{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 730,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 5.681818181818182e-07,
      "loss": 1.6669,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.1363636363636364e-06,
      "loss": 1.4274,
      "step": 2
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.7045454545454546e-06,
      "loss": 1.5736,
      "step": 3
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 1.6031,
      "step": 4
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.840909090909091e-06,
      "loss": 1.4013,
      "step": 5
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.409090909090909e-06,
      "loss": 1.4383,
      "step": 6
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.9772727272727275e-06,
      "loss": 1.4783,
      "step": 7
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 1.4839,
      "step": 8
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.113636363636364e-06,
      "loss": 1.2941,
      "step": 9
    },
    {
      "epoch": 0.07,
      "learning_rate": 5.681818181818182e-06,
      "loss": 1.5711,
      "step": 10
    },
    {
      "epoch": 0.08,
      "learning_rate": 6.25e-06,
      "loss": 1.5195,
      "step": 11
    },
    {
      "epoch": 0.08,
      "learning_rate": 6.818181818181818e-06,
      "loss": 1.535,
      "step": 12
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.386363636363637e-06,
      "loss": 1.3355,
      "step": 13
    },
    {
      "epoch": 0.1,
      "learning_rate": 7.954545454545455e-06,
      "loss": 1.5082,
      "step": 14
    },
    {
      "epoch": 0.1,
      "learning_rate": 8.522727272727273e-06,
      "loss": 1.4538,
      "step": 15
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.090909090909091e-06,
      "loss": 1.3486,
      "step": 16
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.659090909090909e-06,
      "loss": 1.4842,
      "step": 17
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.0227272727272729e-05,
      "loss": 1.4197,
      "step": 18
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.0795454545454547e-05,
      "loss": 1.4488,
      "step": 19
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.1363636363636365e-05,
      "loss": 1.4491,
      "step": 20
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.1931818181818183e-05,
      "loss": 1.3609,
      "step": 21
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.25e-05,
      "loss": 1.4129,
      "step": 22
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.3068181818181819e-05,
      "loss": 1.1582,
      "step": 23
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.3636363636363637e-05,
      "loss": 1.4224,
      "step": 24
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.4204545454545456e-05,
      "loss": 1.4216,
      "step": 25
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.4772727272727274e-05,
      "loss": 1.4054,
      "step": 26
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.534090909090909e-05,
      "loss": 1.3476,
      "step": 27
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.590909090909091e-05,
      "loss": 1.4123,
      "step": 28
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.6477272727272726e-05,
      "loss": 1.2541,
      "step": 29
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.7045454545454546e-05,
      "loss": 1.3971,
      "step": 30
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.7613636363636366e-05,
      "loss": 1.4469,
      "step": 31
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 1.3165,
      "step": 32
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 1.3095,
      "step": 33
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9318181818181818e-05,
      "loss": 1.2892,
      "step": 34
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9886363636363638e-05,
      "loss": 1.2706,
      "step": 35
    },
    {
      "epoch": 0.25,
      "learning_rate": 2.0454545454545457e-05,
      "loss": 1.3491,
      "step": 36
    },
    {
      "epoch": 0.25,
      "learning_rate": 2.1022727272727274e-05,
      "loss": 1.3282,
      "step": 37
    },
    {
      "epoch": 0.26,
      "learning_rate": 2.1590909090909093e-05,
      "loss": 1.1812,
      "step": 38
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.215909090909091e-05,
      "loss": 1.3438,
      "step": 39
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.272727272727273e-05,
      "loss": 1.3827,
      "step": 40
    },
    {
      "epoch": 0.28,
      "learning_rate": 2.3295454545454546e-05,
      "loss": 1.2663,
      "step": 41
    },
    {
      "epoch": 0.29,
      "learning_rate": 2.3863636363636365e-05,
      "loss": 1.3777,
      "step": 42
    },
    {
      "epoch": 0.29,
      "learning_rate": 2.4431818181818185e-05,
      "loss": 1.2316,
      "step": 43
    },
    {
      "epoch": 0.3,
      "learning_rate": 2.5e-05,
      "loss": 1.3528,
      "step": 44
    },
    {
      "epoch": 0.31,
      "learning_rate": 2.5568181818181817e-05,
      "loss": 1.379,
      "step": 45
    },
    {
      "epoch": 0.32,
      "learning_rate": 2.6136363636363637e-05,
      "loss": 1.3162,
      "step": 46
    },
    {
      "epoch": 0.32,
      "learning_rate": 2.6704545454545453e-05,
      "loss": 1.2006,
      "step": 47
    },
    {
      "epoch": 0.33,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 1.4458,
      "step": 48
    },
    {
      "epoch": 0.34,
      "learning_rate": 2.784090909090909e-05,
      "loss": 1.2852,
      "step": 49
    },
    {
      "epoch": 0.34,
      "learning_rate": 2.8409090909090912e-05,
      "loss": 1.2553,
      "step": 50
    },
    {
      "epoch": 0.35,
      "learning_rate": 2.8977272727272732e-05,
      "loss": 1.4476,
      "step": 51
    },
    {
      "epoch": 0.36,
      "learning_rate": 2.954545454545455e-05,
      "loss": 1.2309,
      "step": 52
    },
    {
      "epoch": 0.36,
      "learning_rate": 3.0113636363636365e-05,
      "loss": 1.3228,
      "step": 53
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.068181818181818e-05,
      "loss": 1.3367,
      "step": 54
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.125e-05,
      "loss": 1.2866,
      "step": 55
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.181818181818182e-05,
      "loss": 1.3093,
      "step": 56
    },
    {
      "epoch": 0.39,
      "learning_rate": 3.238636363636364e-05,
      "loss": 1.3932,
      "step": 57
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.295454545454545e-05,
      "loss": 1.1568,
      "step": 58
    },
    {
      "epoch": 0.4,
      "learning_rate": 3.352272727272727e-05,
      "loss": 1.2535,
      "step": 59
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.409090909090909e-05,
      "loss": 1.2189,
      "step": 60
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.465909090909091e-05,
      "loss": 1.149,
      "step": 61
    },
    {
      "epoch": 0.42,
      "learning_rate": 3.522727272727273e-05,
      "loss": 1.3026,
      "step": 62
    },
    {
      "epoch": 0.43,
      "learning_rate": 3.579545454545455e-05,
      "loss": 1.2468,
      "step": 63
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 1.4326,
      "step": 64
    },
    {
      "epoch": 0.45,
      "learning_rate": 3.6931818181818184e-05,
      "loss": 1.2512,
      "step": 65
    },
    {
      "epoch": 0.45,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 1.3661,
      "step": 66
    },
    {
      "epoch": 0.46,
      "learning_rate": 3.8068181818181816e-05,
      "loss": 1.2677,
      "step": 67
    },
    {
      "epoch": 0.47,
      "learning_rate": 3.8636363636363636e-05,
      "loss": 1.354,
      "step": 68
    },
    {
      "epoch": 0.47,
      "learning_rate": 3.9204545454545456e-05,
      "loss": 1.2258,
      "step": 69
    },
    {
      "epoch": 0.48,
      "learning_rate": 3.9772727272727275e-05,
      "loss": 1.1839,
      "step": 70
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.034090909090909e-05,
      "loss": 1.4097,
      "step": 71
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.0909090909090915e-05,
      "loss": 1.2499,
      "step": 72
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.1477272727272734e-05,
      "loss": 1.3991,
      "step": 73
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.204545454545455e-05,
      "loss": 1.2102,
      "step": 74
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.261363636363637e-05,
      "loss": 1.3301,
      "step": 75
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.318181818181819e-05,
      "loss": 1.2872,
      "step": 76
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.375e-05,
      "loss": 1.2391,
      "step": 77
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.431818181818182e-05,
      "loss": 1.2094,
      "step": 78
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.488636363636364e-05,
      "loss": 1.1217,
      "step": 79
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.545454545454546e-05,
      "loss": 1.3957,
      "step": 80
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.602272727272727e-05,
      "loss": 1.1267,
      "step": 81
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.659090909090909e-05,
      "loss": 1.0954,
      "step": 82
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.715909090909091e-05,
      "loss": 1.412,
      "step": 83
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.772727272727273e-05,
      "loss": 1.076,
      "step": 84
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.829545454545455e-05,
      "loss": 1.2825,
      "step": 85
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.886363636363637e-05,
      "loss": 1.1066,
      "step": 86
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.943181818181818e-05,
      "loss": 1.2088,
      "step": 87
    },
    {
      "epoch": 0.6,
      "learning_rate": 5e-05,
      "loss": 1.1819,
      "step": 88
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.999998461763134e-05,
      "loss": 1.2009,
      "step": 89
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.9999938470544264e-05,
      "loss": 1.1569,
      "step": 90
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.999986155879557e-05,
      "loss": 0.9333,
      "step": 91
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.9999753882479914e-05,
      "loss": 1.2257,
      "step": 92
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.9999615441729785e-05,
      "loss": 1.26,
      "step": 93
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.999944623671556e-05,
      "loss": 1.185,
      "step": 94
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.9999246267645464e-05,
