{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9990250243743906,
  "eval_steps": 256,
  "global_step": 2307,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8447,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4e-05,
      "loss": 2.1865,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 6e-05,
      "loss": 2.3418,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 8e-05,
      "loss": 2.2393,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001,
      "loss": 1.9549,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00012,
      "loss": 2.2225,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00014,
      "loss": 2.0327,
      "step": 7
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00016,
      "loss": 2.0837,
      "step": 8
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018,
      "loss": 1.9674,
      "step": 9
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0002,
      "loss": 1.6951,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001999999941924048,
      "loss": 1.6578,
      "step": 11
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999997676961985,
      "loss": 1.576,
      "step": 12
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001999999477316472,
      "loss": 1.6541,
      "step": 13
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999990707849022,
      "loss": 1.3535,
      "step": 14
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999985481015364,
      "loss": 1.2641,
      "step": 15
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999979092664348,
      "loss": 1.2216,
      "step": 16
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999971542796721,
      "loss": 1.421,
      "step": 17
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001999996283141336,
      "loss": 1.1499,
      "step": 18
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019999952958515274,
      "loss": 0.9146,
      "step": 19
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019999941924103611,
      "loss": 1.1572,
      "step": 20
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001999992972817965,
      "loss": 1.1196,
      "step": 21
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019999916370744814,
      "loss": 1.4986,
      "step": 22
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001999990185180065,
      "loss": 0.9607,
      "step": 23
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019999886171348847,
      "loss": 1.0656,
      "step": 24
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019999869329391222,
      "loss": 1.0145,
      "step": 25
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019999851325929735,
      "loss": 0.9555,
      "step": 26
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019999832160966477,
      "loss": 1.2241,
      "step": 27
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001999981183450367,
      "loss": 1.0528,
      "step": 28
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001999979034654368,
      "loss": 0.9701,
      "step": 29
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019999767697089,
      "loss": 0.8233,
      "step": 30
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019999743886142262,
      "loss": 1.0239,
      "step": 31
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001999971891370623,
      "loss": 1.0671,
      "step": 32
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019999692779783807,
      "loss": 0.7999,
      "step": 33
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001999966548437803,
      "loss": 0.8779,
      "step": 34
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019999637027492062,
      "loss": 0.8969,
      "step": 35
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019999607409129216,
      "loss": 0.8077,
      "step": 36
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019999576629292927,
      "loss": 0.9768,
      "step": 37
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019999544687986776,
      "loss": 0.8677,
      "step": 38
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019999511585214465,
      "loss": 0.9755,
      "step": 39
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019999477320979846,
      "loss": 0.9795,
      "step": 40
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.000199994418952869,
      "loss": 0.9113,
      "step": 41
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019999405308139735,
      "loss": 0.7379,
      "step": 42
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019999367559542602,
      "loss": 0.7477,
      "step": 43
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019999328649499887,
      "loss": 0.7874,
      "step": 44
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001999928857801611,
      "loss": 0.8461,
      "step": 45
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019999247345095928,
      "loss": 0.8191,
      "step": 46
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019999204950744127,
      "loss": 0.7752,
      "step": 47
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001999916139496563,
      "loss": 0.6585,
      "step": 48
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019999116677765498,
      "loss": 0.6603,
      "step": 49
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019999070799148923,
      "loss": 0.9971,
      "step": 50
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019999023759121237,
      "loss": 0.8029,
      "step": 51
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019998975557687906,
      "loss": 0.6454,
      "step": 52
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001999892619485452,
      "loss": 0.6136,
      "step": 53
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019998875670626819,
      "loss": 1.047,
      "step": 54
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019998823985010672,
      "loss": 0.5221,
      "step": 55
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001999877113801208,
      "loss": 0.9518,
      "step": 56
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001999871712963718,
      "loss": 0.5893,
      "step": 57
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0001999866195989225,
