618e-05,
      "loss": 1.2928,
      "step": 20
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.999344418328162e-05,
      "loss": 0.7608,
      "step": 21
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.998834541281798e-05,
      "loss": 1.058,
      "step": 22
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.998179010548088e-05,
      "loss": 1.3221,
      "step": 23
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.997377845227576e-05,
      "loss": 0.8712,
      "step": 24
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.996431068664237e-05,
      "loss": 0.5889,
      "step": 25
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.995338708444804e-05,
      "loss": 1.0206,
      "step": 26
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.994100796397954e-05,
      "loss": 1.0047,
      "step": 27
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.992717368593385e-05,
      "loss": 0.6275,
      "step": 28
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.991188465340766e-05,
      "loss": 1.0235,
      "step": 29
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.989514131188559e-05,
      "loss": 1.1663,
      "step": 30
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.987694414922724e-05,
      "loss": 0.5028,
      "step": 31
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.985729369565299e-05,
      "loss": 0.1488,
      "step": 32
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.983619052372848e-05,
      "loss": 0.5894,
      "step": 33
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.9813635248348e-05,
      "loss": 0.1516,
      "step": 34
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.978962852671656e-05,
      "loss": 0.6211,
      "step": 35
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.97641710583307e-05,
      "loss": 0.4634,
      "step": 36
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.973726358495817e-05,
      "loss": 1.3872,
      "step": 37
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.970890689061622e-05,
      "loss": 1.3923,
      "step": 38
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.967910180154889e-05,
      "loss": 1.1016,
      "step": 39
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.964784918620282e-05,
      "loss": 0.2436,
      "step": 40
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.961514995520201e-05,
      "loss": 1.2478,
      "step": 41
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.958100506132127e-05,
      "loss": 0.4752,
      "step": 42
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.95454154994584e-05,
      "loss": 0.7473,
      "step": 43
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.950838230660534e-05,
      "loss": 0.5468,
      "step": 44
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.946990656181781e-05,
      "loss": 0.9922,
      "step": 45
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.942998938618394e-05,
      "loss": 0.7146,
      "step": 46
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.938863194279163e-05,
      "loss": 0.941,
      "step": 47
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.934583543669453e-05,
      "loss": 1.1213,
      "step": 48
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.930160111487716e-05,
      "loss": 1.0282,
      "step": 49
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.925593026621833e-05,
      "loss": 0.8123,
      "step": 50
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.920882422145372e-05,
      "loss": 0.5971,
      "step": 51
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.916028435313708e-05,
      "loss": 0.644,
      "step": 52
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.911031207560027e-05,
      "loss": 0.5748,
      "step": 53
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.905890884491195e-05,
      "loss": 1.1141,
      "step": 54
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.900607615883528e-05,
      "loss": 1.2696,
      "step": 55
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.895181555678418e-05,
      "loss": 1.0389,
      "step": 56
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.889612861977853e-05,
      "loss": 0.6419,
      "step": 57
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.883901697039808e-05,
      "loss": 0.6725,
      "step": 58
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.87804822727352e-05,
      "loss": 0.746,
      "step": 59
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.872052623234632e-05,
      "loss": 0.6614,
      "step": 60
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.865915059620232e-05,
      "loss": 0.5904,
      "step": 61
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.85963571526376e-05,
      "loss": 1.119,
      "step": 62
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.853214773129796e-05,
      "loss": 1.0983,
      "step": 63
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.846652420308728e-05,
      "loss": 1.093,
      "step": 64
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.839948848011304e-05,
      "loss": 0.7193,
      "step": 65
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.833104251563056e-05,
      "loss": 1.0884,
      "step": 66
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.826118830398615e-05,
      "loss": 1.0729,
      "step": 67
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.818992788055889e-05,
      "loss": 0.6625,
      "step": 68
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.811726332170153e-05,
      "loss": 0.9081,
      "step": 69
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.80431967446797e-05,
      "loss": 0.9791,
      "step": 70
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.796773030761047e-05,
      "loss": 0.868,
      "step": 71
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.789086620939936e-05,
      "loss": 0.8326,
      "step": 72
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.781260668967628e-05,
      "loss": 0.6327,
      "step": 73
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.773295402873026e-05,
      "loss": 0.9782,
      "step": 74
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.765191054744305e-05,
      "loss": 1.169,
      "step": 75
    },
    {
      "epoch": 0.76,
      "learning_rate": 9.756947860722143e-05,
      "loss": 0.524,
      "step": 76
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.748566060992847e-05,
      "loss": 0.9121,
      "step": 77
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.740045899781352e-05,
      "loss": 0.7816,
      "step": 78
    },
    {
      "epoch": 0.79,
      "learning_rate": 9.731387625344104e-05,
      "loss": 0.1256,
      "step": 79
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.722591489961827e-05,
      "loss": 1.1603,
      "step": 80
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.713657749932172e-05,
      "loss": 1.1672,
      "step": 81
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.70458666556225e-05,
      "loss": 0.7319,
      "step": 82
    },
    {
      "epoch": 0.83,
      "learning_rate": 9.695378501161045e-05,
      "loss": 0.8472,
      "step": 83
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.686033525031719e-05,
      "loss": 1.0569,
      "step": 84
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.676552009463783e-05,
      "loss": 0.7236,
      "step": 85
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.66693423072518e-05,
      "loss": 0.9983,
      "step": 86
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.657180469054213e-05,
      "loss": 0.5939,
      "step": 87
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.647291008651398e-05,
      "loss": 0.6493,
      "step": 88
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.637266137671177e-05,
      "loss": 0.5846,
      "step": 89
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.627106148213522e-05,
      "loss": 0.7209,
      "step": 90
    },
    {
      "epoch": 0.91,
      "learning_rate": 9.61681133631542e-05,
      "loss": 1.147,
      "step": 91
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.606382001942255e-05,
      "loss": 1.1926,
      "step": 92
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.595818448979061e-05,
      "loss": 0.6786,
      "step": 93
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.585120985221671e-05,
      "loss": 0.7093,
      "step": 94
    },
    {
      "epoch": 0.95,
      "learning_rate": 9.574289922367749e-05,
      "loss": 0.5315,
      "step": 95