      "loss": 1.2355,
      "step": 95
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.999901553476555e-05,
      "loss": 1.323,
      "step": 96
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.9998754038359785e-05,
      "loss": 1.2167,
      "step": 97
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.999846177874995e-05,
      "loss": 1.2381,
      "step": 98
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.9998138756295704e-05,
      "loss": 1.1787,
      "step": 99
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.999778497139455e-05,
      "loss": 1.2369,
      "step": 100
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.9997400424481844e-05,
      "loss": 1.265,
      "step": 101
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.999698511603082e-05,
      "loss": 1.1829,
      "step": 102
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.9996539046552546e-05,
      "loss": 1.2228,
      "step": 103
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.999606221659595e-05,
      "loss": 1.4031,
      "step": 104
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.999555462674781e-05,
      "loss": 1.2798,
      "step": 105
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.999501627763277e-05,
      "loss": 1.1808,
      "step": 106
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.999444716991332e-05,
      "loss": 1.2457,
      "step": 107
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.9993847304289774e-05,
      "loss": 1.3655,
      "step": 108
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.9993216681500346e-05,
      "loss": 1.187,
      "step": 109
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.999255530232105e-05,
      "loss": 1.1713,
      "step": 110
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.99918631675658e-05,
      "loss": 1.1947,
      "step": 111
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.9991140278086316e-05,
      "loss": 1.1169,
      "step": 112
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.999038663477218e-05,
      "loss": 1.2369,
      "step": 113
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.99896022385508e-05,
      "loss": 1.2075,
      "step": 114
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.998878709038748e-05,
      "loss": 1.2208,
      "step": 115
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.99879411912853e-05,
      "loss": 1.1711,
      "step": 116
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.998706454228524e-05,
      "loss": 1.2532,
      "step": 117
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.998615714446608e-05,
      "loss": 1.1868,
      "step": 118
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.998521899894446e-05,
      "loss": 1.2798,
      "step": 119
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.998425010687484e-05,
      "loss": 1.1243,
      "step": 120
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.998325046944955e-05,
      "loss": 1.3059,
      "step": 121
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.9982220087898715e-05,
      "loss": 1.2788,
      "step": 122
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.998115896349032e-05,
      "loss": 1.1768,
      "step": 123
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.998006709753017e-05,
      "loss": 1.2633,
      "step": 124
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.997894449136191e-05,
      "loss": 1.0365,
      "step": 125
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.9977791146367005e-05,
      "loss": 1.2884,
      "step": 126
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.997660706396474e-05,
      "loss": 1.2045,
      "step": 127
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.9975392245612254e-05,
      "loss": 1.2226,
      "step": 128
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.997414669280446e-05,
      "loss": 1.237,
      "step": 129
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.997287040707415e-05,
      "loss": 1.1576,
      "step": 130
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.99715633899919e-05,
      "loss": 1.2579,
      "step": 131
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.99702256431661e-05,
      "loss": 1.3043,
      "step": 132
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.996885716824299e-05,
      "loss": 1.2905,
      "step": 133
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.996745796690657e-05,
      "loss": 1.1184,
      "step": 134
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.996602804087871e-05,
      "loss": 1.0862,
      "step": 135
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.996456739191905e-05,
      "loss": 1.2715,
      "step": 136
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.996307602182505e-05,
      "loss": 1.0201,
      "step": 137
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.996155393243198e-05,
      "loss": 1.3834,
      "step": 138
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.996000112561289e-05,
      "loss": 1.1772,
      "step": 139
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.995841760327867e-05,
      "loss": 1.14,
      "step": 140
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.995680336737797e-05,
      "loss": 1.3104,
      "step": 141
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.995515841989726e-05,
      "loss": 1.1304,
      "step": 142
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.9953482762860796e-05,
      "loss": 1.0273,
      "step": 143
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.995177639833062e-05,
      "loss": 1.2024,
      "step": 144
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.995003932840657e-05,
      "loss": 1.2857,
      "step": 145
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.994827155522625e-05,
      "loss": 1.1091,
      "step": 146
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.994647308096509e-05,
      "loss": 1.3138,
      "step": 147
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.994464390783625e-05,
      "loss": 1.1692,
      "step": 148
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.994278403809071e-05,
      "loss": 1.2849,
      "step": 149
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.994089347401719e-05,
      "loss": 1.3361,
      "step": 150
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.993897221794221e-05,
      "loss": 1.1463,
      "step": 151
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.993702027223004e-05,
      "loss": 1.1808,
      "step": 152
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.9935037639282725e-05,
      "loss": 1.1949,
      "step": 153
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.993302432154008e-05,
      "loss": 1.2023,
      "step": 154
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.993098032147966e-05,
      "loss": 0.9441,
      "step": 155
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.9928905641616794e-05,
      "loss": 1.26,
      "step": 156
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.992680028450457e-05,
      "loss": 1.2594,
      "step": 157
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.9924664252733816e-05,
      "loss": 1.2523,
      "step": 158
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.99224975489331e-05,
      "loss": 1.0497,
      "step": 159
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.992030017576875e-05,
      "loss": 1.2882,
      "step": 160
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.991807213594484e-05,
      "loss": 1.1562,
      "step": 161
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.9915813432203165e-05,
      "loss": 1.1235,
      "step": 162
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.991352406732325e-05,
      "loss": 1.2257,
      "step": 163
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.991120404412238e-05,
      "loss": 1.2162,
      "step": 164
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.990885336545555e-05,
      "loss": 1.2173,
      "step": 165
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.9906472034215466e-05,
      "loss": 1.2527,
      "step": 166
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.990406005333259e-05,
      "loss": 1.1254,
      "step": 167
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.9901617425775067e-05,
      "loss": 1.2175,
      "step": 168
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.989914415454877e-05,
      "loss": 0.9104,
      "step": 169
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.9896640242697276e-05,
      "loss": 1.2549,
      "step": 170
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.9894105693301896e-05,
      "loss": 1.2029,
      "step": 171
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.989154050948159e-05,
      "loss": 1.1945,
      "step": 172
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.988894469439306e-05,
      "loss": 1.1285,
      "step": 173
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.988631825123069e-05,
      "loss": 1.2458,
      "step": 174
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.988366118322655e-05,
      "loss": 1.0894,
      "step": 175
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.9880973493650394e-05,
      "loss": 1.2417,
      "step": 176
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.9878255185809684e-05,
      "loss": 1.2962,
      "step": 177
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.987550626304952e-05,
      "loss": 1.1519,
      "step": 178
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.987272672875271e-05,
      "loss": 1.116,
      "step": 179
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.986991658633972e-05,
      "loss": 1.0908,
      "step": 180
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.986707583926867e-05,
      "loss": 1.0814,
      "step": 181
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.986420449103536e-05,
      "loss": 1.182,
      "step": 182
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.986130254517325e-05,
      "loss": 1.1839,
      "step": 183
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.985837000525343e-05,
      "loss": 1.0252,
      "step": 184
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.985540687488466e-05,
      "loss": 1.219,
      "step": 185
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.985241315771334e-05,
      "loss": 1.2573,
      "step": 186
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.98493888574235e-05,
      "loss": 1.1408,
      "step": 187
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.9846333977736813e-05,
      "loss": 1.2265,
      "step": 188