      "loss": 0.6162,
      "step": 58
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019998605628783692,
      "loss": 0.9023,
      "step": 59
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019998548136318056,
      "loss": 0.7126,
      "step": 60
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019998489482502013,
      "loss": 0.8642,
      "step": 61
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019998429667342378,
      "loss": 0.665,
      "step": 62
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019998368690846098,
      "loss": 0.6577,
      "step": 63
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019998306553020262,
      "loss": 0.7237,
      "step": 64
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.00019998243253872077,
      "loss": 0.7588,
      "step": 65
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019998178793408905,
      "loss": 0.7006,
      "step": 66
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019998113171638228,
      "loss": 0.7269,
      "step": 67
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019998046388567664,
      "loss": 0.7561,
      "step": 68
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001999797844420498,
      "loss": 0.6634,
      "step": 69
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001999790933855806,
      "loss": 0.705,
      "step": 70
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019997839071634935,
      "loss": 0.4444,
      "step": 71
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019997767643443762,
      "loss": 0.9072,
      "step": 72
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.00019997695053992844,
      "loss": 0.6258,
      "step": 73
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0001999762130329061,
      "loss": 0.644,
      "step": 74
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019997546391345623,
      "loss": 0.6299,
      "step": 75
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019997470318166587,
      "loss": 0.4758,
      "step": 76
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019997393083762336,
      "loss": 0.6336,
      "step": 77
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019997314688141845,
      "loss": 0.4333,
      "step": 78
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019997235131314215,
      "loss": 0.5814,
      "step": 79
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019997154413288693,
      "loss": 0.6756,
      "step": 80
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019997072534074645,
      "loss": 0.7424,
      "step": 81
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019996989493681588,
      "loss": 0.6102,
      "step": 82
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001999690529211917,
      "loss": 0.7806,
      "step": 83
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019996819929397165,
      "loss": 0.6747,
      "step": 84
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0001999673340552549,
      "loss": 0.8397,
      "step": 85
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019996645720514195,
      "loss": 0.5725,
      "step": 86
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019996556874373467,
      "loss": 0.6423,
      "step": 87
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019996466867113624,
      "loss": 0.5559,
      "step": 88
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019996375698745115,
      "loss": 0.6251,
      "step": 89
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019996283369278541,
      "loss": 0.5611,
      "step": 90
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019996189878724617,
      "loss": 0.5144,
      "step": 91
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019996095227094204,
      "loss": 0.7947,
      "step": 92
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.000199959994143983,
      "loss": 0.8718,
      "step": 93
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0001999590244064803,
      "loss": 0.5949,
      "step": 94
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019995804305854655,
      "loss": 0.7059,
      "step": 95
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019995705010029583,
      "loss": 0.7815,
      "step": 96
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019995604553184335,
      "loss": 0.7278,
      "step": 97
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001999550293533059,
      "loss": 0.4676,
      "step": 98
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019995400156480147,
      "loss": 0.6948,
      "step": 99
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019995296216644942,
      "loss": 0.4889,
      "step": 100
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019995191115837048,
      "loss": 0.7367,
      "step": 101
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019995084854068678,
      "loss": 0.6689,
      "step": 102
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019994977431352166,
      "loss": 0.5654,
      "step": 103
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019994868847699997,
      "loss": 0.661,
      "step": 104
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001999475910312478,
      "loss": 0.7194,
      "step": 105
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019994648197639263,
      "loss": 0.6786,
      "step": 106
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019994536131256324,
      "loss": 0.5634,
      "step": 107
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001999442290398899,
      "loss": 0.6238,
      "step": 108
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019994308515850398,
      "loss": 0.7237,
      "step": 109
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019994192966853844,
      "loss": 0.569,
      "step": 110
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001999407625701275,
      "loss": 0.5945,
      "step": 111
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019993958386340665,
      "loss": 0.5952,
      "step": 112
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019993839354851288,
      "loss": 0.3274,
      "step": 113
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019993719162558437,