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.563325576007701e-05,
      "loss": 0.743,
      "step": 96
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.552228265615492e-05,
      "loss": 0.9921,
      "step": 97
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.540998314539328e-05,
      "loss": 0.7962,
      "step": 98
    },
    {
      "epoch": 0.99,
      "learning_rate": 9.529636049992234e-05,
      "loss": 0.7813,
      "step": 99
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.518141803042527e-05,
      "loss": 1.0127,
      "step": 100
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.506515908604162e-05,
      "loss": 1.0064,
      "step": 101
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.494758705426978e-05,
      "loss": 1.1139,
      "step": 102
    },
    {
      "epoch": 1.03,
      "learning_rate": 9.482870536086823e-05,
      "loss": 0.5693,
      "step": 103
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.470851746975582e-05,
      "loss": 0.5606,
      "step": 104
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.458702688291073e-05,
      "loss": 0.5032,
      "step": 105
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.446423714026846e-05,
      "loss": 0.5778,
      "step": 106
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.434015181961873e-05,
      "loss": 0.6076,
      "step": 107
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.421477453650118e-05,
      "loss": 0.5808,
      "step": 108
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.408810894410009e-05,
      "loss": 0.5576,
      "step": 109
    },
    {
      "epoch": 1.1,
      "learning_rate": 9.396015873313781e-05,
      "loss": 1.0834,
      "step": 110
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.38309276317674e-05,
      "loss": 0.8611,
      "step": 111
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.37004194054638e-05,
      "loss": 0.9198,
      "step": 112
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.356863785691428e-05,
      "loss": 0.0983,
      "step": 113
    },
    {
      "epoch": 1.14,
      "learning_rate": 9.343558682590756e-05,
      "loss": 0.5039,
      "step": 114
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.330127018922194e-05,
      "loss": 0.5061,
      "step": 115
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.316569186051234e-05,
      "loss": 1.0256,
      "step": 116
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.302885579019627e-05,
      "loss": 0.9337,
      "step": 117
    },
    {
      "epoch": 1.18,
      "learning_rate": 9.289076596533872e-05,
      "loss": 0.5748,
      "step": 118
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.2751426409536e-05,
      "loss": 0.587,
      "step": 119
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.261084118279847e-05,
      "loss": 1.1003,
      "step": 120
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.24690143814323e-05,
      "loss": 0.6208,
      "step": 121
    },
    {
      "epoch": 1.22,
      "learning_rate": 9.232595013792002e-05,
      "loss": 0.1508,
      "step": 122
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.218165262080023e-05,
      "loss": 0.522,
      "step": 123
    },
    {
      "epoch": 1.24,
      "learning_rate": 9.203612603454604e-05,
      "loss": 1.036,
      "step": 124
    },
    {
      "epoch": 1.25,
      "learning_rate": 9.18893746194426e-05,
      "loss": 0.609,
      "step": 125
    },
    {
      "epoch": 1.26,
      "learning_rate": 9.174140265146356e-05,
      "loss": 1.0418,
      "step": 126
    },
    {
      "epoch": 1.27,
      "learning_rate": 9.159221444214645e-05,
      "loss": 0.4341,
      "step": 127
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.144181433846707e-05,
      "loss": 0.8735,
      "step": 128
    },
    {
      "epoch": 1.29,
      "learning_rate": 9.129020672271283e-05,
      "loss": 0.4226,
      "step": 129
    },
    {
      "epoch": 1.3,
      "learning_rate": 9.113739601235507e-05,
      "loss": 0.9683,
      "step": 130
    },
    {
      "epoch": 1.31,
      "learning_rate": 9.09833866599203e-05,
      "loss": 0.5853,
      "step": 131
    },
    {
      "epoch": 1.32,
      "learning_rate": 9.082818315286055e-05,
      "loss": 0.3069,
      "step": 132
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.067179001342252e-05,
      "loss": 0.6285,
      "step": 133
    },
    {
      "epoch": 1.34,
      "learning_rate": 9.051421179851588e-05,
      "loss": 0.592,
      "step": 134
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.035545309958046e-05,
      "loss": 1.1428,
      "step": 135
    },
    {
      "epoch": 1.36,
      "learning_rate": 9.01955185424525e-05,
      "loss": 0.407,
      "step": 136
    },
    {
      "epoch": 1.37,
      "learning_rate": 9.003441278722981e-05,
      "loss": 0.4145,
      "step": 137
    },
    {
      "epoch": 1.38,
      "learning_rate": 8.987214052813604e-05,
      "loss": 1.0347,
      "step": 138
    },
    {
      "epoch": 1.39,
      "learning_rate": 8.970870649338387e-05,
      "loss": 1.2144,
      "step": 139
    },
    {
      "epoch": 1.4,
      "learning_rate": 8.954411544503729e-05,
      "loss": 1.0854,
      "step": 140
    },
    {
      "epoch": 1.41,
      "learning_rate": 8.937837217887273e-05,
      "loss": 0.5885,
      "step": 141
    },
    {
      "epoch": 1.42,
      "learning_rate": 8.921148152423946e-05,
      "loss": 1.0299,
      "step": 142
    },
    {
      "epoch": 1.43,
      "learning_rate": 8.904344834391882e-05,
      "loss": 1.1687,
      "step": 143
    },
    {
      "epoch": 1.44,
      "learning_rate": 8.887427753398248e-05,
      "loss": 1.1054,
      "step": 144
    },
    {
      "epoch": 1.45,
      "learning_rate": 8.870397402364984e-05,
      "loss": 0.9101,
      "step": 145
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.853254277514446e-05,
      "loss": 0.5874,
      "step": 146
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.835998878354931e-05,
      "loss": 0.4957,
      "step": 147
    },
    {
      "epoch": 1.48,
      "learning_rate": 8.818631707666135e-05,
      "loss": 0.5342,
      "step": 148
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.801153271484502e-05,
      "loss": 1.0376,
      "step": 149
    },
    {
      "epoch": 1.5,
      "learning_rate": 8.783564079088477e-05,
      "loss": 0.8963,
      "step": 150
    },
    {
      "epoch": 1.51,
      "learning_rate": 8.765864642983665e-05,
      "loss": 0.7895,
      "step": 151
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.748055478887904e-05,
      "loss": 0.9246,
      "step": 152
    },
    {
      "epoch": 1.53,
      "learning_rate": 8.73013710571623e-05,
      "loss": 1.1028,
      "step": 153
    },
    {
      "epoch": 1.54,
      "learning_rate": 8.712110045565768e-05,
      "loss": 0.896,
      "step": 154
    },
    {
      "epoch": 1.55,
      "learning_rate": 8.693974823700506e-05,
      "loss": 0.9723,
      "step": 155
    },
    {
      "epoch": 1.56,
      "learning_rate": 8.675731968536002e-05,
      "loss": 1.2289,
      "step": 156
    },
    {
      "epoch": 1.57,
      "learning_rate": 8.657382011623981e-05,
      "loss": 1.1148,
      "step": 157
    },
    {
      "epoch": 1.58,
      "learning_rate": 8.638925487636848e-05,
      "loss": 0.6829,
      "step": 158
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.620362934352109e-05,
      "loss": 0.9895,
      "step": 159
    },
    {
      "epoch": 1.6,
      "learning_rate": 8.6016948926367e-05,
      "loss": 0.5653,
      "step": 160
    },
    {
      "epoch": 1.61,
      "learning_rate": 8.582921906431237e-05,
      "loss": 0.8259,
      "step": 161
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.564044522734147e-05,
      "loss": 0.7777,
      "step": 162
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.545063291585752e-05,
      "loss": 0.9909,
      "step": 163
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.52597876605223e-05,
      "loss": 0.4626,
      "step": 164
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.506791502209496e-05,
      "loss": 1.051,
      "step": 165
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.487502059127015e-05,
      "loss": 1.0609,
      "step": 166
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.468110998851496e-05,
      "loss": 0.6606,
      "step": 167
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.448618886390522e-05,
      "loss": 0.7334,
      "step": 168
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.429026289696091e-05,