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.984324852241259e-05,
      "loss": 1.0908,
      "step": 189
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.984013249524775e-05,
      "loss": 1.2468,
      "step": 190
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.9836985900076844e-05,
      "loss": 1.2172,
      "step": 191
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.983380874077204e-05,
      "loss": 1.2108,
      "step": 192
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.9830601021243125e-05,
      "loss": 1.0567,
      "step": 193
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.982736274543748e-05,
      "loss": 1.3061,
      "step": 194
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.982409391734009e-05,
      "loss": 1.1814,
      "step": 195
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.982079454097354e-05,
      "loss": 1.1392,
      "step": 196
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.981746462039801e-05,
      "loss": 1.3159,
      "step": 197
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.981410415971127e-05,
      "loss": 1.1255,
      "step": 198
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.981071316304867e-05,
      "loss": 1.2056,
      "step": 199
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.980729163458312e-05,
      "loss": 1.2303,
      "step": 200
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.980383957852512e-05,
      "loss": 1.1711,
      "step": 201
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.9800356999122746e-05,
      "loss": 1.203,
      "step": 202
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.9796843900661613e-05,
      "loss": 1.2845,
      "step": 203
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.979330028746491e-05,
      "loss": 1.0433,
      "step": 204
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.978972616389337e-05,
      "loss": 1.1515,
      "step": 205
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.9786121534345265e-05,
      "loss": 1.1155,
      "step": 206
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.978248640325641e-05,
      "loss": 1.0288,
      "step": 207
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.9778820775100185e-05,
      "loss": 1.1976,
      "step": 208
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.977512465438744e-05,
      "loss": 1.1593,
      "step": 209
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.9771398045666606e-05,
      "loss": 1.3097,
      "step": 210
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.97676409535236e-05,
      "loss": 1.1394,
      "step": 211
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.976385338258186e-05,
      "loss": 1.2662,
      "step": 212
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.976003533750233e-05,
      "loss": 1.1382,
      "step": 213
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.9756186822983464e-05,
      "loss": 1.248,
      "step": 214
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.975230784376119e-05,
      "loss": 1.1389,
      "step": 215
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.974839840460895e-05,
      "loss": 1.0841,
      "step": 216
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.974445851033765e-05,
      "loss": 1.317,
      "step": 217
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.97404881657957e-05,
      "loss": 1.1366,
      "step": 218
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.973648737586894e-05,
      "loss": 1.2806,
      "step": 219
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.973245614548072e-05,
      "loss": 1.1281,
      "step": 220
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.972839447959181e-05,
      "loss": 1.2346,
      "step": 221
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.9724302383200477e-05,
      "loss": 1.2117,
      "step": 222
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.97201798613424e-05,
      "loss": 1.1455,
      "step": 223
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.9716026919090705e-05,
      "loss": 1.0981,
      "step": 224
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.971184356155597e-05,
      "loss": 1.023,
      "step": 225
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.970762979388618e-05,
      "loss": 1.2858,
      "step": 226
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.9703385621266766e-05,
      "loss": 1.0097,
      "step": 227
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.9699111048920554e-05,
      "loss": 1.0166,
      "step": 228
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.969480608210779e-05,
      "loss": 1.3232,
      "step": 229
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.9690470726126115e-05,
      "loss": 0.9815,
      "step": 230
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.968610498631058e-05,
      "loss": 1.175,
      "step": 231
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.9681708868033616e-05,
      "loss": 1.0065,
      "step": 232
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.967728237670504e-05,
      "loss": 1.0978,
      "step": 233
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.967282551777205e-05,
      "loss": 1.0995,
      "step": 234
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.96683382967192e-05,
      "loss": 1.1151,
      "step": 235
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.966382071906843e-05,
      "loss": 1.0575,
      "step": 236
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.965927279037901e-05,
      "loss": 0.8452,
      "step": 237
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.965469451624759e-05,
      "loss": 1.1209,
      "step": 238
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.965008590230812e-05,
      "loss": 1.171,
      "step": 239
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.9645446954231936e-05,
      "loss": 1.0801,
      "step": 240
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.9640777677727674e-05,
      "loss": 1.1266,
      "step": 241
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.963607807854129e-05,
      "loss": 1.2221,
      "step": 242
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.963134816245606e-05,
      "loss": 1.1198,
      "step": 243
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.962658793529258e-05,
      "loss": 1.1369,
      "step": 244
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.962179740290873e-05,
      "loss": 1.0799,
      "step": 245
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.961697657119968e-05,
      "loss": 1.1408,
      "step": 246
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.9612125446097905e-05,
      "loss": 1.1629,
      "step": 247
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.9607244033573156e-05,
      "loss": 1.0933,
      "step": 248
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.960233233963242e-05,
      "loss": 1.1474,
      "step": 249
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.9597390370320006e-05,
      "loss": 1.3268,
      "step": 250
    },
    {
      "epoch": 1.72,
      "learning_rate": 4.959241813171743e-05,
      "loss": 1.1895,
      "step": 251
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.95874156299435e-05,
      "loss": 1.0743,
      "step": 252
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.958238287115421e-05,
      "loss": 1.16,
      "step": 253
    },
    {
      "epoch": 1.74,
      "learning_rate": 4.957731986154285e-05,
      "loss": 1.2787,
      "step": 254
    },
    {
      "epoch": 1.75,
      "learning_rate": 4.9572226607339886e-05,
      "loss": 1.0935,
      "step": 255
    },
    {
      "epoch": 1.75,
      "learning_rate": 4.956710311481303e-05,
      "loss": 1.0916,
      "step": 256
    },
    {
      "epoch": 1.76,
      "learning_rate": 4.9561949390267196e-05,
      "loss": 1.1169,
      "step": 257
    },
    {
      "epoch": 1.77,
      "learning_rate": 4.95567654400445e-05,
      "loss": 1.0298,
      "step": 258
    },
    {
      "epoch": 1.77,
      "learning_rate": 4.955155127052428e-05,
      "loss": 1.1667,
      "step": 259
    },
    {
      "epoch": 1.78,
      "learning_rate": 4.9546306888123003e-05,
      "loss": 1.1274,
      "step": 260
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.9541032299294375e-05,
      "loss": 1.1245,
      "step": 261
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.953572751052924e-05,
      "loss": 1.0894,
      "step": 262
    },
    {
      "epoch": 1.8,
      "learning_rate": 4.9530392528355626e-05,
      "loss": 1.1729,
      "step": 263
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.9525027359338696e-05,
      "loss": 1.096,
      "step": 264
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.951963201008076e-05,
      "loss": 1.1825,
      "step": 265
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.9514206487221304e-05,
      "loss": 1.0211,
      "step": 266
    },
    {
      "epoch": 1.83,
      "learning_rate": 4.95087507974369e-05,
      "loss": 1.2285,
      "step": 267
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.9503264947441275e-05,
      "loss": 1.2042,
      "step": 268
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.949774894398524e-05,
      "loss": 1.088,
      "step": 269
    },
    {
      "epoch": 1.85,
      "learning_rate": 4.949220279385676e-05,
      "loss": 1.1779,
      "step": 270
    },
    {
      "epoch": 1.86,
      "learning_rate": 4.948662650388084e-05,
      "loss": 0.9534,
      "step": 271
    },
    {
      "epoch": 1.86,
      "learning_rate": 4.948102008091962e-05,
      "loss": 1.1841,
      "step": 272
    },
    {
      "epoch": 1.87,
      "learning_rate": 4.94753835318723e-05,
      "loss": 1.1233,
      "step": 273
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.946971686367516e-05,
      "loss": 1.1338,
      "step": 274
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.9464020083301544e-05,
      "loss": 1.1575,
      "step": 275
    },
    {
      "epoch": 1.89,
      "learning_rate": 4.945829319776184e-05,
      "loss": 1.0758,
      "step": 276
    },
    {
      "epoch": 1.9,
      "learning_rate": 4.945253621410351e-05,
      "loss": 1.172,
      "step": 277
    },
    {
      "epoch": 1.9,
      "learning_rate": 4.944674913941102e-05,
      "loss": 1.2216,
      "step": 278
    },
    {
      "epoch": 1.91,
      "learning_rate": 4.944093198080589e-05,
      "loss": 1.2167,
      "step": 279
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.9435084745446666e-05,
      "loss": 1.0337,
      "step": 280
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.9429207440528876e-05,