      "loss": 0.6255,
      "step": 114
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001999359780947608,
      "loss": 0.7292,
      "step": 115
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019993475295618308,
      "loss": 0.8216,
      "step": 116
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019993351620999352,
      "loss": 0.5401,
      "step": 117
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019993226785633574,
      "loss": 0.6242,
      "step": 118
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0001999310078953548,
      "loss": 0.7058,
      "step": 119
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.000199929736327197,
      "loss": 0.5044,
      "step": 120
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019992845315201007,
      "loss": 0.5582,
      "step": 121
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.000199927158369943,
      "loss": 0.489,
      "step": 122
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019992585198114627,
      "loss": 0.5382,
      "step": 123
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001999245339857715,
      "loss": 0.4422,
      "step": 124
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019992320438397187,
      "loss": 0.6081,
      "step": 125
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001999218631759018,
      "loss": 0.5615,
      "step": 126
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019992051036171708,
      "loss": 0.5493,
      "step": 127
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001999191459415748,
      "loss": 0.4734,
      "step": 128
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019991776991563348,
      "loss": 0.5567,
      "step": 129
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001999163822840529,
      "loss": 0.3175,
      "step": 130
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019991498304699427,
      "loss": 0.6372,
      "step": 131
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019991357220462013,
      "loss": 0.586,
      "step": 132
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019991214975709435,
      "loss": 0.5545,
      "step": 133
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001999107157045821,
      "loss": 0.4699,
      "step": 134
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019990927004725,
      "loss": 0.5774,
      "step": 135
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019990781278526597,
      "loss": 0.3848,
      "step": 136
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019990634391879922,
      "loss": 0.5031,
      "step": 137
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019990486344802043,
      "loss": 0.7628,
      "step": 138
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019990337137310146,
      "loss": 0.7127,
      "step": 139
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019990186769421574,
      "loss": 0.5708,
      "step": 140
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019990035241153784,
      "loss": 0.598,
      "step": 141
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019989882552524376,
      "loss": 0.7069,
      "step": 142
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019989728703551093,
      "loss": 0.7696,
      "step": 143
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019989573694251798,
      "loss": 0.8118,
      "step": 144
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019989417524644498,
      "loss": 0.4401,
      "step": 145
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001998926019474733,
      "loss": 0.5229,
      "step": 146
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001998910170457857,
      "loss": 0.4623,
      "step": 147
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019988942054156633,
      "loss": 0.7658,
      "step": 148
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0001998878124350005,
      "loss": 0.4783,
      "step": 149
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019988619272627507,
      "loss": 0.5119,
      "step": 150
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019988456141557817,
      "loss": 0.5611,
      "step": 151
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019988291850309928,
      "loss": 0.7363,
      "step": 152
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001998812639890292,
      "loss": 0.5845,
      "step": 153
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019987959787356015,
      "loss": 0.6208,
      "step": 154
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019987792015688563,
      "loss": 0.4857,
      "step": 155
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001998762308392005,
      "loss": 0.3632,
      "step": 156
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019987452992070097,
      "loss": 0.5116,
      "step": 157
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019987281740158462,
      "loss": 0.6581,
      "step": 158
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001998710932820504,
      "loss": 0.417,
      "step": 159
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001998693575622985,
      "loss": 0.4106,
      "step": 160
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019986761024253056,
      "loss": 0.4868,
      "step": 161
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019986585132294953,
      "loss": 0.5194,
      "step": 162
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019986408080375974,
      "loss": 0.6545,
      "step": 163
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001998622986851668,
      "loss": 0.6197,
      "step": 164
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00019986050496737773,
      "loss": 0.6387,
      "step": 165
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019985869965060085,
      "loss": 0.7138,
      "step": 166
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019985688273504586,
      "loss": 0.488,
      "step": 167
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019985505422092384,
      "loss": 0.4416,
      "step": 168
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001998532141084471,
      "loss": 0.5069,