      "loss": 1.0334,
      "step": 169
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.40933377964806e-05,
      "loss": 0.4235,
      "step": 170
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.389541930037516e-05,
      "loss": 1.0154,
      "step": 171
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.369651317550054e-05,
      "loss": 0.165,
      "step": 172
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.349662521748977e-05,
      "loss": 0.9675,
      "step": 173
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.329576125058406e-05,
      "loss": 0.351,
      "step": 174
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.309392712746308e-05,
      "loss": 0.8936,
      "step": 175
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.289112872907454e-05,
      "loss": 0.4929,
      "step": 176
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.268737196446264e-05,
      "loss": 0.8426,
      "step": 177
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.248266277059607e-05,
      "loss": 0.8089,
      "step": 178
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.227700711219493e-05,
      "loss": 0.4637,
      "step": 179
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.2070410981557e-05,
      "loss": 0.1139,
      "step": 180
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.186288039838304e-05,
      "loss": 0.0912,
      "step": 181
    },
    {
      "epoch": 1.82,
      "learning_rate": 8.16544214096015e-05,
      "loss": 1.0137,
      "step": 182
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.144504008919222e-05,
      "loss": 0.721,
      "step": 183
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.123474253800957e-05,
      "loss": 0.6093,
      "step": 184
    },
    {
      "epoch": 1.85,
      "learning_rate": 8.102353488360454e-05,
      "loss": 1.1467,
      "step": 185
    },
    {
      "epoch": 1.86,
      "learning_rate": 8.081142328004637e-05,
      "loss": 1.042,
      "step": 186
    },
    {
      "epoch": 1.87,
      "learning_rate": 8.059841390774307e-05,
      "loss": 0.5714,
      "step": 187
    },
    {
      "epoch": 1.88,
      "learning_rate": 8.038451297326145e-05,
      "loss": 0.6604,
      "step": 188
    },
    {
      "epoch": 1.89,
      "learning_rate": 8.016972670914624e-05,
      "loss": 1.0334,
      "step": 189
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.995406137373846e-05,
      "loss": 0.9352,
      "step": 190
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.973752325099314e-05,
      "loss": 0.3907,
      "step": 191
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.952011865029614e-05,
      "loss": 0.3872,
      "step": 192
    },
    {
      "epoch": 1.93,
      "learning_rate": 7.930185390628035e-05,
      "loss": 0.5052,
      "step": 193
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.908273537864113e-05,
      "loss": 0.9008,
      "step": 194
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.886276945195099e-05,
      "loss": 0.1056,
      "step": 195
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.86419625354735e-05,
      "loss": 0.8095,
      "step": 196
    },
    {
      "epoch": 1.97,
      "learning_rate": 7.842032106297666e-05,
      "loss": 1.1502,
      "step": 197
    },
    {
      "epoch": 1.98,
      "learning_rate": 7.819785149254532e-05,
      "loss": 1.1409,
      "step": 198
    },
    {
      "epoch": 1.99,
      "learning_rate": 7.797456030639313e-05,
      "loss": 0.0934,
      "step": 199
    },
    {
      "epoch": 2.0,
      "learning_rate": 7.77504540106735e-05,
      "loss": 0.9729,
      "step": 200
    },
    {
      "epoch": 2.01,
      "learning_rate": 7.752553913529018e-05,
      "loss": 0.6731,
      "step": 201
    },
    {
      "epoch": 2.02,
      "learning_rate": 7.729982223370691e-05,
      "loss": 0.6876,
      "step": 202
    },
    {
      "epoch": 2.03,
      "learning_rate": 7.707330988275651e-05,
      "loss": 0.8572,
      "step": 203
    },
    {
      "epoch": 2.04,
      "learning_rate": 7.68460086824492e-05,
      "loss": 0.7632,
      "step": 204
    },
    {
      "epoch": 2.05,
      "learning_rate": 7.661792525578035e-05,
      "loss": 0.8381,
      "step": 205
    },
    {
      "epoch": 2.06,
      "learning_rate": 7.638906624853743e-05,
      "loss": 0.1438,
      "step": 206
    },
    {
      "epoch": 2.07,
      "learning_rate": 7.61594383291065e-05,
      "loss": 0.9388,
      "step": 207
    },
    {
      "epoch": 2.08,
      "learning_rate": 7.592904818827775e-05,
      "loss": 0.7623,
      "step": 208
    },
    {
      "epoch": 2.09,
      "learning_rate": 7.569790253905059e-05,
      "loss": 0.7193,
      "step": 209
    },
    {
      "epoch": 2.1,
      "learning_rate": 7.546600811643816e-05,
      "loss": 0.4692,
      "step": 210
    },
    {
      "epoch": 2.11,
      "learning_rate": 7.523337167727095e-05,
      "loss": 0.852,
      "step": 211
    },
    {
      "epoch": 2.12,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.4806,
      "step": 212
    },
    {
      "epoch": 2.13,
      "learning_rate": 7.476589988449939e-05,
      "loss": 0.9571,
      "step": 213
    },
    {
      "epoch": 2.14,
      "learning_rate": 7.453107815186803e-05,
      "loss": 0.8643,
      "step": 214
    },
    {
      "epoch": 2.15,
      "learning_rate": 7.429554164423102e-05,
      "loss": 0.1021,
      "step": 215
    },
    {
      "epoch": 2.16,
      "learning_rate": 7.405929722454026e-05,
      "loss": 0.8053,
      "step": 216
    },
    {
      "epoch": 2.17,
      "learning_rate": 7.382235177637437e-05,
      "loss": 0.7002,
      "step": 217
    },
    {
      "epoch": 2.18,
      "learning_rate": 7.358471220373832e-05,
      "loss": 0.9525,
      "step": 218
    },
    {
      "epoch": 2.19,
      "learning_rate": 7.334638543086203e-05,
      "loss": 0.5114,
      "step": 219
    },
    {
      "epoch": 2.2,
      "learning_rate": 7.310737840199885e-05,
      "loss": 1.1477,
      "step": 220
    },
    {
      "epoch": 2.21,
      "learning_rate": 7.286769808122304e-05,
      "loss": 0.9718,
      "step": 221
    },
    {
      "epoch": 2.22,
      "learning_rate": 7.262735145222696e-05,
      "loss": 0.4707,
      "step": 222
    },
    {
      "epoch": 2.23,
      "learning_rate": 7.238634551811749e-05,
      "loss": 0.3755,
      "step": 223
    },
    {
      "epoch": 2.24,
      "learning_rate": 7.214468730121208e-05,
      "loss": 0.4897,
      "step": 224
    },
    {
      "epoch": 2.25,
      "learning_rate": 7.190238384283412e-05,
      "loss": 1.0034,
      "step": 225
    },
    {
      "epoch": 2.26,
      "learning_rate": 7.165944220310767e-05,
      "loss": 1.0291,
      "step": 226
    },
    {
      "epoch": 2.27,
      "learning_rate": 7.141586946075183e-05,
      "loss": 0.7632,
      "step": 227
    },
    {
      "epoch": 2.28,
      "learning_rate": 7.117167271287453e-05,
      "loss": 0.4562,
      "step": 228
    },
    {
      "epoch": 2.29,
      "learning_rate": 7.092685907476558e-05,
      "loss": 0.9181,
      "step": 229
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.068143567968957e-05,
      "loss": 0.1491,
      "step": 230
    },
    {
      "epoch": 2.31,
      "learning_rate": 7.043540967867782e-05,
      "loss": 0.5085,
      "step": 231
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.018878824032009e-05,
      "loss": 0.4929,
      "step": 232
    },
    {
      "epoch": 2.33,
      "learning_rate": 6.994157855055576e-05,
      "loss": 0.9621,
      "step": 233
    },
    {
      "epoch": 2.34,
      "learning_rate": 6.969378781246436e-05,
      "loss": 0.4147,
      "step": 234
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.944542324605578e-05,
      "loss": 0.9537,
      "step": 235
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.919649208805981e-05,
      "loss": 0.3851,
      "step": 236
    },
    {
      "epoch": 2.37,
      "learning_rate": 6.894700159171534e-05,
      "loss": 0.6313,
      "step": 237
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.869695902655897e-05,
      "loss": 1.0456,
      "step": 238
    },
    {
      "epoch": 2.39,
      "learning_rate": 6.844637167821326e-05,
      "loss": 0.8106,
      "step": 239
    },
    {
      "epoch": 2.4,
      "learning_rate": 6.819524684817438e-05,
      "loss": 0.6425,
      "step": 240
    },
    {
      "epoch": 2.41,
      "learning_rate": 6.794359185359938e-05,
      "loss": 0.5893,
      "step": 241
    },
    {
      "epoch": 2.42,
      "learning_rate": 6.769141402709305e-05,