      "loss": 1.0044,
      "step": 281
    },
    {
      "epoch": 1.93,
      "learning_rate": 4.942330007328509e-05,
      "loss": 1.1916,
      "step": 282
    },
    {
      "epoch": 1.94,
      "learning_rate": 4.9417362650984835e-05,
      "loss": 0.9445,
      "step": 283
    },
    {
      "epoch": 1.95,
      "learning_rate": 4.941139518093464e-05,
      "loss": 1.3121,
      "step": 284
    },
    {
      "epoch": 1.95,
      "learning_rate": 4.9405397670478014e-05,
      "loss": 1.1049,
      "step": 285
    },
    {
      "epoch": 1.96,
      "learning_rate": 4.9399370126995444e-05,
      "loss": 1.062,
      "step": 286
    },
    {
      "epoch": 1.97,
      "learning_rate": 4.939331255790434e-05,
      "loss": 1.2379,
      "step": 287
    },
    {
      "epoch": 1.97,
      "learning_rate": 4.93872249706591e-05,
      "loss": 1.0523,
      "step": 288
    },
    {
      "epoch": 1.98,
      "learning_rate": 4.938110737275104e-05,
      "loss": 0.9423,
      "step": 289
    },
    {
      "epoch": 1.99,
      "learning_rate": 4.937495977170841e-05,
      "loss": 1.1224,
      "step": 290
    },
    {
      "epoch": 1.99,
      "learning_rate": 4.936878217509637e-05,
      "loss": 1.2134,
      "step": 291
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.936257459051703e-05,
      "loss": 0.9984,
      "step": 292
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.9356337025609365e-05,
      "loss": 1.1952,
      "step": 293
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.935006948804927e-05,
      "loss": 1.0955,
      "step": 294
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.934377198554948e-05,
      "loss": 1.1969,
      "step": 295
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.933744452585966e-05,
      "loss": 1.2647,
      "step": 296
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.933108711676632e-05,
      "loss": 1.0782,
      "step": 297
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.9324699766092806e-05,
      "loss": 1.0923,
      "step": 298
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.931828248169933e-05,
      "loss": 1.1035,
      "step": 299
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.9311835271482943e-05,
      "loss": 1.1281,
      "step": 300
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.93053581433775e-05,
      "loss": 0.8405,
      "step": 301
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.9298851105353696e-05,
      "loss": 1.1676,
      "step": 302
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.929231416541901e-05,
      "loss": 1.192,
      "step": 303
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.9285747331617746e-05,
      "loss": 1.183,
      "step": 304
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.927915061203099e-05,
      "loss": 0.9712,
      "step": 305
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.927252401477657e-05,
      "loss": 1.1965,
      "step": 306
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.926586754800912e-05,
      "loss": 1.0713,
      "step": 307
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.925918121992002e-05,
      "loss": 1.0439,
      "step": 308
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.925246503873739e-05,
      "loss": 1.1367,
      "step": 309
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.92457190127261e-05,
      "loss": 1.1343,
      "step": 310
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.923894315018773e-05,
      "loss": 1.1216,
      "step": 311
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.923213745946059e-05,
      "loss": 1.1822,
      "step": 312
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.922530194891969e-05,
      "loss": 1.0442,
      "step": 313
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.921843662697673e-05,
      "loss": 1.139,
      "step": 314
    },
    {
      "epoch": 2.16,
      "learning_rate": 4.921154150208012e-05,
      "loss": 0.8287,
      "step": 315
    },
    {
      "epoch": 2.16,
      "learning_rate": 4.920461658271492e-05,
      "loss": 1.1756,
      "step": 316
    },
    {
      "epoch": 2.17,
      "learning_rate": 4.919766187740286e-05,
      "loss": 1.1312,
      "step": 317
    },
    {
      "epoch": 2.18,
      "learning_rate": 4.9190677394702324e-05,
      "loss": 1.122,
      "step": 318
    },
    {
      "epoch": 2.18,
      "learning_rate": 4.9183663143208355e-05,
      "loss": 1.0514,
      "step": 319
    },
    {
      "epoch": 2.19,
      "learning_rate": 4.9176619131552604e-05,
      "loss": 1.1719,
      "step": 320
    },
    {
      "epoch": 2.2,
      "learning_rate": 4.916954536840337e-05,
      "loss": 1.0202,
      "step": 321
    },
    {
      "epoch": 2.21,
      "learning_rate": 4.916244186246555e-05,
      "loss": 1.1547,
      "step": 322
    },
    {
      "epoch": 2.21,
      "learning_rate": 4.9155308622480625e-05,
      "loss": 1.2142,
      "step": 323
    },
    {
      "epoch": 2.22,
      "learning_rate": 4.914814565722671e-05,
      "loss": 1.0676,
      "step": 324
    },
    {
      "epoch": 2.23,
      "learning_rate": 4.9140952975518465e-05,
      "loss": 1.0258,
      "step": 325
    },
    {
      "epoch": 2.23,
      "learning_rate": 4.9133730586207124e-05,
      "loss": 1.0105,
      "step": 326
    },
    {
      "epoch": 2.24,
      "learning_rate": 4.9133730586207124e-05,
      "loss": 0.9735,
      "step": 327
    },
    {
      "epoch": 2.25,
      "learning_rate": 4.912647849818049e-05,
      "loss": 1.1046,
      "step": 328
    },
    {
      "epoch": 2.25,
      "learning_rate": 4.91191967203629e-05,
      "loss": 1.1104,
      "step": 329
    },
    {
      "epoch": 2.26,
      "learning_rate": 4.911188526171524e-05,
      "loss": 0.9328,
      "step": 330
    },
    {
      "epoch": 2.27,
      "learning_rate": 4.910454413123491e-05,
      "loss": 1.1438,
      "step": 331
    },
    {
      "epoch": 2.27,
      "learning_rate": 4.909717333795584e-05,
      "loss": 1.1995,
      "step": 332
    },
    {
      "epoch": 2.28,
      "learning_rate": 4.908977289094843e-05,
      "loss": 1.057,
      "step": 333
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.90823427993196e-05,
      "loss": 1.1436,
      "step": 334
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.9074883072212744e-05,
      "loss": 1.0117,
      "step": 335
    },
    {
      "epoch": 2.3,
      "learning_rate": 4.906739371880773e-05,
      "loss": 1.1685,
      "step": 336
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.9059874748320876e-05,
      "loss": 1.1363,
      "step": 337
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.9052326170004936e-05,
      "loss": 1.1357,
      "step": 338
    },
    {
      "epoch": 2.32,
      "learning_rate": 4.904474799314913e-05,
      "loss": 0.974,
      "step": 339
    },
    {
      "epoch": 2.33,
      "learning_rate": 4.9037140227079065e-05,
      "loss": 1.2257,
      "step": 340
    },
    {
      "epoch": 2.34,
      "learning_rate": 4.902950288115679e-05,
      "loss": 1.1061,
      "step": 341
    },
    {
      "epoch": 2.34,
      "learning_rate": 4.902183596478073e-05,
      "loss": 1.0701,
      "step": 342
    },
    {
      "epoch": 2.35,
      "learning_rate": 4.901413948738572e-05,
      "loss": 1.2365,
      "step": 343
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.900641345844298e-05,
      "loss": 1.0575,
      "step": 344
    },
    {
      "epoch": 2.36,
      "learning_rate": 4.899865788746005e-05,
      "loss": 1.1325,
      "step": 345
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.899087278398088e-05,
      "loss": 1.1504,
      "step": 346
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.8983058157585717e-05,
      "loss": 1.0893,
      "step": 347
    },
    {
      "epoch": 2.38,
      "learning_rate": 4.897521401789116e-05,
      "loss": 1.1198,
      "step": 348
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.896734037455014e-05,
      "loss": 1.1958,
      "step": 349
    },
    {
      "epoch": 2.4,
      "learning_rate": 4.895943723725187e-05,
      "loss": 0.958,
      "step": 350
    },
    {
      "epoch": 2.4,
      "learning_rate": 4.895150461572187e-05,
      "loss": 1.0629,
      "step": 351
    },
    {
      "epoch": 2.41,
      "learning_rate": 4.894354251972193e-05,
      "loss": 1.0446,
      "step": 352
    },
    {
      "epoch": 2.42,
      "learning_rate": 4.893555095905014e-05,
      "loss": 0.9373,
      "step": 353
    },
    {
      "epoch": 2.42,
      "learning_rate": 4.892752994354082e-05,
      "loss": 1.11,
      "step": 354
    },
    {
      "epoch": 2.43,
      "learning_rate": 4.8919479483064544e-05,
      "loss": 1.0709,
      "step": 355
    },
    {
      "epoch": 2.44,
      "learning_rate": 4.891139958752813e-05,
      "loss": 1.2324,
      "step": 356
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.890329026687462e-05,
      "loss": 1.0608,
      "step": 357
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.889515153108324e-05,
      "loss": 1.1808,
      "step": 358
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.888698339016945e-05,
      "loss": 1.0626,
      "step": 359
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.887878585418487e-05,
      "loss": 1.1625,
      "step": 360
    },
    {
      "epoch": 2.47,
      "learning_rate": 4.88705589332173e-05,
      "loss": 1.0463,
      "step": 361
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.8862302637390714e-05,
      "loss": 0.9999,
      "step": 362
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.88540169768652e-05,
      "loss": 1.2328,
      "step": 363
    },
    {
      "epoch": 2.49,
      "learning_rate": 4.884570196183703e-05,
      "loss": 1.063,
      "step": 364
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.883735760253856e-05,
      "loss": 1.1949,
      "step": 365
    },
    {
      "epoch": 2.51,
      "learning_rate": 4.882898390923828e-05,
      "loss": 1.0521,
      "step": 366
    },
    {
      "epoch": 2.51,
      "learning_rate": 4.882058089224075e-05,
      "loss": 1.1375,
      "step": 367
    },
    {
      "epoch": 2.52,
      "learning_rate": 4.881214856188666e-05,
      "loss": 1.1363,
      "step": 368
    },
    {
      "epoch": 2.53,
      "learning_rate": 4.8803686928552736e-05,
      "loss": 1.0669,
      "step": 369
    },
    {
      "epoch": 2.53,
      "learning_rate": 4.879519600265177e-05,
      "loss": 1.0104,
      "step": 370
    },
    {
      "epoch": 2.54,
      "learning_rate": 4.8786675794632606e-05,
      "loss": 0.943,
      "step": 371
    },
    {
      "epoch": 2.55,
      "learning_rate": 4.8778126314980136e-05,