      "step": 169
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019985136239782943,
      "loss": 0.5029,
      "step": 170
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001998494990892859,
      "loss": 0.5658,
      "step": 171
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001998476241830329,
      "loss": 0.5266,
      "step": 172
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00019984573767928824,
      "loss": 0.4776,
      "step": 173
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019984383957827106,
      "loss": 0.5413,
      "step": 174
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001998419298802018,
      "loss": 0.4012,
      "step": 175
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019984000858530227,
      "loss": 0.6573,
      "step": 176
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001998380756937956,
      "loss": 0.5688,
      "step": 177
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001998361312059064,
      "loss": 0.6729,
      "step": 178
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019983417512186042,
      "loss": 0.4086,
      "step": 179
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00019983220744188493,
      "loss": 0.5498,
      "step": 180
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00019983022816620846,
      "loss": 0.4674,
      "step": 181
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001998282372950609,
      "loss": 0.5661,
      "step": 182
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001998262348286735,
      "loss": 0.3976,
      "step": 183
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00019982422076727884,
      "loss": 0.3531,
      "step": 184
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00019982219511111083,
      "loss": 0.5082,
      "step": 185
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001998201578604048,
      "loss": 0.5398,
      "step": 186
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00019981810901539745,
      "loss": 0.4306,
      "step": 187
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001998160485763266,
      "loss": 0.4772,
      "step": 188
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019981397654343167,
      "loss": 0.5306,
      "step": 189
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019981189291695328,
      "loss": 0.4842,
      "step": 190
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0001998097976971335,
      "loss": 0.6483,
      "step": 191
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0001998076908842157,
      "loss": 0.5152,
      "step": 192
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0001998055724784445,
      "loss": 0.5388,
      "step": 193
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019980344248006607,
      "loss": 0.7442,
      "step": 194
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019980130088932774,
      "loss": 0.6341,
      "step": 195
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00019979914770647828,
      "loss": 0.4562,
      "step": 196
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001997969829317678,
      "loss": 0.395,
      "step": 197
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019979480656544772,
      "loss": 0.6145,
      "step": 198
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019979261860777084,
      "loss": 0.754,
      "step": 199
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001997904190589913,
      "loss": 0.6445,
      "step": 200
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001997882079193646,
      "loss": 0.4616,
      "step": 201
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001997859851891475,
      "loss": 0.5271,
      "step": 202
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00019978375086859824,
      "loss": 0.5929,
      "step": 203
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019978150495797632,
      "loss": 0.4455,
      "step": 204
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019977924745754263,
      "loss": 0.5503,
      "step": 205
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019977697836755934,
      "loss": 0.5417,
      "step": 206
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019977469768829,
      "loss": 0.4149,
      "step": 207
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019977240541999957,
      "loss": 0.5015,
      "step": 208
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019977010156295427,
      "loss": 0.6789,
      "step": 209
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019976778611742173,
      "loss": 0.4961,
      "step": 210
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00019976545908367084,
      "loss": 0.4563,
      "step": 211
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019976312046197193,
      "loss": 0.5287,
      "step": 212
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001997607702525966,
      "loss": 0.5581,
      "step": 213
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019975840845581785,
      "loss": 0.5591,
      "step": 214
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019975603507191002,
      "loss": 0.7441,
      "step": 215
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019975365010114877,
      "loss": 0.5907,
      "step": 216
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019975125354381113,
      "loss": 0.4914,
      "step": 217
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019974884540017545,
      "loss": 0.539,
      "step": 218
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00019974642567052145,
      "loss": 0.7504,
      "step": 219
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001997439943551302,
      "loss": 0.6565,
      "step": 220
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019974155145428406,
      "loss": 0.4329,
      "step": 221
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001997390969682668,
      "loss": 0.683,
      "step": 222
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0001997366308973635,
      "loss": 0.5974,
      "step": 223
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019973415324186066,
      "loss": 0.408,
      "step": 224