      "loss": 0.3764,
      "step": 242
    },
    {
      "epoch": 2.43,
      "learning_rate": 6.743872071649411e-05,
      "loss": 0.8726,
      "step": 243
    },
    {
      "epoch": 2.44,
      "learning_rate": 6.718551928466132e-05,
      "loss": 0.4516,
      "step": 244
    },
    {
      "epoch": 2.45,
      "learning_rate": 6.693181710925878e-05,
      "loss": 0.1484,
      "step": 245
    },
    {
      "epoch": 2.46,
      "learning_rate": 6.667762158254104e-05,
      "loss": 0.4944,
      "step": 246
    },
    {
      "epoch": 2.47,
      "learning_rate": 6.642294011113764e-05,
      "loss": 0.5293,
      "step": 247
    },
    {
      "epoch": 2.48,
      "learning_rate": 6.616778011583743e-05,
      "loss": 0.8189,
      "step": 248
    },
    {
      "epoch": 2.49,
      "learning_rate": 6.59121490313722e-05,
      "loss": 0.3628,
      "step": 249
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.565605430620013e-05,
      "loss": 0.8436,
      "step": 250
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.539950340228877e-05,
      "loss": 0.9848,
      "step": 251
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.514250379489753e-05,
      "loss": 0.4029,
      "step": 252
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.488506297236003e-05,
      "loss": 0.5044,
      "step": 253
    },
    {
      "epoch": 2.54,
      "learning_rate": 6.462718843586571e-05,
      "loss": 0.0983,
      "step": 254
    },
    {
      "epoch": 2.55,
      "learning_rate": 6.436888769924142e-05,
      "loss": 0.4742,
      "step": 255
    },
    {
      "epoch": 2.56,
      "learning_rate": 6.411016828873239e-05,
      "loss": 0.8615,
      "step": 256
    },
    {
      "epoch": 2.57,
      "learning_rate": 6.385103774278303e-05,
      "loss": 1.0037,
      "step": 257
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.359150361181715e-05,
      "loss": 0.3709,
      "step": 258
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.333157345801809e-05,
      "loss": 0.7423,
      "step": 259
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.307125485510828e-05,
      "loss": 0.5818,
      "step": 260
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.281055538812861e-05,
      "loss": 0.4394,
      "step": 261
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.254948265321744e-05,
      "loss": 0.8806,
      "step": 262
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.228804425738917e-05,
      "loss": 0.6338,
      "step": 263
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.202624781831268e-05,
      "loss": 0.9208,
      "step": 264
    },
    {
      "epoch": 2.65,
      "learning_rate": 6.176410096408938e-05,
      "loss": 0.6143,
      "step": 265
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.150161133303089e-05,
      "loss": 0.665,
      "step": 266
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.123878657343648e-05,
      "loss": 0.1202,
      "step": 267
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.0975634343370256e-05,
      "loss": 0.7635,
      "step": 268
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.071216231043799e-05,
      "loss": 0.3946,
      "step": 269
    },
    {
      "epoch": 2.7,
      "learning_rate": 6.044837815156377e-05,
      "loss": 0.9457,
      "step": 270
    },
    {
      "epoch": 2.71,
      "learning_rate": 6.018428955276617e-05,
      "loss": 0.4996,
      "step": 271
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.99199042089345e-05,
      "loss": 0.0955,
      "step": 272
    },
    {
      "epoch": 2.73,
      "learning_rate": 5.9655229823604406e-05,
      "loss": 1.0307,
      "step": 273
    },
    {
      "epoch": 2.74,
      "learning_rate": 5.939027410873351e-05,
      "loss": 0.397,
      "step": 274
    },
    {
      "epoch": 2.75,
      "learning_rate": 5.912504478447669e-05,
      "loss": 0.9636,
      "step": 275
    },
    {
      "epoch": 2.76,
      "learning_rate": 5.885954957896115e-05,
      "loss": 0.4741,
      "step": 276
    },
    {
      "epoch": 2.77,
      "learning_rate": 5.859379622806115e-05,
      "loss": 0.8797,
      "step": 277
    },
    {
      "epoch": 2.78,
      "learning_rate": 5.832779247517273e-05,
      "loss": 0.7577,
      "step": 278
    },
    {
      "epoch": 2.79,
      "learning_rate": 5.8061546070987994e-05,
      "loss": 0.7404,
      "step": 279
    },
    {
      "epoch": 2.8,
      "learning_rate": 5.779506477326933e-05,
      "loss": 0.8952,
      "step": 280
    },
    {
      "epoch": 2.81,
      "learning_rate": 5.752835634662331e-05,
      "loss": 0.7155,
      "step": 281
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.726142856227452e-05,
      "loss": 0.9328,
      "step": 282
    },
    {
      "epoch": 2.83,
      "learning_rate": 5.699428919783906e-05,
      "loss": 0.9198,
      "step": 283
    },
    {
      "epoch": 2.84,
      "learning_rate": 5.672694603709794e-05,
      "loss": 0.9023,
      "step": 284
    },
    {
      "epoch": 2.85,
      "learning_rate": 5.645940686977033e-05,
      "loss": 0.4921,
      "step": 285
    },
    {
      "epoch": 2.86,
      "learning_rate": 5.619167949128652e-05,
      "loss": 0.5109,
      "step": 286
    },
    {
      "epoch": 2.87,
      "learning_rate": 5.59237717025608e-05,
      "loss": 0.6515,
      "step": 287
    },
    {
      "epoch": 2.88,
      "learning_rate": 5.565569130976422e-05,
      "loss": 0.8769,
      "step": 288
    },
    {
      "epoch": 2.89,
      "learning_rate": 5.538744612409701e-05,
      "loss": 0.718,
      "step": 289
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.5119043961561136e-05,
      "loss": 0.9196,
      "step": 290
    },
    {
      "epoch": 2.91,
      "learning_rate": 5.4850492642732406e-05,
      "loss": 0.1985,
      "step": 291
    },
    {
      "epoch": 2.92,
      "learning_rate": 5.458179999253275e-05,
      "loss": 0.9993,
      "step": 292
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.4312973840002045e-05,
      "loss": 0.789,
      "step": 293
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.4044022018070214e-05,
      "loss": 0.2474,
      "step": 294
    },
    {
      "epoch": 2.95,
      "learning_rate": 5.37749523633288e-05,
      "loss": 0.1283,
      "step": 295
    },
    {
      "epoch": 2.96,
      "learning_rate": 5.3505772715802704e-05,
      "loss": 0.2069,
      "step": 296
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.3236490918721794e-05,
      "loss": 0.7092,
      "step": 297
    },
    {
      "epoch": 2.98,
      "learning_rate": 5.296711481829226e-05,
      "loss": 0.4774,
      "step": 298
    },
    {
      "epoch": 2.99,
      "learning_rate": 5.2697652263468125e-05,
      "loss": 0.6437,
      "step": 299
    },
    {
      "epoch": 3.0,
      "learning_rate": 5.242811110572242e-05,
      "loss": 0.8045,
      "step": 300
    },
    {
      "epoch": 3.01,
      "learning_rate": 5.2158499198818503e-05,
      "loss": 0.6945,
      "step": 301
    },
    {
      "epoch": 3.02,
      "learning_rate": 5.188882439858117e-05,
      "loss": 0.7385,
      "step": 302
    },
    {
      "epoch": 3.03,
      "learning_rate": 5.1619094562667804e-05,
      "loss": 0.455,
      "step": 303
    },
    {
      "epoch": 3.04,
      "learning_rate": 5.134931755033936e-05,
      "loss": 0.5975,
      "step": 304
    },
    {
      "epoch": 3.05,
      "learning_rate": 5.107950122223139e-05,
      "loss": 0.1079,
      "step": 305
    },
    {
      "epoch": 3.06,
      "learning_rate": 5.080965344012508e-05,
      "loss": 0.888,
      "step": 306
    },
    {
      "epoch": 3.07,
      "learning_rate": 5.053978206671801e-05,
      "loss": 0.4048,
      "step": 307
    },
    {
      "epoch": 3.08,
      "learning_rate": 5.0269894965395225e-05,
      "loss": 0.2892,
      "step": 308
    },
    {
      "epoch": 3.09,
      "learning_rate": 5e-05,
      "loss": 0.7617,
      "step": 309
    },
    {
      "epoch": 3.1,
      "learning_rate": 4.973010503460479e-05,
      "loss": 0.507,
      "step": 310
    },
    {
      "epoch": 3.11,
      "learning_rate": 4.946021793328201e-05,
      "loss": 0.8717,
      "step": 311
    },
    {
      "epoch": 3.12,
      "learning_rate": 4.919034655987493e-05,
      "loss": 0.7561,
      "step": 312
    },
    {
      "epoch": 3.13,
      "learning_rate": 4.892049877776861e-05,
      "loss": 0.5775,
      "step": 313
    },
    {
      "epoch": 3.14,
      "learning_rate": 4.865068244966066e-05,
      "loss": 0.4048,
      "step": 314
    },
    {
      "epoch": 3.15,