      "loss": 1.2132,
      "step": 372
    },
    {
      "epoch": 2.55,
      "learning_rate": 4.876954757421523e-05,
      "loss": 0.9231,
      "step": 373
    },
    {
      "epoch": 2.56,
      "learning_rate": 4.876093958289484e-05,
      "loss": 0.9422,
      "step": 374
    },
    {
      "epoch": 2.57,
      "learning_rate": 4.8752302351611836e-05,
      "loss": 1.2366,
      "step": 375
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.8743635890995124e-05,
      "loss": 0.904,
      "step": 376
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.873494021170953e-05,
      "loss": 1.0879,
      "step": 377
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.8726215324455905e-05,
      "loss": 0.9262,
      "step": 378
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.8717461239970975e-05,
      "loss": 1.0145,
      "step": 379
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.870867796902743e-05,
      "loss": 1.0186,
      "step": 380
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.8699865522433884e-05,
      "loss": 1.0292,
      "step": 381
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.8691023911034826e-05,
      "loss": 0.9748,
      "step": 382
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.868215314571065e-05,
      "loss": 0.7651,
      "step": 383
    },
    {
      "epoch": 2.63,
      "learning_rate": 4.867325323737765e-05,
      "loss": 1.0372,
      "step": 384
    },
    {
      "epoch": 2.64,
      "learning_rate": 4.866432419698792e-05,
      "loss": 1.0903,
      "step": 385
    },
    {
      "epoch": 2.64,
      "learning_rate": 4.8655366035529483e-05,
      "loss": 1.0043,
      "step": 386
    },
    {
      "epoch": 2.65,
      "learning_rate": 4.8646378764026126e-05,
      "loss": 1.0399,
      "step": 387
    },
    {
      "epoch": 2.66,
      "learning_rate": 4.863736239353752e-05,
      "loss": 1.1461,
      "step": 388
    },
    {
      "epoch": 2.66,
      "learning_rate": 4.862831693515909e-05,
      "loss": 1.0312,
      "step": 389
    },
    {
      "epoch": 2.67,
      "learning_rate": 4.8619242400022096e-05,
      "loss": 1.0527,
      "step": 390
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.861013879929357e-05,
      "loss": 0.9872,
      "step": 391
    },
    {
      "epoch": 2.68,
      "learning_rate": 4.8601006144176284e-05,
      "loss": 1.0633,
      "step": 392
    },
    {
      "epoch": 2.69,
      "learning_rate": 4.859184444590882e-05,
      "loss": 1.0893,
      "step": 393
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.859184444590882e-05,
      "loss": 1.0137,
      "step": 394
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.858265371576545e-05,
      "loss": 1.0704,
      "step": 395
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.857343396505618e-05,
      "loss": 1.2629,
      "step": 396
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.856418520512676e-05,
      "loss": 1.1081,
      "step": 397
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.85549074473586e-05,
      "loss": 0.9953,
      "step": 398
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.8545600703168824e-05,
      "loss": 1.0844,
      "step": 399
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.85362649840102e-05,
      "loss": 1.1951,
      "step": 400
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.852690030137118e-05,
      "loss": 1.018,
      "step": 401
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.851750666677584e-05,
      "loss": 1.0113,
      "step": 402
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.8508084091783874e-05,
      "loss": 1.039,
      "step": 403
    },
    {
      "epoch": 2.77,
      "learning_rate": 4.8498632587990625e-05,
      "loss": 0.953,
      "step": 404
    },
    {
      "epoch": 2.77,
      "learning_rate": 4.8489152167026995e-05,
      "loss": 1.1071,
      "step": 405
    },
    {
      "epoch": 2.78,
      "learning_rate": 4.8479642840559505e-05,
      "loss": 1.048,
      "step": 406
    },
    {
      "epoch": 2.79,
      "learning_rate": 4.847010462029022e-05,
      "loss": 1.0457,
      "step": 407
    },
    {
      "epoch": 2.79,
      "learning_rate": 4.8460537517956794e-05,
      "loss": 1.0223,
      "step": 408
    },
    {
      "epoch": 2.8,
      "learning_rate": 4.845094154533239e-05,
      "loss": 1.1077,
      "step": 409
    },
    {
      "epoch": 2.81,
      "learning_rate": 4.84413167142257e-05,
      "loss": 1.0166,
      "step": 410
    },
    {
      "epoch": 2.82,
      "learning_rate": 4.8431663036480955e-05,
      "loss": 1.0992,
      "step": 411
    },
    {
      "epoch": 2.82,
      "learning_rate": 4.8421980523977864e-05,
      "loss": 0.9356,
      "step": 412
    },
    {
      "epoch": 2.83,
      "learning_rate": 4.841226918863162e-05,
      "loss": 1.1566,
      "step": 413
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.840252904239291e-05,
      "loss": 1.1303,
      "step": 414
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.839276009724783e-05,
      "loss": 1.0215,
      "step": 415
    },
    {
      "epoch": 2.85,
      "learning_rate": 4.838296236521796e-05,
      "loss": 1.087,
      "step": 416
    },
    {
      "epoch": 2.86,
      "learning_rate": 4.837313585836027e-05,
      "loss": 0.8805,
      "step": 417
    },
    {
      "epoch": 2.86,
      "learning_rate": 4.836328058876717e-05,
      "loss": 1.0968,
      "step": 418
    },
    {
      "epoch": 2.87,
      "learning_rate": 4.8353396568566454e-05,
      "loss": 1.0483,
      "step": 419
    },
    {
      "epoch": 2.88,
      "learning_rate": 4.834348380992127e-05,
      "loss": 1.0528,
      "step": 420
    },
    {
      "epoch": 2.88,
      "learning_rate": 4.833354232503019e-05,
      "loss": 1.0813,
      "step": 421
    },
    {
      "epoch": 2.89,
      "learning_rate": 4.832357212612707e-05,
      "loss": 1.0018,
      "step": 422
    },
    {
      "epoch": 2.9,
      "learning_rate": 4.8313573225481143e-05,
      "loss": 1.088,
      "step": 423
    },
    {
      "epoch": 2.9,
      "learning_rate": 4.830354563539696e-05,
      "loss": 1.1499,
      "step": 424
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.8293489368214354e-05,
      "loss": 1.13,
      "step": 425
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.8283404436308464e-05,
      "loss": 0.952,
      "step": 426
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.8273290852089704e-05,
      "loss": 0.9332,
      "step": 427
    },
    {
      "epoch": 2.93,
      "learning_rate": 4.826314862800375e-05,
      "loss": 1.1237,
      "step": 428
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.825297777653151e-05,
      "loss": 0.874,
      "step": 429
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.824277831018913e-05,
      "loss": 1.2359,
      "step": 430
    },
    {
      "epoch": 2.95,
      "learning_rate": 4.823255024152796e-05,
      "loss": 1.0285,
      "step": 431
    },
    {
      "epoch": 2.96,
      "learning_rate": 4.822229358313456e-05,
      "loss": 0.9887,
      "step": 432
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.8212008347630664e-05,
      "loss": 1.1521,
      "step": 433
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.820169454767318e-05,
      "loss": 0.9784,
      "step": 434
    },
    {
      "epoch": 2.98,
      "learning_rate": 4.819135219595416e-05,
      "loss": 0.8608,
      "step": 435
    },
    {
      "epoch": 2.99,
      "learning_rate": 4.818098130520078e-05,
      "loss": 1.0483,
      "step": 436
    },
    {
      "epoch": 2.99,
      "learning_rate": 4.817058188817536e-05,
      "loss": 1.1391,
      "step": 437
    },
    {
      "epoch": 3.0,
      "learning_rate": 4.8160153957675316e-05,
      "loss": 0.8631,
      "step": 438
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.8149697526533145e-05,
      "loss": 1.0571,
      "step": 439
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.813921260761642e-05,
      "loss": 1.022,
      "step": 440
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.8128699213827785e-05,
      "loss": 1.1186,
      "step": 441
    },
    {
      "epoch": 3.03,
      "learning_rate": 4.81181573581049e-05,
      "loss": 1.1906,
      "step": 442
    },
    {
      "epoch": 3.03,
      "learning_rate": 4.8107587053420455e-05,
      "loss": 1.0026,
      "step": 443
    },
    {
      "epoch": 3.04,
      "learning_rate": 4.8096988312782174e-05,
      "loss": 1.011,
      "step": 444
    },
    {
      "epoch": 3.05,
      "learning_rate": 4.808636114923274e-05,
      "loss": 1.0215,
      "step": 445
    },
    {
      "epoch": 3.05,
      "learning_rate": 4.807570557584984e-05,
      "loss": 1.0448,
      "step": 446
    },
    {
      "epoch": 3.06,
      "learning_rate": 4.8065021605746104e-05,
      "loss": 0.7449,
      "step": 447
    },
    {
      "epoch": 3.07,
      "learning_rate": 4.8054309252069114e-05,
      "loss": 1.0893,
      "step": 448
    },
    {
      "epoch": 3.08,
      "learning_rate": 4.8043568528001384e-05,
      "loss": 1.1073,
      "step": 449
    },
    {
      "epoch": 3.08,
      "learning_rate": 4.803279944676032e-05,
      "loss": 1.1046,
      "step": 450
    },
    {
      "epoch": 3.09,
      "learning_rate": 4.802200202159827e-05,
      "loss": 0.8914,
      "step": 451
    },
    {
      "epoch": 3.1,
      "learning_rate": 4.8011176265802415e-05,
      "loss": 1.1037,
      "step": 452
    },
    {
      "epoch": 3.1,
      "learning_rate": 4.8000322192694814e-05,
      "loss": 0.9823,
      "step": 453
    },
    {
      "epoch": 3.11,
      "learning_rate": 4.798943981563238e-05,
      "loss": 0.9635,
      "step": 454
    },
    {
      "epoch": 3.12,
      "learning_rate": 4.797852914800684e-05,
      "loss": 1.0584,
      "step": 455
    },
    {
      "epoch": 3.12,
      "learning_rate": 4.796759020324477e-05,
      "loss": 1.0476,
      "step": 456
    },
    {
      "epoch": 3.13,
      "learning_rate": 4.7956622994807495e-05,
      "loss": 1.0305,
      "step": 457
    },
    {
      "epoch": 3.14,
      "learning_rate": 4.7945627536191166e-05,
      "loss": 1.1159,
      "step": 458
    },
    {
      "epoch": 3.14,
      "learning_rate": 4.7934603840926675e-05,
      "loss": 0.9567,
      "step": 459
    },
    {
      "epoch": 3.15,
      "learning_rate": 4.7923551922579656e-05,
      "loss": 1.0536,
      "step": 460
    },
    {
      "epoch": 3.16,
      "learning_rate": 4.7912471794750496e-05,
      "loss": 0.7428,
      "step": 461
    },
    {
      "epoch": 3.16,
      "learning_rate": 4.790136347107427e-05,
      "loss": 1.0945,
      "step": 462
    },
    {
      "epoch": 3.17,
      "learning_rate": 4.7890226965220785e-05,
      "loss": 1.0525,
      "step": 463
    },
    {
      "epoch": 3.18,
      "learning_rate": 4.787906229089448e-05,
      "loss": 1.0444,
      "step": 464
    },
    {
      "epoch": 3.18,
      "learning_rate": 4.7867869461834514e-05,
      "loss": 0.9727,
      "step": 465
    },
    {
      "epoch": 3.19,
      "learning_rate": 4.785664849181465e-05,
      "loss": 1.097,
      "step": 466
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.78453993946433e-05,
      "loss": 0.953,
      "step": 467
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.783412218416348e-05,
      "loss": 1.0736,
      "step": 468
    },
    {
      "epoch": 3.21,
      "learning_rate": 4.78228168742528e-05,
      "loss": 1.1363,
      "step": 469
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.781148347882347e-05,
      "loss": 0.9819,
      "step": 470
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.780012201182225e-05,
      "loss": 0.9414,
      "step": 471
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.7788732487230434e-05,
      "loss": 0.9277,
      "step": 472
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.777731491906384e-05,
      "loss": 0.9638,
      "step": 473
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.7765869321372836e-05,
      "loss": 1.0214,
      "step": 474
    },
    {
      "epoch": 3.25,
      "learning_rate": 4.7754395708242226e-05,
      "loss": 1.0234,
      "step": 475
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.774289409379133e-05,
      "loss": 0.8438,
      "step": 476
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.773136449217391e-05,
      "loss": 1.0616,
      "step": 477
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.771980691757819e-05,
      "loss": 1.1344,
      "step": 478
    },
    {
      "epoch": 3.28,
      "learning_rate": 4.770822138422677e-05,
      "loss": 0.9735,
      "step": 479
    },
    {
      "epoch": 3.29,
      "learning_rate": 4.769660790637671e-05,
      "loss": 1.061,
      "step": 480
    },
    {
      "epoch": 3.29,
      "learning_rate": 4.768496649831942e-05,
      "loss": 0.9322,
      "step": 481
    },
    {
      "epoch": 3.3,
      "learning_rate": 4.767329717438071e-05,
      "loss": 1.0904,
      "step": 482
    },
    {
      "epoch": 3.31,
      "learning_rate": 4.76615999489207e-05,
      "loss": 1.0558,
      "step": 483
    },
    {
      "epoch": 3.32,
      "learning_rate": 4.76498748363339e-05,
      "loss": 1.0602,
      "step": 484
    },
    {
      "epoch": 3.32,
      "learning_rate": 4.76381218510491e-05,
      "loss": 0.9079,
      "step": 485
    },
    {
      "epoch": 3.33,
      "learning_rate": 4.76263410075294e-05,
      "loss": 1.1412,
      "step": 486
    },
    {
      "epoch": 3.34,
      "learning_rate": 4.7614532320272174e-05,
      "loss": 1.0225,
      "step": 487
    },
    {
      "epoch": 3.34,
      "learning_rate": 4.760269580380909e-05,
      "loss": 1.0012,
      "step": 488
    },
    {
      "epoch": 3.35,
      "learning_rate": 4.759083147270602e-05,
      "loss": 1.1793,
      "step": 489
    },
    {
      "epoch": 3.36,
      "learning_rate": 4.7578939341563095e-05,
      "loss": 0.9858,
      "step": 490
    },
    {
      "epoch": 3.36,
      "learning_rate": 4.7567019425014644e-05,
      "loss": 1.059,
      "step": 491
    },
    {
      "epoch": 3.37,
      "learning_rate": 4.755507173772919e-05,
      "loss": 1.0591,
      "step": 492
    },
    {
      "epoch": 3.38,
      "learning_rate": 4.754309629440943e-05,
      "loss": 1.0188,
      "step": 493
    },
    {
      "epoch": 3.38,
      "learning_rate": 4.753109310979224e-05,
      "loss": 1.0426,
      "step": 494
    },
    {
      "epoch": 3.39,
      "learning_rate": 4.751906219864857e-05,
      "loss": 1.1116,
      "step": 495
    },
    {
      "epoch": 3.4,
      "learning_rate": 4.750700357578357e-05,
      "loss": 0.8753,
      "step": 496
    },
    {
      "epoch": 3.4,
      "learning_rate": 4.749491725603644e-05,
      "loss": 0.9712,
      "step": 497
    },
    {
      "epoch": 3.41,
      "learning_rate": 4.7482803254280484e-05,
      "loss": 0.9633,
      "step": 498
    },
    {
      "epoch": 3.42,
      "learning_rate": 4.747066158542306e-05,
      "loss": 0.8521,
      "step": 499
    },
    {
      "epoch": 3.42,
      "learning_rate": 4.7458492264405574e-05,
      "loss": 1.0322,
      "step": 500
    },
    {
      "epoch": 3.43,
      "learning_rate": 4.7446295306203474e-05,
      "loss": 0.9904,
      "step": 501
    },
    {
      "epoch": 3.44,
      "learning_rate": 4.743407072582621e-05,
      "loss": 1.1605,
      "step": 502
    },
    {
      "epoch": 3.45,
      "learning_rate": 4.742181853831721e-05,
      "loss": 0.9827,
      "step": 503
    },
    {
      "epoch": 3.45,
      "learning_rate": 4.74095387587539e-05,
      "loss": 1.0952,
      "step": 504
    },
    {
      "epoch": 3.46,
      "learning_rate": 4.739723140224763e-05,
      "loss": 0.9814,
      "step": 505
    },
    {
      "epoch": 3.47,
      "learning_rate": 4.738489648394373e-05,
      "loss": 1.0741,
      "step": 506
    },
    {
      "epoch": 3.47,
      "learning_rate": 4.73725340190214e-05,
      "loss": 0.9566,
      "step": 507
    },
    {
      "epoch": 3.48,
      "learning_rate": 4.736014402269376e-05,
      "loss": 0.913,
      "step": 508
    },
    {
      "epoch": 3.49,
      "learning_rate": 4.734772651020782e-05,
      "loss": 1.151,
      "step": 509
    },
    {
      "epoch": 3.49,
      "learning_rate": 4.733528149684444e-05,
      "loss": 0.9932,
      "step": 510
    },
    {
      "epoch": 3.5,
      "learning_rate": 4.732280899791832e-05,
      "loss": 1.1216,
      "step": 511
    },
    {
      "epoch": 3.51,
      "learning_rate": 4.7310309028777976e-05,
      "loss": 0.9726,
      "step": 512
    },
    {
      "epoch": 3.51,
      "learning_rate": 4.729778160480576e-05,
      "loss": 1.0382,
      "step": 513
    },
    {
      "epoch": 3.52,
      "learning_rate": 4.728522674141776e-05,
      "loss": 1.0765,
      "step": 514
    },
    {
      "epoch": 3.53,
      "learning_rate": 4.727264445406388e-05,
      "loss": 1.0005,
      "step": 515
    },
    {
      "epoch": 3.53,
      "learning_rate": 4.726003475822775e-05,
      "loss": 0.931,
      "step": 516
    },
    {
      "epoch": 3.54,
      "learning_rate": 4.7247397669426716e-05,
      "loss": 0.8656,
      "step": 517
    },
    {
      "epoch": 3.55,
      "learning_rate": 4.723473320321186e-05,
      "loss": 1.1407,
      "step": 518
    },
    {
      "epoch": 3.55,
      "learning_rate": 4.7222041375167936e-05,
      "loss": 0.845,
      "step": 519
    },
    {
      "epoch": 3.56,
      "learning_rate": 4.720932220091337e-05,
      "loss": 0.8612,
      "step": 520
    },
    {
      "epoch": 3.57,
      "learning_rate": 4.7196575696100254e-05,
      "loss": 1.1551,
      "step": 521
    },
    {
      "epoch": 3.58,
      "learning_rate": 4.7183801876414294e-05,
      "loss": 0.8392,
      "step": 522
    },
    {
      "epoch": 3.58,
      "learning_rate": 4.717100075757482e-05,
      "loss": 1.0083,
      "step": 523
    },
    {
      "epoch": 3.59,
      "learning_rate": 4.715817235533476e-05,
      "loss": 0.8475,
      "step": 524
    },
    {
      "epoch": 3.6,
      "learning_rate": 4.71453166854806e-05,
      "loss": 0.936,
      "step": 525
    },
    {
      "epoch": 3.6,
      "learning_rate": 4.7132433763832404e-05,
      "loss": 0.9306,
      "step": 526
    },
    {
      "epoch": 3.61,
      "learning_rate": 4.711952360624376e-05,
      "loss": 0.9495,
      "step": 527
    },
    {
      "epoch": 3.62,
      "learning_rate": 4.710658622860176e-05,
      "loss": 0.8872,
      "step": 528
    },
    {
      "epoch": 3.62,
      "learning_rate": 4.709362164682702e-05,
      "loss": 0.6822,
      "step": 529
    },
    {
      "epoch": 3.63,
      "learning_rate": 4.70806298768736e-05,
      "loss": 0.9525,
      "step": 530
    },
    {
      "epoch": 3.64,
      "learning_rate": 4.706761093472906e-05,
      "loss": 1.0159,
      "step": 531
    },
    {
      "epoch": 3.64,
      "learning_rate": 4.705456483641435e-05,
      "loss": 0.9192,
      "step": 532
    },
    {
      "epoch": 3.65,
      "learning_rate": 4.704149159798387e-05,
      "loss": 0.9538,
      "step": 533
    },
    {
      "epoch": 3.66,
      "learning_rate": 4.702839123552541e-05,
      "loss": 1.0676,
      "step": 534
    },
    {
      "epoch": 3.66,
      "learning_rate": 4.7015263765160154e-05,
      "loss": 0.9409,
      "step": 535
    },
    {
      "epoch": 3.67,
      "learning_rate": 4.70021092030426e-05,
      "loss": 0.9745,
      "step": 536
    },
    {
      "epoch": 3.68,
      "learning_rate": 4.698892756536064e-05,
      "loss": 0.8958,
      "step": 537
    },
    {
      "epoch": 3.68,
      "learning_rate": 4.697571886833544e-05,
      "loss": 1.0001,
      "step": 538
    },
    {
      "epoch": 3.69,
      "learning_rate": 4.696248312822149e-05,
      "loss": 1.026,
      "step": 539
    },
    {
      "epoch": 3.7,
      "learning_rate": 4.6949220361306555e-05,
      "loss": 1.0302,
      "step": 540
    },
    {
      "epoch": 3.71,
      "learning_rate": 4.693593058391165e-05,
      "loss": 0.9987,
      "step": 541
    },
    {
      "epoch": 3.71,
      "learning_rate": 4.692261381239105e-05,
      "loss": 1.1951,
      "step": 542
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.690927006313222e-05,
      "loss": 1.0323,
      "step": 543
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.689589935255585e-05,
      "loss": 0.9272,
      "step": 544
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.688250169711578e-05,
      "loss": 0.9996,
      "step": 545
    },
    {
      "epoch": 3.74,
      "learning_rate": 4.686907711329903e-05,
      "loss": 1.1128,
      "step": 546
    },
    {
      "epoch": 3.75,
      "learning_rate": 4.6855625617625763e-05,
      "loss": 0.9461,
      "step": 547
    },
    {
      "epoch": 3.75,
      "learning_rate": 4.684214722664924e-05,
      "loss": 0.9449,
      "step": 548
    },
    {
      "epoch": 3.76,
      "learning_rate": 4.682864195695582e-05,
      "loss": 0.9528,
      "step": 549
    },
    {
      "epoch": 3.77,
      "learning_rate": 4.681510982516496e-05,
      "loss": 0.8721,
      "step": 550
    },
    {
      "epoch": 3.77,
      "learning_rate": 4.680155084792914e-05,
      "loss": 1.0428,
      "step": 551
    },
    {
      "epoch": 3.78,
      "learning_rate": 4.678796504193392e-05,
      "loss": 1.0111,
      "step": 552
    },
    {
      "epoch": 3.79,
      "learning_rate": 4.677435242389784e-05,