      "learning_rate": 4.838090543733222e-05,
      "loss": 0.771,
      "step": 315
    },
    {
      "epoch": 3.16,
      "learning_rate": 4.8111175601418844e-05,
      "loss": 0.4579,
      "step": 316
    },
    {
      "epoch": 3.17,
      "learning_rate": 4.784150080118152e-05,
      "loss": 0.5625,
      "step": 317
    },
    {
      "epoch": 3.18,
      "learning_rate": 4.7571888894277604e-05,
      "loss": 0.4475,
      "step": 318
    },
    {
      "epoch": 3.19,
      "learning_rate": 4.730234773653188e-05,
      "loss": 0.1464,
      "step": 319
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.703288518170774e-05,
      "loss": 0.3964,
      "step": 320
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.676350908127822e-05,
      "loss": 0.4403,
      "step": 321
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.6494227284197294e-05,
      "loss": 0.6092,
      "step": 322
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.622504763667122e-05,
      "loss": 0.3271,
      "step": 323
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.59559779819298e-05,
      "loss": 0.647,
      "step": 324
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.568702615999797e-05,
      "loss": 0.582,
      "step": 325
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.541820000746727e-05,
      "loss": 0.404,
      "step": 326
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.51495073572676e-05,
      "loss": 0.4076,
      "step": 327
    },
    {
      "epoch": 3.28,
      "learning_rate": 4.4880956038438876e-05,
      "loss": 0.5281,
      "step": 328
    },
    {
      "epoch": 3.29,
      "learning_rate": 4.461255387590299e-05,
      "loss": 0.5991,
      "step": 329
    },
    {
      "epoch": 3.3,
      "learning_rate": 4.434430869023579e-05,
      "loss": 0.5625,
      "step": 330
    },
    {
      "epoch": 3.31,
      "learning_rate": 4.4076228297439204e-05,
      "loss": 0.4749,
      "step": 331
    },
    {
      "epoch": 3.32,
      "learning_rate": 4.38083205087135e-05,
      "loss": 0.8395,
      "step": 332
    },
    {
      "epoch": 3.33,
      "learning_rate": 4.35405931302297e-05,
      "loss": 0.1073,
      "step": 333
    },
    {
      "epoch": 3.34,
      "learning_rate": 4.3273053962902076e-05,
      "loss": 0.7115,
      "step": 334
    },
    {
      "epoch": 3.35,
      "learning_rate": 4.3005710802160965e-05,
      "loss": 0.7178,
      "step": 335
    },
    {
      "epoch": 3.36,
      "learning_rate": 4.27385714377255e-05,
      "loss": 0.4655,
      "step": 336
    },
    {
      "epoch": 3.37,
      "learning_rate": 4.2471643653376685e-05,
      "loss": 0.3833,
      "step": 337
    },
    {
      "epoch": 3.38,
      "learning_rate": 4.220493522673067e-05,
      "loss": 0.8855,
      "step": 338
    },
    {
      "epoch": 3.39,
      "learning_rate": 4.193845392901201e-05,
      "loss": 0.2677,
      "step": 339
    },
    {
      "epoch": 3.4,
      "learning_rate": 4.1672207524827275e-05,
      "loss": 0.7882,
      "step": 340
    },
    {
      "epoch": 3.41,
      "learning_rate": 4.140620377193885e-05,
      "loss": 0.5953,
      "step": 341
    },
    {
      "epoch": 3.42,
      "learning_rate": 4.114045042103887e-05,
      "loss": 0.8057,
      "step": 342
    },
    {
      "epoch": 3.43,
      "learning_rate": 4.087495521552331e-05,
      "loss": 0.9334,
      "step": 343
    },
    {
      "epoch": 3.44,
      "learning_rate": 4.06097258912665e-05,
      "loss": 0.5721,
      "step": 344
    },
    {
      "epoch": 3.45,
      "learning_rate": 4.0344770176395606e-05,
      "loss": 0.7482,
      "step": 345
    },
    {
      "epoch": 3.46,
      "learning_rate": 4.0080095791065505e-05,
      "loss": 0.3746,
      "step": 346
    },
    {
      "epoch": 3.47,
      "learning_rate": 3.9815710447233836e-05,
      "loss": 0.6298,
      "step": 347
    },
    {
      "epoch": 3.48,
      "learning_rate": 3.955162184843625e-05,
      "loss": 0.6947,
      "step": 348
    },
    {
      "epoch": 3.49,
      "learning_rate": 3.9287837689562016e-05,
      "loss": 0.6928,
      "step": 349
    },
    {
      "epoch": 3.5,
      "learning_rate": 3.902436565662977e-05,
      "loss": 0.2789,
      "step": 350
    },
    {
      "epoch": 3.51,
      "learning_rate": 3.876121342656355e-05,
      "loss": 0.326,
      "step": 351
    },
    {
      "epoch": 3.52,
      "learning_rate": 3.849838866696913e-05,
      "loss": 0.7283,
      "step": 352
    },
    {
      "epoch": 3.53,
      "learning_rate": 3.823589903591063e-05,
      "loss": 0.3768,
      "step": 353
    },
    {
      "epoch": 3.54,
      "learning_rate": 3.7973752181687335e-05,
      "loss": 0.6908,
      "step": 354
    },
    {
      "epoch": 3.55,
      "learning_rate": 3.771195574261084e-05,
      "loss": 0.655,
      "step": 355
    },
    {
      "epoch": 3.56,
      "learning_rate": 3.745051734678256e-05,
      "loss": 0.665,
      "step": 356
    },
    {
      "epoch": 3.57,
      "learning_rate": 3.718944461187138e-05,
      "loss": 0.7143,
      "step": 357
    },
    {
      "epoch": 3.58,
      "learning_rate": 3.692874514489173e-05,
      "loss": 0.4508,
      "step": 358
    },
    {
      "epoch": 3.59,
      "learning_rate": 3.666842654198191e-05,
      "loss": 0.7157,
      "step": 359
    },
    {
      "epoch": 3.6,
      "learning_rate": 3.640849638818286e-05,
      "loss": 0.5367,
      "step": 360
    },
    {
      "epoch": 3.61,
      "learning_rate": 3.614896225721699e-05,
      "loss": 0.5101,
      "step": 361
    },
    {
      "epoch": 3.62,
      "learning_rate": 3.588983171126762e-05,
      "loss": 0.6966,
      "step": 362
    },
    {
      "epoch": 3.63,
      "learning_rate": 3.5631112300758595e-05,
      "loss": 0.9672,
      "step": 363
    },
    {
      "epoch": 3.64,
      "learning_rate": 3.53728115641343e-05,
      "loss": 0.1973,
      "step": 364
    },
    {
      "epoch": 3.65,
      "learning_rate": 3.5114937027639985e-05,
      "loss": 0.769,
      "step": 365
    },
    {
      "epoch": 3.66,
      "learning_rate": 3.4857496205102474e-05,
      "loss": 0.9575,
      "step": 366
    },
    {
      "epoch": 3.67,
      "learning_rate": 3.460049659771124e-05,
      "loss": 0.0895,
      "step": 367
    },
    {
      "epoch": 3.68,
      "learning_rate": 3.434394569379988e-05,
      "loss": 0.5265,
      "step": 368
    },
    {
      "epoch": 3.69,
      "learning_rate": 3.408785096862782e-05,
      "loss": 0.9364,
      "step": 369
    },
    {
      "epoch": 3.7,
      "learning_rate": 3.3832219884162585e-05,
      "loss": 0.4791,
      "step": 370
    },
    {
      "epoch": 3.71,
      "learning_rate": 3.3577059888862364e-05,
      "loss": 0.5245,
      "step": 371
    },
    {
      "epoch": 3.72,
      "learning_rate": 3.332237841745898e-05,
      "loss": 0.8515,
      "step": 372
    },
    {
      "epoch": 3.73,
      "learning_rate": 3.30681828907412e-05,
      "loss": 0.8348,
      "step": 373
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.281448071533867e-05,
      "loss": 0.3995,
      "step": 374
    },
    {
      "epoch": 3.75,
      "learning_rate": 3.2561279283505883e-05,
      "loss": 0.3216,
      "step": 375
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.2308585972906966e-05,
      "loss": 0.567,
      "step": 376
    },
    {
      "epoch": 3.77,
      "learning_rate": 3.2056408146400614e-05,
      "loss": 0.7622,
      "step": 377
    },
    {
      "epoch": 3.78,
      "learning_rate": 3.180475315182563e-05,
      "loss": 0.7857,
      "step": 378
    },
    {
      "epoch": 3.79,
      "learning_rate": 3.1553628321786745e-05,
      "loss": 0.4483,
      "step": 379
    },
    {
      "epoch": 3.8,
      "learning_rate": 3.130304097344103e-05,
      "loss": 0.3763,
      "step": 380
    },
    {
      "epoch": 3.81,
      "learning_rate": 3.105299840828466e-05,
      "loss": 0.3648,
      "step": 381
    },
    {
      "epoch": 3.82,
      "learning_rate": 3.080350791194019e-05,
      "loss": 0.3112,
      "step": 382
    },
    {
      "epoch": 3.83,
      "learning_rate": 3.055457675394423e-05,
      "loss": 0.0819,
      "step": 383
    },
    {
      "epoch": 3.84,
      "learning_rate": 3.0306212187535653e-05,
      "loss": 0.268,
      "step": 384
    },
    {
      "epoch": 3.85,
      "learning_rate": 3.005842144944425e-05,
      "loss": 0.7788,
      "step": 385
    },
    {
      "epoch": 3.86,
      "learning_rate": 2.9811211759679924e-05,
      "loss": 0.7045,
      "step": 386
    },
    {
      "epoch": 3.87,
      "learning_rate": 2.9564590321322207e-05,
      "loss": 0.4414,
      "step": 387
    },
    {
      "epoch": 3.88,
      "learning_rate": 2.9318564320310444e-05,
      "loss": 0.6668,
      "step": 388
    },
    {
      "epoch": 3.89,
      "learning_rate": 2.907314092523442e-05,
      "loss": 0.481,
      "step": 389
    },
    {
      "epoch": 3.9,
      "learning_rate": 2.882832728712551e-05,
      "loss": 0.3447,
      "step": 390
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.8584130539248166e-05,
      "loss": 0.0927,
      "step": 391
    },
    {
      "epoch": 3.92,
      "learning_rate": 2.8340557796892354e-05,
      "loss": 0.8,
      "step": 392
    },
    {
      "epoch": 3.93,
      "learning_rate": 2.8097616157165883e-05,
      "loss": 0.3228,
      "step": 393
    },
    {
      "epoch": 3.94,
      "learning_rate": 2.7855312698787904e-05,
      "loss": 0.8464,
      "step": 394
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.761365448188253e-05,
      "loss": 0.0815,
      "step": 395
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.737264854777306e-05,
      "loss": 0.396,
      "step": 396
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.7132301918776977e-05,
      "loss": 0.883,
      "step": 397
    },
    {
      "epoch": 3.98,
      "learning_rate": 2.6892621598001156e-05,
      "loss": 0.2582,
      "step": 398
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.6653614569137968e-05,
      "loss": 0.728,
      "step": 399
    },
    {
      "epoch": 4.0,
      "learning_rate": 2.6415287796261706e-05,
      "loss": 0.7127,
      "step": 400
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.617764822362563e-05,
      "loss": 0.3259,
      "step": 401
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.5940702775459747e-05,
      "loss": 0.3665,
      "step": 402
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.5704458355768986e-05,
      "loss": 0.4499,
      "step": 403
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.5468921848131983e-05,
      "loss": 0.6581,
      "step": 404
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.5234100115500643e-05,
      "loss": 0.4261,
      "step": 405
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.500000000000001e-05,
      "loss": 0.7749,
      "step": 406
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.4766628322729064e-05,
      "loss": 0.6572,
      "step": 407
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.4533991883561868e-05,
      "loss": 0.3524,
      "step": 408
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.430209746094943e-05,
      "loss": 0.9622,
      "step": 409
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.407095181172227e-05,
      "loss": 0.604,
      "step": 410
    },
    {
      "epoch": 4.11,
      "learning_rate": 2.3840561670893496e-05,
      "loss": 0.3321,
      "step": 411
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.3610933751462553e-05,
      "loss": 0.6422,
      "step": 412
    },
    {
      "epoch": 4.13,
      "learning_rate": 2.3382074744219668e-05,
      "loss": 0.4829,
      "step": 413
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.315399131755081e-05,
      "loss": 0.3551,
      "step": 414
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.292669011724351e-05,
      "loss": 0.6241,
      "step": 415
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.2700177766293096e-05,
      "loss": 0.5875,
      "step": 416
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.2474460864709824e-05,
      "loss": 0.3197,
      "step": 417
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.2249545989326514e-05,
      "loss": 0.3756,
      "step": 418
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.2025439693606882e-05,
      "loss": 0.534,
      "step": 419
    },
    {
      "epoch": 4.2,
      "learning_rate": 2.180214850745467e-05,
      "loss": 0.5832,
      "step": 420
    },
    {
      "epoch": 4.21,
      "learning_rate": 2.1579678937023363e-05,
      "loss": 0.5484,
      "step": 421
    },
    {
      "epoch": 4.22,
      "learning_rate": 2.1358037464526515e-05,
      "loss": 0.7997,
      "step": 422
    },
    {
      "epoch": 4.23,
      "learning_rate": 2.1137230548049043e-05,
      "loss": 0.7011,
      "step": 423
    },
    {
      "epoch": 4.24,
      "learning_rate": 2.091726462135888e-05,
      "loss": 0.7455,
      "step": 424
    },
    {
      "epoch": 4.25,
      "learning_rate": 2.0698146093719656e-05,
      "loss": 0.3987,
      "step": 425
    },
    {
      "epoch": 4.26,
      "learning_rate": 2.0479881349703883e-05,
      "loss": 0.545,
      "step": 426
    },
    {
      "epoch": 4.27,
      "learning_rate": 2.0262476749006877e-05,
      "loss": 0.8259,
      "step": 427
    },
    {
      "epoch": 4.28,
      "learning_rate": 2.0045938626261546e-05,
      "loss": 0.3419,
      "step": 428
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.983027329085377e-05,
      "loss": 0.4308,
      "step": 429
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.9615487026738543e-05,
      "loss": 0.6206,
      "step": 430
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.940158609225694e-05,
      "loss": 0.4265,
      "step": 431
    },
    {
      "epoch": 4.32,
      "learning_rate": 1.9188576719953633e-05,
      "loss": 0.6544,
      "step": 432
    },
    {
      "epoch": 4.33,
      "learning_rate": 1.8976465116395464e-05,
      "loss": 0.5376,
      "step": 433
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.8765257461990442e-05,
      "loss": 0.2594,
      "step": 434
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.8554959910807775e-05,
      "loss": 0.2927,
      "step": 435
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.834557859039851e-05,
      "loss": 0.3391,
      "step": 436
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.813711960161696e-05,
      "loss": 0.6223,
      "step": 437
    },
    {
      "epoch": 4.38,
      "learning_rate": 1.7929589018443016e-05,
      "loss": 0.5188,
      "step": 438
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.772299288780508e-05,
      "loss": 0.3451,
      "step": 439
    },
    {
      "epoch": 4.4,
      "learning_rate": 1.7517337229403946e-05,
      "loss": 0.7354,
      "step": 440
    },
    {
      "epoch": 4.41,
      "learning_rate": 1.7312628035537387e-05,
      "loss": 0.3843,
      "step": 441
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.710887127092548e-05,
      "loss": 0.6396,
      "step": 442
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.6906072872536917e-05,
      "loss": 0.6972,
      "step": 443
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.6704238749415957e-05,
      "loss": 0.4035,
      "step": 444
    },
    {
      "epoch": 4.45,
      "learning_rate": 1.6503374782510234e-05,
      "loss": 0.1665,
      "step": 445
    },
    {
      "epoch": 4.46,
      "learning_rate": 1.6303486824499458e-05,
      "loss": 0.6172,
      "step": 446
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.6104580699624837e-05,
      "loss": 0.2412,
      "step": 447
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.5906662203519412e-05,
      "loss": 0.505,
      "step": 448
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.5709737103039103e-05,
      "loss": 0.4971,
      "step": 449
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.5513811136094787e-05,
      "loss": 0.7349,
      "step": 450
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.5318890011485065e-05,
      "loss": 0.3637,
      "step": 451
    },
    {
      "epoch": 4.52,
      "learning_rate": 1.5124979408729861e-05,
      "loss": 0.4416,
      "step": 452
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.4932084977905042e-05,
      "loss": 0.5081,
      "step": 453
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.4740212339477721e-05,
      "loss": 0.4887,
      "step": 454
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.4549367084142484e-05,
      "loss": 0.5103,
      "step": 455
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.4359554772658552e-05,
      "loss": 0.6833,
      "step": 456
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.4170780935687656e-05,
      "loss": 0.5815,
      "step": 457
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.3983051073632997e-05,
      "loss": 0.5264,
      "step": 458
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.3796370656478935e-05,
      "loss": 0.352,
      "step": 459
    },
    {
      "epoch": 4.6,
      "learning_rate": 1.3610745123631535e-05,