      "loss": 0.9557,
      "step": 553
    },
    {
      "epoch": 3.79,
      "learning_rate": 4.676071301057243e-05,
      "loss": 0.9475,
      "step": 554
    },
    {
      "epoch": 3.8,
      "learning_rate": 4.674704681874223e-05,
      "loss": 1.039,
      "step": 555
    },
    {
      "epoch": 3.81,
      "learning_rate": 4.6733353865224694e-05,
      "loss": 0.9335,
      "step": 556
    },
    {
      "epoch": 3.82,
      "learning_rate": 4.671963416687024e-05,
      "loss": 1.0222,
      "step": 557
    },
    {
      "epoch": 3.82,
      "learning_rate": 4.670588774056218e-05,
      "loss": 0.8481,
      "step": 558
    },
    {
      "epoch": 3.83,
      "learning_rate": 4.669211460321673e-05,
      "loss": 1.0872,
      "step": 559
    },
    {
      "epoch": 3.84,
      "learning_rate": 4.667831477178295e-05,
      "loss": 1.0565,
      "step": 560
    },
    {
      "epoch": 3.84,
      "learning_rate": 4.666448826324278e-05,
      "loss": 0.9574,
      "step": 561
    },
    {
      "epoch": 3.85,
      "learning_rate": 4.665063509461097e-05,
      "loss": 1.0013,
      "step": 562
    },
    {
      "epoch": 3.86,
      "learning_rate": 4.663675528293509e-05,
      "loss": 0.807,
      "step": 563
    },
    {
      "epoch": 3.86,
      "learning_rate": 4.662284884529549e-05,
      "loss": 1.0153,
      "step": 564
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.660891579880528e-05,
      "loss": 0.9717,
      "step": 565
    },
    {
      "epoch": 3.88,
      "learning_rate": 4.6594956160610325e-05,
      "loss": 0.9686,
      "step": 566
    },
    {
      "epoch": 3.88,
      "learning_rate": 4.6580969947889216e-05,
      "loss": 1.0015,
      "step": 567
    },
    {
      "epoch": 3.89,
      "learning_rate": 4.656695717785323e-05,
      "loss": 0.9288,
      "step": 568
    },
    {
      "epoch": 3.9,
      "learning_rate": 4.6552917867746324e-05,
      "loss": 1.0121,
      "step": 569
    },
    {
      "epoch": 3.9,
      "learning_rate": 4.653885203484515e-05,
      "loss": 1.0764,
      "step": 570
    },
    {
      "epoch": 3.91,
      "learning_rate": 4.652475969645896e-05,
      "loss": 1.0444,
      "step": 571
    },
    {
      "epoch": 3.92,
      "learning_rate": 4.651064086992965e-05,
      "loss": 0.8733,
      "step": 572
    },
    {
      "epoch": 3.92,
      "learning_rate": 4.6496495572631675e-05,
      "loss": 0.8512,
      "step": 573
    },
    {
      "epoch": 3.93,
      "learning_rate": 4.6482323821972105e-05,
      "loss": 1.0505,
      "step": 574
    },
    {
      "epoch": 3.94,
      "learning_rate": 4.6468125635390556e-05,
      "loss": 0.7957,
      "step": 575
    },
    {
      "epoch": 3.95,
      "learning_rate": 4.6453901030359154e-05,
      "loss": 1.1614,
      "step": 576
    },
    {
      "epoch": 3.95,
      "learning_rate": 4.6439650024382547e-05,
      "loss": 0.9545,
      "step": 577
    },
    {
      "epoch": 3.96,
      "learning_rate": 4.6439650024382547e-05,
      "loss": 0.9162,
      "step": 578
    },
    {
      "epoch": 3.97,
      "learning_rate": 4.642537263499788e-05,
      "loss": 1.0761,
      "step": 579
    },
    {
      "epoch": 3.97,
      "learning_rate": 4.6411068879774754e-05,
      "loss": 0.9022,
      "step": 580
    },
    {
      "epoch": 3.98,
      "learning_rate": 4.639673877631523e-05,
      "loss": 0.7747,
      "step": 581
    },
    {
      "epoch": 3.99,
      "learning_rate": 4.6382382342253775e-05,
      "loss": 0.9697,
      "step": 582
    },
    {
      "epoch": 3.99,
      "learning_rate": 4.636799959525726e-05,
      "loss": 1.0663,
      "step": 583
    },
    {
      "epoch": 4.0,
      "learning_rate": 4.635359055302495e-05,
      "loss": 0.7292,
      "step": 584
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.633915523328846e-05,
      "loss": 0.9178,
      "step": 585
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.632469365381174e-05,
      "loss": 0.9554,
      "step": 586
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.631020583239107e-05,
      "loss": 1.0515,
      "step": 587
    },
    {
      "epoch": 4.03,
      "learning_rate": 4.629569178685499e-05,
      "loss": 1.1199,
      "step": 588
    },
    {
      "epoch": 4.03,
      "learning_rate": 4.628115153506435e-05,
      "loss": 0.9295,
      "step": 589
    },
    {
      "epoch": 4.04,
      "learning_rate": 4.626658509491222e-05,
      "loss": 0.9321,
      "step": 590
    },
    {
      "epoch": 4.05,
      "learning_rate": 4.625199248432391e-05,
      "loss": 0.9355,
      "step": 591
    },
    {
      "epoch": 4.05,
      "learning_rate": 4.6237373721256935e-05,
      "loss": 0.9687,
      "step": 592
    },
    {
      "epoch": 4.06,
      "learning_rate": 4.6222728823700995e-05,
      "loss": 0.6614,
      "step": 593
    },
    {
      "epoch": 4.07,
      "learning_rate": 4.620805780967794e-05,
      "loss": 1.0145,
      "step": 594
    },
    {
      "epoch": 4.08,
      "learning_rate": 4.619336069724177e-05,
      "loss": 1.0242,
      "step": 595
    },
    {
      "epoch": 4.08,
      "learning_rate": 4.617863750447861e-05,
      "loss": 1.0289,
      "step": 596
    },
    {
      "epoch": 4.09,
      "learning_rate": 4.616388824950664e-05,
      "loss": 0.8058,
      "step": 597
    },
    {
      "epoch": 4.1,
      "learning_rate": 4.614911295047615e-05,
      "loss": 1.0225,
      "step": 598
    },
    {
      "epoch": 4.1,
      "learning_rate": 4.613431162556947e-05,
      "loss": 0.8832,
      "step": 599
    },
    {
      "epoch": 4.11,
      "learning_rate": 4.6119484293000955e-05,
      "loss": 0.8861,
      "step": 600
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.610463097101696e-05,
      "loss": 0.9833,
      "step": 601
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.6089751677895836e-05,
      "loss": 0.9703,
      "step": 602
    },
    {
      "epoch": 4.13,
      "learning_rate": 4.607484643194788e-05,
      "loss": 0.9511,
      "step": 603
    },
    {
      "epoch": 4.14,
      "learning_rate": 4.605991525151533e-05,
      "loss": 1.0516,
      "step": 604
    },
    {
      "epoch": 4.14,
      "learning_rate": 4.604495815497234e-05,
      "loss": 0.8705,
      "step": 605
    },
    {
      "epoch": 4.15,
      "learning_rate": 4.6029975160724945e-05,
      "loss": 0.9725,
      "step": 606
    },
    {
      "epoch": 4.16,
      "learning_rate": 4.6014966287211084e-05,
      "loss": 0.6535,
      "step": 607
    },
    {
      "epoch": 4.16,
      "learning_rate": 4.59999315529005e-05,
      "loss": 1.013,
      "step": 608
    },
    {
      "epoch": 4.17,
      "learning_rate": 4.598487097629479e-05,
      "loss": 0.9731,
      "step": 609
    },
    {
      "epoch": 4.18,
      "learning_rate": 4.5969784575927324e-05,
      "loss": 0.9715,
      "step": 610
    },
    {
      "epoch": 4.18,
      "learning_rate": 4.595467237036329e-05,
      "loss": 0.898,
      "step": 611
    },
    {
      "epoch": 4.19,
      "learning_rate": 4.59395343781996e-05,
      "loss": 1.0253,
      "step": 612
    },
    {
      "epoch": 4.2,
      "learning_rate": 4.5924370618064913e-05,
      "loss": 0.8818,
      "step": 613
    },
    {
      "epoch": 4.21,
      "learning_rate": 4.590918110861958e-05,
      "loss": 0.9952,
      "step": 614
    },
    {
      "epoch": 4.21,
      "learning_rate": 4.5893965868555664e-05,
      "loss": 1.0598,
      "step": 615
    },
    {
      "epoch": 4.22,
      "learning_rate": 4.5878724916596874e-05,
      "loss": 0.9002,
      "step": 616
    },
    {
      "epoch": 4.23,
      "learning_rate": 4.586345827149856e-05,
      "loss": 0.8679,
      "step": 617
    },
    {
      "epoch": 4.23,
      "learning_rate": 4.58481659520477e-05,
      "loss": 0.8458,
      "step": 618
    },
    {
      "epoch": 4.24,
      "learning_rate": 4.5832847977062874e-05,
      "loss": 0.8581,
      "step": 619
    },
    {
      "epoch": 4.25,
      "learning_rate": 4.581750436539421e-05,
      "loss": 0.9438,
      "step": 620
    },
    {
      "epoch": 4.25,
      "learning_rate": 4.5802135135923386e-05,
      "loss": 0.9411,
      "step": 621
    },
    {
      "epoch": 4.26,
      "learning_rate": 4.5786740307563636e-05,
      "loss": 0.7669,
      "step": 622
    },
    {
      "epoch": 4.27,
      "learning_rate": 4.5771319899259656e-05,
      "loss": 0.9775,
      "step": 623
    },
    {
      "epoch": 4.27,
      "learning_rate": 4.5755873929987634e-05,
      "loss": 1.0761,
      "step": 624
    },
    {
      "epoch": 4.28,
      "learning_rate": 4.5740402418755246e-05,
      "loss": 0.8978,
      "step": 625
    },
    {
      "epoch": 4.29,
      "learning_rate": 4.572490538460154e-05,
      "loss": 0.9787,
      "step": 626
    },
    {
      "epoch": 4.29,
      "learning_rate": 4.570938284659702e-05,
      "loss": 0.8622,
      "step": 627
    },
    {
      "epoch": 4.3,
      "learning_rate": 4.5693834823843556e-05,
      "loss": 1.0193,
      "step": 628
    },
    {
      "epoch": 4.31,
      "learning_rate": 4.5678261335474384e-05,
      "loss": 0.9681,
      "step": 629
    },
    {
      "epoch": 4.32,
      "learning_rate": 4.566266240065406e-05,
      "loss": 0.9882,
      "step": 630
    },
    {
      "epoch": 4.32,
      "learning_rate": 4.564703803857849e-05,
      "loss": 0.8362,
      "step": 631
    },
    {
      "epoch": 4.33,
      "learning_rate": 4.5631388268474837e-05,
      "loss": 1.0627,
      "step": 632
    },
    {
      "epoch": 4.34,
      "learning_rate": 4.5615713109601544e-05,
      "loss": 0.9333,
      "step": 633
    },
    {
      "epoch": 4.34,
      "learning_rate": 4.56000125812483e-05,
      "loss": 0.93,
      "step": 634
    },
    {
      "epoch": 4.35,
      "learning_rate": 4.558428670273601e-05,
      "loss": 1.1188,
      "step": 635
    },
    {
      "epoch": 4.36,
      "learning_rate": 4.556853549341679e-05,
      "loss": 0.9112,
      "step": 636
    },
    {
      "epoch": 4.36,
      "learning_rate": 4.555275897267388e-05,
      "loss": 0.9868,
      "step": 637
    },
    {
      "epoch": 4.37,
      "learning_rate": 4.553695715992172e-05,
      "loss": 0.974,
      "step": 638
    },
    {
      "epoch": 4.38,
      "learning_rate": 4.552113007460586e-05,
      "loss": 0.952,
      "step": 639
    },
    {
      "epoch": 4.38,
      "learning_rate": 4.550527773620293e-05,
      "loss": 0.9677,
      "step": 640
    },
    {
      "epoch": 4.39,
      "learning_rate": 4.5489400164220666e-05,
      "loss": 1.0329,
      "step": 641
    },
    {
      "epoch": 4.4,
      "learning_rate": 4.5473497378197835e-05,
      "loss": 0.7958,
      "step": 642
    },
    {
      "epoch": 4.4,