      "loss": 0.441,
      "step": 460
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.3426179883760192e-05,
      "loss": 0.2207,
      "step": 461
    },
    {
      "epoch": 4.62,
      "learning_rate": 1.3242680314639993e-05,
      "loss": 0.3624,
      "step": 462
    },
    {
      "epoch": 4.63,
      "learning_rate": 1.3060251762994946e-05,
      "loss": 0.4462,
      "step": 463
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.2878899544342327e-05,
      "loss": 0.3813,
      "step": 464
    },
    {
      "epoch": 4.65,
      "learning_rate": 1.2698628942837699e-05,
      "loss": 0.2802,
      "step": 465
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.2519445211120979e-05,
      "loss": 0.5328,
      "step": 466
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.2341353570163355e-05,
      "loss": 0.6202,
      "step": 467
    },
    {
      "epoch": 4.68,
      "learning_rate": 1.2164359209115234e-05,
      "loss": 0.4611,
      "step": 468
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.1988467285154987e-05,
      "loss": 0.3121,
      "step": 469
    },
    {
      "epoch": 4.7,
      "learning_rate": 1.1813682923338653e-05,
      "loss": 0.5383,
      "step": 470
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.1640011216450691e-05,
      "loss": 0.253,
      "step": 471
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.1467457224855544e-05,
      "loss": 0.6717,
      "step": 472
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.129602597635015e-05,
      "loss": 0.6211,
      "step": 473
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.1125722466017547e-05,
      "loss": 0.3007,
      "step": 474
    },
    {
      "epoch": 4.75,
      "learning_rate": 1.09565516560812e-05,
      "loss": 0.5946,
      "step": 475
    },
    {
      "epoch": 4.76,
      "learning_rate": 1.0788518475760545e-05,
      "loss": 0.6937,
      "step": 476
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.0621627821127289e-05,
      "loss": 0.3618,
      "step": 477
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.0455884554962725e-05,
      "loss": 0.3505,
      "step": 478
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.0291293506616133e-05,
      "loss": 0.6379,
      "step": 479
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.012785947186397e-05,
      "loss": 0.6715,
      "step": 480
    },
    {
      "epoch": 4.81,
      "learning_rate": 9.965587212770195e-06,
      "loss": 0.6154,
      "step": 481
    },
    {
      "epoch": 4.82,
      "learning_rate": 9.804481457547498e-06,
      "loss": 0.6582,
      "step": 482
    },
    {
      "epoch": 4.83,
      "learning_rate": 9.644546900419533e-06,
      "loss": 0.5089,
      "step": 483
    },
    {
      "epoch": 4.84,
      "learning_rate": 9.485788201484126e-06,
      "loss": 0.1242,
      "step": 484
    },
    {
      "epoch": 4.85,
      "learning_rate": 9.32820998657748e-06,
      "loss": 0.2903,
      "step": 485
    },
    {
      "epoch": 4.86,
      "learning_rate": 9.171816847139448e-06,
      "loss": 0.3307,
      "step": 486
    },
    {
      "epoch": 4.87,
      "learning_rate": 9.016613340079705e-06,
      "loss": 0.5639,
      "step": 487
    },
    {
      "epoch": 4.88,
      "learning_rate": 8.86260398764494e-06,
      "loss": 0.1877,
      "step": 488
    },
    {
      "epoch": 4.89,
      "learning_rate": 8.70979327728718e-06,
      "loss": 0.8098,
      "step": 489
    },
    {
      "epoch": 4.9,
      "learning_rate": 8.558185661532941e-06,
      "loss": 0.7206,
      "step": 490
    },
    {
      "epoch": 4.91,
      "learning_rate": 8.407785557853553e-06,
      "loss": 0.6704,
      "step": 491
    },
    {
      "epoch": 4.92,
      "learning_rate": 8.25859734853645e-06,
      "loss": 0.5953,
      "step": 492
    },
    {
      "epoch": 4.93,
      "learning_rate": 8.110625380557402e-06,
      "loss": 0.3174,
      "step": 493
    },
    {
      "epoch": 4.94,
      "learning_rate": 7.96387396545396e-06,
      "loss": 0.3217,
      "step": 494
    },
    {
      "epoch": 4.95,
      "learning_rate": 7.81834737919978e-06,
      "loss": 0.3145,
      "step": 495
    },
    {
      "epoch": 4.96,
      "learning_rate": 7.67404986207999e-06,
      "loss": 0.2236,
      "step": 496
    },
    {
      "epoch": 4.97,
      "learning_rate": 7.530985618567726e-06,
      "loss": 0.4421,
      "step": 497
    },
    {
      "epoch": 4.98,
      "learning_rate": 7.389158817201542e-06,
      "loss": 0.3936,
      "step": 498
    },
    {
      "epoch": 4.99,
      "learning_rate": 7.248573590464014e-06,
      "loss": 0.5713,
      "step": 499
    },
    {
      "epoch": 5.0,
      "learning_rate": 7.109234034661289e-06,
      "loss": 0.6414,
      "step": 500
    },
    {
      "epoch": 5.01,
      "learning_rate": 6.9711442098037375e-06,
      "loss": 0.6305,
      "step": 501
    },
    {
      "epoch": 5.02,
      "learning_rate": 6.8343081394876715e-06,
      "loss": 0.663,
      "step": 502
    },
    {
      "epoch": 5.03,
      "learning_rate": 6.698729810778065e-06,
      "loss": 0.197,
      "step": 503
    },
    {
      "epoch": 5.04,
      "learning_rate": 6.564413174092443e-06,
      "loss": 0.3142,
      "step": 504
    },
    {
      "epoch": 5.05,
      "learning_rate": 6.431362143085734e-06,
      "loss": 0.2621,
      "step": 505
    },
    {
      "epoch": 5.06,
      "learning_rate": 6.299580594536214e-06,
      "loss": 0.4311,
      "step": 506
    },
    {
      "epoch": 5.07,
      "learning_rate": 6.16907236823262e-06,
      "loss": 0.4916,
      "step": 507
    },
    {
      "epoch": 5.08,
      "learning_rate": 6.0398412668621895e-06,
      "loss": 0.2118,
      "step": 508
    },
    {
      "epoch": 5.09,
      "learning_rate": 5.91189105589992e-06,
      "loss": 0.3151,
      "step": 509
    },
    {
      "epoch": 5.1,
      "learning_rate": 5.785225463498828e-06,
      "loss": 0.457,
      "step": 510
    },
    {
      "epoch": 5.11,
      "learning_rate": 5.659848180381283e-06,
      "loss": 0.3601,
      "step": 511
    },
    {
      "epoch": 5.12,
      "learning_rate": 5.535762859731547e-06,
      "loss": 0.232,
      "step": 512
    },
    {
      "epoch": 5.13,
      "learning_rate": 5.412973117089287e-06,
      "loss": 0.5667,
      "step": 513
    },
    {
      "epoch": 5.14,
      "learning_rate": 5.291482530244179e-06,
      "loss": 0.5585,
      "step": 514
    },
    {
      "epoch": 5.15,
      "learning_rate": 5.171294639131779e-06,
      "loss": 0.4921,
      "step": 515
    },
    {
      "epoch": 5.16,
      "learning_rate": 5.05241294573024e-06,
      "loss": 0.4778,
      "step": 516
    },
    {
      "epoch": 5.17,
      "learning_rate": 4.934840913958388e-06,
      "loss": 0.5271,
      "step": 517
    },
    {
      "epoch": 5.18,
      "learning_rate": 4.818581969574742e-06,
      "loss": 0.2727,
      "step": 518
    },
    {
      "epoch": 5.19,
      "learning_rate": 4.703639500077656e-06,
      "loss": 0.6403,
      "step": 519
    },
    {
      "epoch": 5.2,
      "learning_rate": 4.590016854606727e-06,
      "loss": 0.4939,
      "step": 520
    },
    {
      "epoch": 5.21,
      "learning_rate": 4.477717343845078e-06,
      "loss": 0.5571,
      "step": 521
    },
    {
      "epoch": 5.22,
      "learning_rate": 4.366744239922998e-06,
      "loss": 0.2923,
      "step": 522
    },
    {
      "epoch": 5.23,
      "learning_rate": 4.257100776322525e-06,
      "loss": 0.336,
      "step": 523
    },
    {
      "epoch": 5.24,
      "learning_rate": 4.148790147783288e-06,
      "loss": 0.4136,
      "step": 524
    },
    {
      "epoch": 5.25,
      "learning_rate": 4.041815510209396e-06,
      "loss": 0.6364,
      "step": 525
    },
    {
      "epoch": 5.26,
      "learning_rate": 3.936179980577453e-06,
      "loss": 0.5695,
      "step": 526
    },
    {
      "epoch": 5.27,
      "learning_rate": 3.8318866368458e-06,
      "loss": 0.2892,
      "step": 527
    },
    {
      "epoch": 5.28,
      "learning_rate": 3.728938517864794e-06,
      "loss": 0.2887,
      "step": 528
    },
    {
      "epoch": 5.29,
      "learning_rate": 3.6273386232882343e-06,
      "loss": 0.2571,
      "step": 529
    },
    {
      "epoch": 5.3,
      "learning_rate": 3.527089913486037e-06,
      "loss": 0.5463,
      "step": 530
    },
    {
      "epoch": 5.31,
      "learning_rate": 3.4281953094578877e-06,
      "loss": 0.4527,
      "step": 531
    },
    {
      "epoch": 5.32,
      "learning_rate": 3.3306576927482126e-06,
      "loss": 0.2259,