      "learning_rate": 4.545756939770423e-05,
      "loss": 0.8831,
      "step": 643
    },
    {
      "epoch": 4.41,
      "learning_rate": 4.5441616242340665e-05,
      "loss": 0.8805,
      "step": 644
    },
    {
      "epoch": 4.42,
      "learning_rate": 4.542563793173893e-05,
      "loss": 0.7746,
      "step": 645
    },
    {
      "epoch": 4.42,
      "learning_rate": 4.540963448556176e-05,
      "loss": 0.9577,
      "step": 646
    },
    {
      "epoch": 4.43,
      "learning_rate": 4.539360592350282e-05,
      "loss": 0.9146,
      "step": 647
    },
    {
      "epoch": 4.44,
      "learning_rate": 4.537755226528671e-05,
      "loss": 1.1038,
      "step": 648
    },
    {
      "epoch": 4.45,
      "learning_rate": 4.5361473530668874e-05,
      "loss": 0.9022,
      "step": 649
    },
    {
      "epoch": 4.45,
      "learning_rate": 4.534536973943564e-05,
      "loss": 1.0197,
      "step": 650
    },
    {
      "epoch": 4.46,
      "learning_rate": 4.532924091140417e-05,
      "loss": 0.9027,
      "step": 651
    },
    {
      "epoch": 4.47,
      "learning_rate": 4.531308706642243e-05,
      "loss": 0.9918,
      "step": 652
    },
    {
      "epoch": 4.47,
      "learning_rate": 4.529690822436916e-05,
      "loss": 0.875,
      "step": 653
    },
    {
      "epoch": 4.48,
      "learning_rate": 4.528070440515388e-05,
      "loss": 0.8319,
      "step": 654
    },
    {
      "epoch": 4.49,
      "learning_rate": 4.526447562871685e-05,
      "loss": 1.0758,
      "step": 655
    },
    {
      "epoch": 4.49,
      "learning_rate": 4.5248221915029014e-05,
      "loss": 0.9287,
      "step": 656
    },
    {
      "epoch": 4.5,
      "learning_rate": 4.523194328409203e-05,
      "loss": 1.047,
      "step": 657
    },
    {
      "epoch": 4.51,
      "learning_rate": 4.5215639755938214e-05,
      "loss": 0.9021,
      "step": 658
    },
    {
      "epoch": 4.51,
      "learning_rate": 4.519931135063051e-05,
      "loss": 0.9434,
      "step": 659
    },
    {
      "epoch": 4.52,
      "learning_rate": 4.518295808826249e-05,
      "loss": 1.0068,
      "step": 660
    },
    {
      "epoch": 4.53,
      "learning_rate": 4.5166579988958296e-05,
      "loss": 0.9381,
      "step": 661
    },
    {
      "epoch": 4.53,
      "learning_rate": 4.515017707287265e-05,
      "loss": 0.8563,
      "step": 662
    },
    {
      "epoch": 4.54,
      "learning_rate": 4.5133749360190805e-05,
      "loss": 0.7951,
      "step": 663
    },
    {
      "epoch": 4.55,
      "learning_rate": 4.5117296871128546e-05,
      "loss": 1.0672,
      "step": 664
    },
    {
      "epoch": 4.55,
      "learning_rate": 4.510081962593211e-05,
      "loss": 0.7677,
      "step": 665
    },
    {
      "epoch": 4.56,
      "learning_rate": 4.508431764487824e-05,
      "loss": 0.774,
      "step": 666
    },
    {
      "epoch": 4.57,
      "learning_rate": 4.5067790948274094e-05,
      "loss": 1.0797,
      "step": 667
    },
    {
      "epoch": 4.58,
      "learning_rate": 4.5051239556457244e-05,
      "loss": 0.7631,
      "step": 668
    },
    {
      "epoch": 4.58,
      "learning_rate": 4.503466348979568e-05,
      "loss": 0.9337,
      "step": 669
    },
    {
      "epoch": 4.59,
      "learning_rate": 4.501806276868772e-05,
      "loss": 0.7778,
      "step": 670
    },
    {
      "epoch": 4.6,
      "learning_rate": 4.500143741356203e-05,
      "loss": 0.8628,
      "step": 671
    },
    {
      "epoch": 4.6,
      "learning_rate": 4.4984787444877616e-05,
      "loss": 0.8446,
      "step": 672
    },
    {
      "epoch": 4.61,
      "learning_rate": 4.4968112883123734e-05,
      "loss": 0.8722,
      "step": 673
    },
    {
      "epoch": 4.62,
      "learning_rate": 4.495141374881995e-05,
      "loss": 0.8005,
      "step": 674
    },
    {
      "epoch": 4.62,
      "learning_rate": 4.493469006251601e-05,
      "loss": 0.5906,
      "step": 675
    },
    {
      "epoch": 4.63,
      "learning_rate": 4.491794184479194e-05,
      "loss": 0.8642,
      "step": 676
    },
    {
      "epoch": 4.64,
      "learning_rate": 4.49011691162579e-05,
      "loss": 0.9482,
      "step": 677
    },
    {
      "epoch": 4.64,
      "learning_rate": 4.488437189755424e-05,
      "loss": 0.8409,
      "step": 678
    },
    {
      "epoch": 4.65,
      "learning_rate": 4.486755020935144e-05,
      "loss": 0.8732,
      "step": 679
    },
    {
      "epoch": 4.66,
      "learning_rate": 4.485070407235009e-05,
      "loss": 0.9934,
      "step": 680
    },
    {
      "epoch": 4.66,
      "learning_rate": 4.4833833507280884e-05,
      "loss": 0.858,
      "step": 681
    },
    {
      "epoch": 4.67,
      "learning_rate": 4.481693853490454e-05,
      "loss": 0.8971,
      "step": 682
    },
    {
      "epoch": 4.68,
      "learning_rate": 4.480001917601185e-05,
      "loss": 0.8115,
      "step": 683
    },
    {
      "epoch": 4.68,
      "learning_rate": 4.478307545142359e-05,
      "loss": 0.9325,
      "step": 684
    },
    {
      "epoch": 4.69,
      "learning_rate": 4.476610738199053e-05,
      "loss": 0.956,
      "step": 685
    },
    {
      "epoch": 4.7,
      "learning_rate": 4.4749114988593396e-05,
      "loss": 0.9499,
      "step": 686
    },
    {
      "epoch": 4.71,
      "learning_rate": 4.473209829214286e-05,
      "loss": 0.9287,
      "step": 687
    },
    {
      "epoch": 4.71,
      "learning_rate": 4.471505731357949e-05,
      "loss": 1.1292,
      "step": 688
    },
    {
      "epoch": 4.72,
      "learning_rate": 4.4697992073873724e-05,
      "loss": 0.9526,
      "step": 689
    },
    {
      "epoch": 4.73,
      "learning_rate": 4.468090259402587e-05,
      "loss": 0.8573,
      "step": 690
    },
    {
      "epoch": 4.73,
      "learning_rate": 4.466378889506607e-05,
      "loss": 0.9222,
      "step": 691
    },
    {
      "epoch": 4.74,
      "learning_rate": 4.464665099805424e-05,
      "loss": 1.0377,
      "step": 692
    },
    {
      "epoch": 4.75,
      "learning_rate": 4.462948892408012e-05,
      "loss": 0.8685,
      "step": 693
    },
    {
      "epoch": 4.75,
      "learning_rate": 4.4612302694263174e-05,
      "loss": 0.8759,
      "step": 694
    },
    {
      "epoch": 4.76,
      "learning_rate": 4.4595092329752583e-05,
      "loss": 0.8718,
      "step": 695
    },
    {
      "epoch": 4.77,
      "learning_rate": 4.457785785172726e-05,
      "loss": 0.7879,
      "step": 696
    },
    {
      "epoch": 4.77,
      "learning_rate": 4.4560599281395755e-05,
      "loss": 0.9711,
      "step": 697
    },
    {
      "epoch": 4.78,
      "learning_rate": 4.454331663999629e-05,
      "loss": 0.9297,
      "step": 698
    },
    {
      "epoch": 4.79,
      "learning_rate": 4.4526009948796703e-05,
      "loss": 0.8714,
      "step": 699
    },
    {
      "epoch": 4.79,
      "learning_rate": 4.4508679229094425e-05,
      "loss": 0.8773,
      "step": 700
    },
    {
      "epoch": 4.8,
      "learning_rate": 4.449132450221646e-05,
      "loss": 0.973,
      "step": 701
    },
    {
      "epoch": 4.81,
      "learning_rate": 4.447394578951935e-05,
      "loss": 0.8554,
      "step": 702
    },
    {
      "epoch": 4.82,
      "learning_rate": 4.445654311238915e-05,
      "loss": 0.955,
      "step": 703
    },
    {
      "epoch": 4.82,
      "learning_rate": 4.443911649224143e-05,
      "loss": 0.7657,
      "step": 704
    },
    {
      "epoch": 4.83,
      "learning_rate": 4.442166595052118e-05,
      "loss": 1.0131,
      "step": 705
    },
    {
      "epoch": 4.84,
      "learning_rate": 4.4404191508702875e-05,
      "loss": 0.9818,
      "step": 706
    },
    {
      "epoch": 4.84,
      "learning_rate": 4.4386693188290376e-05,
      "loss": 0.8895,
      "step": 707
    },
    {
      "epoch": 4.85,
      "learning_rate": 4.4369171010816925e-05,
      "loss": 0.9271,
      "step": 708
    },
    {
      "epoch": 4.86,
      "learning_rate": 4.435162499784513e-05,
      "loss": 0.7359,
      "step": 709
    },
    {
      "epoch": 4.86,
      "learning_rate": 4.433405517096693e-05,
      "loss": 0.9381,
      "step": 710
    },
    {
      "epoch": 4.87,
      "learning_rate": 4.431646155180358e-05,
      "loss": 0.899,
      "step": 711
    },
    {
      "epoch": 4.88,
      "learning_rate": 4.4298844162005585e-05,
      "loss": 0.8745,
      "step": 712
    },
    {
      "epoch": 4.88,
      "learning_rate": 4.4281203023252735e-05,
      "loss": 0.9215,
      "step": 713
    },
    {
      "epoch": 4.89,
      "learning_rate": 4.426353815725403e-05,
      "loss": 0.8616,
      "step": 714
    },
    {
      "epoch": 4.9,
      "learning_rate": 4.4245849585747654e-05,
      "loss": 0.9401,
      "step": 715
    },
    {
      "epoch": 4.9,
      "learning_rate": 4.422813733050099e-05,
      "loss": 1.0099,
      "step": 716
    },
    {
      "epoch": 4.91,
      "learning_rate": 4.4210401413310556e-05,
      "loss": 0.9682,
      "step": 717
    },
    {
      "epoch": 4.92,
      "learning_rate": 4.4192641856001976e-05,
      "loss": 0.7944,
      "step": 718
    },
    {
      "epoch": 4.92,
      "learning_rate": 4.417485868042998e-05,
      "loss": 0.7715,
      "step": 719
    },
    {
      "epoch": 4.93,
      "learning_rate": 4.415705190847835e-05,
      "loss": 0.9758,
      "step": 720
    },
    {
      "epoch": 4.94,
      "learning_rate": 4.413922156205992e-05,
      "loss": 0.7187,
      "step": 721
    },
    {
      "epoch": 4.95,
      "learning_rate": 4.412136766311652e-05,
      "loss": 1.0898,
      "step": 722
    },
    {
      "epoch": 4.95,
      "learning_rate": 4.410349023361898e-05,
      "loss": 0.8747,
      "step": 723
    },
    {
      "epoch": 4.96,
      "learning_rate": 4.4085589295567065e-05,
      "loss": 0.9458,
      "step": 724
    },
    {
      "epoch": 4.97,
      "learning_rate": 4.406766487098949e-05,
      "loss": 1.0059,
      "step": 725
    },
    {
      "epoch": 4.97,
      "learning_rate": 4.4049716981943866e-05,
      "loss": 0.8319,
      "step": 726
    },
    {
      "epoch": 4.98,
      "learning_rate": 4.4031745650516666e-05,
      "loss": 0.7012,
      "step": 727
    },
    {
      "epoch": 4.99,
      "learning_rate": 4.401375089882324e-05,
      "loss": 0.9013,
      "step": 728
    },
    {
      "epoch": 4.99,
      "learning_rate": 4.399573274900771e-05,
      "loss": 0.9979,
      "step": 729
    },
    {
      "epoch": 5.0,
      "learning_rate": 4.397769122324305e-05,
      "loss": 0.6219,
      "step": 730
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2920,
  "num_train_epochs": 20,
  "save_steps": 1000000000,
  "total_flos": 9.79415936794624e+17,
  "trial_name": null,
  "trial_params": null
}