      "step": 532
    },
    {
      "epoch": 5.33,
      "learning_rate": 3.2344799053621646e-06,
      "loss": 0.4806,
      "step": 533
    },
    {
      "epoch": 5.34,
      "learning_rate": 3.1396647496828247e-06,
      "loss": 0.3638,
      "step": 534
    },
    {
      "epoch": 5.35,
      "learning_rate": 3.0462149883895563e-06,
      "loss": 0.5568,
      "step": 535
    },
    {
      "epoch": 5.36,
      "learning_rate": 2.9541333443775243e-06,
      "loss": 0.6148,
      "step": 536
    },
    {
      "epoch": 5.37,
      "learning_rate": 2.8634225006782865e-06,
      "loss": 0.0862,
      "step": 537
    },
    {
      "epoch": 5.38,
      "learning_rate": 2.774085100381735e-06,
      "loss": 0.5631,
      "step": 538
    },
    {
      "epoch": 5.39,
      "learning_rate": 2.686123746558961e-06,
      "loss": 0.6981,
      "step": 539
    },
    {
      "epoch": 5.4,
      "learning_rate": 2.5995410021864787e-06,
      "loss": 0.2124,
      "step": 540
    },
    {
      "epoch": 5.41,
      "learning_rate": 2.5143393900715296e-06,
      "loss": 0.1605,
      "step": 541
    },
    {
      "epoch": 5.42,
      "learning_rate": 2.430521392778573e-06,
      "loss": 0.5083,
      "step": 542
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.3480894525569562e-06,
      "loss": 0.4244,
      "step": 543
    },
    {
      "epoch": 5.44,
      "learning_rate": 2.2670459712697377e-06,
      "loss": 0.259,
      "step": 544
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.187393310323721e-06,
      "loss": 0.088,
      "step": 545
    },
    {
      "epoch": 5.46,
      "learning_rate": 2.1091337906006482e-06,
      "loss": 0.1838,
      "step": 546
    },
    {
      "epoch": 5.47,
      "learning_rate": 2.0322696923895434e-06,
      "loss": 0.5571,
      "step": 547
    },
    {
      "epoch": 5.48,
      "learning_rate": 1.956803255320322e-06,
      "loss": 0.3224,
      "step": 548
    },
    {
      "epoch": 5.49,
      "learning_rate": 1.8827366782984913e-06,
      "loss": 0.7158,
      "step": 549
    },
    {
      "epoch": 5.5,
      "learning_rate": 1.810072119441103e-06,
      "loss": 0.4693,
      "step": 550
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.73881169601387e-06,
      "loss": 0.4278,
      "step": 551
    },
    {
      "epoch": 5.52,
      "learning_rate": 1.6689574843694433e-06,
      "loss": 0.0819,
      "step": 552
    },
    {
      "epoch": 5.53,
      "learning_rate": 1.6005115198869603e-06,
      "loss": 0.4137,
      "step": 553
    },
    {
      "epoch": 5.54,
      "learning_rate": 1.53347579691272e-06,
      "loss": 0.5221,
      "step": 554
    },
    {
      "epoch": 5.55,
      "learning_rate": 1.4678522687020413e-06,
      "loss": 0.5139,
      "step": 555
    },
    {
      "epoch": 5.56,
      "learning_rate": 1.4036428473624019e-06,
      "loss": 0.5131,
      "step": 556
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.3408494037976894e-06,
      "loss": 0.4282,
      "step": 557
    },
    {
      "epoch": 5.58,
      "learning_rate": 1.2794737676536994e-06,
      "loss": 0.6512,
      "step": 558
    },
    {
      "epoch": 5.59,
      "learning_rate": 1.2195177272648127e-06,
      "loss": 0.6242,
      "step": 559
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.1609830296019143e-06,
      "loss": 0.7032,
      "step": 560
    },
    {
      "epoch": 5.61,
      "learning_rate": 1.1038713802214717e-06,
      "loss": 0.3343,
      "step": 561
    },
    {
      "epoch": 5.62,
      "learning_rate": 1.0481844432158161e-06,
      "loss": 0.1955,
      "step": 562
    },
    {
      "epoch": 5.63,
      "learning_rate": 9.939238411647235e-07,
      "loss": 0.0995,
      "step": 563
    },
    {
      "epoch": 5.64,
      "learning_rate": 9.410911550880475e-07,
      "loss": 0.2064,
      "step": 564
    },
    {
      "epoch": 5.65,
      "learning_rate": 8.896879243997347e-07,
      "loss": 0.3112,
      "step": 565
    },
    {
      "epoch": 5.66,
      "learning_rate": 8.397156468629208e-07,
      "loss": 0.4229,
      "step": 566
    },
    {
      "epoch": 5.67,
      "learning_rate": 7.911757785462881e-07,
      "loss": 0.0823,
      "step": 567
    },
    {
      "epoch": 5.68,
      "learning_rate": 7.44069733781677e-07,
      "loss": 0.5938,
      "step": 568
    },
    {
      "epoch": 5.69,
      "learning_rate": 6.983988851228473e-07,
      "loss": 0.7056,
      "step": 569
    },
    {
      "epoch": 5.7,
      "learning_rate": 6.54164563305465e-07,
      "loss": 0.0954,
      "step": 570
    },
    {
      "epoch": 5.71,
      "learning_rate": 6.113680572083946e-07,
      "loss": 0.7448,
      "step": 571
    },
    {
      "epoch": 5.72,
      "learning_rate": 5.700106138160688e-07,
      "loss": 0.5686,
      "step": 572
    },
    {
      "epoch": 5.73,
      "learning_rate": 5.300934381821998e-07,
      "loss": 0.2387,
      "step": 573
    },
    {
      "epoch": 5.74,
      "learning_rate": 4.916176933946693e-07,
      "loss": 0.3178,
      "step": 574
    },
    {
      "epoch": 5.75,
      "learning_rate": 4.545845005415994e-07,
      "loss": 0.4377,
      "step": 575
    },
    {
      "epoch": 5.76,
      "learning_rate": 4.189949386787462e-07,
      "loss": 0.3517,
      "step": 576
    },
    {
      "epoch": 5.77,
      "learning_rate": 3.848500447979908e-07,
      "loss": 0.7012,
      "step": 577
    },
    {
      "epoch": 5.78,
      "learning_rate": 3.5215081379718074e-07,
      "loss": 0.3336,
      "step": 578
    },
    {
      "epoch": 5.79,
      "learning_rate": 3.208981984511195e-07,
      "loss": 0.5298,
      "step": 579
    },
    {
      "epoch": 5.8,
      "learning_rate": 2.9109310938378877e-07,
      "loss": 0.7599,
      "step": 580
    },
    {
      "epoch": 5.81,
      "learning_rate": 2.6273641504184766e-07,
      "loss": 0.6857,
      "step": 581
    },
    {
      "epoch": 5.82,
      "learning_rate": 2.3582894166930268e-07,
      "loss": 0.1058,
      "step": 582
    },
    {
      "epoch": 5.83,
      "learning_rate": 2.1037147328344387e-07,
      "loss": 0.3377,
      "step": 583
    },
    {
      "epoch": 5.84,
      "learning_rate": 1.8636475165200174e-07,
      "loss": 0.2177,
      "step": 584
    },
    {
      "epoch": 5.85,
      "learning_rate": 1.6380947627153143e-07,
      "loss": 0.2113,
      "step": 585
    },
    {
      "epoch": 5.86,
      "learning_rate": 1.427063043470178e-07,
      "loss": 0.6957,
      "step": 586
    },
    {
      "epoch": 5.87,
      "learning_rate": 1.2305585077276306e-07,
      "loss": 0.6408,
      "step": 587
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.0485868811441757e-07,
      "loss": 0.3335,
      "step": 588
    },
    {
      "epoch": 5.89,
      "learning_rate": 8.811534659234899e-08,
      "loss": 0.5002,
      "step": 589
    },
    {
      "epoch": 5.9,
      "learning_rate": 7.282631406615447e-08,
      "loss": 0.7505,
      "step": 590
    },
    {
      "epoch": 5.91,
      "learning_rate": 5.899203602046655e-08,
      "loss": 0.5746,
      "step": 591
    },
    {
      "epoch": 5.92,
      "learning_rate": 4.661291555196345e-08,
      "loss": 0.3938,
      "step": 592
    },
    {
      "epoch": 5.93,
      "learning_rate": 3.5689313357634145e-08,
      "loss": 0.5023,
      "step": 593
    },
    {
      "epoch": 5.94,
      "learning_rate": 2.6221547724253337e-08,
      "loss": 0.6144,
      "step": 594
    },
    {
      "epoch": 5.95,
      "learning_rate": 1.8209894519122252e-08,
      "loss": 0.4173,
      "step": 595
    },
    {
      "epoch": 5.96,
      "learning_rate": 1.1654587182013953e-08,
      "loss": 0.5752,
      "step": 596
    },
    {
      "epoch": 5.97,
      "learning_rate": 6.5558167183898955e-09,
      "loss": 0.0822,
      "step": 597
    },
    {
      "epoch": 5.98,
      "learning_rate": 2.9137316938265825e-09,
      "loss": 0.5562,
      "step": 598
    },
    {
      "epoch": 5.99,
      "learning_rate": 7.284382296801617e-10,
      "loss": 0.1803,
      "step": 599
    },
    {
      "epoch": 6.0,
      "learning_rate": 0.0,
      "loss": 0.5857,
      "step": 600
    },
    {
      "epoch": 6.0,
      "step": 600,
      "total_flos": 1.0238347568039854e+18,
      "train_loss": 0.6276818083102504,
      "train_runtime": 6533.1892,
      "train_samples_per_second": 0.367,
      "train_steps_per_second": 0.092
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 600,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 10000,
  "total_flos": 1.0238347568039854e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}