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.000199731664002046,
      "loss": 0.4712,
      "step": 225
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019972916317820866,
      "loss": 0.5457,
      "step": 226
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019972665077063915,
      "loss": 0.5137,
      "step": 227
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019972412677962922,
      "loss": 0.5647,
      "step": 228
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019972159120547212,
      "loss": 0.5461,
      "step": 229
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001997190440484623,
      "loss": 0.4815,
      "step": 230
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019971648530889567,
      "loss": 0.467,
      "step": 231
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019971391498706936,
      "loss": 0.7947,
      "step": 232
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.000199711333083282,
      "loss": 0.4465,
      "step": 233
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019970873959783344,
      "loss": 0.3713,
      "step": 234
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001997061345310249,
      "loss": 0.4715,
      "step": 235
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.000199703517883159,
      "loss": 0.7857,
      "step": 236
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019970088965453963,
      "loss": 0.6404,
      "step": 237
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001996982498454721,
      "loss": 0.3878,
      "step": 238
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.000199695598456263,
      "loss": 0.5648,
      "step": 239
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019969293548722033,
      "loss": 0.5672,
      "step": 240
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019969026093865342,
      "loss": 0.4332,
      "step": 241
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019968757481087282,
      "loss": 0.4925,
      "step": 242
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019968487710419065,
      "loss": 0.5979,
      "step": 243
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019968216781892015,
      "loss": 0.5599,
      "step": 244
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019967944695537605,
      "loss": 0.407,
      "step": 245
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001996767145138744,
      "loss": 0.3254,
      "step": 246
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019967397049473258,
      "loss": 0.4003,
      "step": 247
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001996712148982693,
      "loss": 0.5426,
      "step": 248
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019966844772480462,
      "loss": 0.506,
      "step": 249
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019966566897465995,
      "loss": 0.5965,
      "step": 250
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019966287864815806,
      "loss": 0.5057,
      "step": 251
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019966007674562306,
      "loss": 0.5373,
      "step": 252
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019965726326738037,
      "loss": 0.6248,
      "step": 253
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001996544382137568,
      "loss": 0.4623,
      "step": 254
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001996516015850805,
      "loss": 0.5483,
      "step": 255
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019964875338168093,
      "loss": 0.7984,
      "step": 256
    },
    {
      "epoch": 0.33,
      "eval_loss": 0.5021235942840576,
      "eval_runtime": 126.1365,
      "eval_samples_per_second": 1.015,
      "eval_steps_per_second": 0.341,
      "step": 256
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001996458936038889,
      "loss": 0.4895,
      "step": 257
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019964302225203658,
      "loss": 0.4388,
      "step": 258
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019964013932645752,
      "loss": 0.3486,
      "step": 259
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019963724482748653,
      "loss": 0.4693,
      "step": 260
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019963433875545985,
      "loss": 0.4754,
      "step": 261
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.000199631421110715,
      "loss": 0.5187,
      "step": 262
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001996284918935909,
      "loss": 0.671,
      "step": 263
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019962555110442779,
      "loss": 0.4173,
      "step": 264
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019962259874356717,
      "loss": 0.6412,
      "step": 265
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019961963481135203,
      "loss": 0.379,
      "step": 266
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019961665930812666,
      "loss": 0.4203,
      "step": 267
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019961367223423658,
      "loss": 0.6542,
      "step": 268
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001996106735900288,
      "loss": 0.5057,
      "step": 269
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019960766337585165,
      "loss": 0.4153,
      "step": 270
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019960464159205472,
      "loss": 0.4218,
      "step": 271
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019960160823898903,
      "loss": 0.433,
      "step": 272
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001995985633170069,
      "loss": 0.5434,
      "step": 273
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019959550682646196,
      "loss": 0.5434,
      "step": 274
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019959243876770927,
      "loss": 0.3629,
      "step": 275
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001995893591411052,
      "loss": 0.2815,
      "step": 276
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019958626794700743,
      "loss": 0.6312,
      "step": 277
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019958316518577506,
      "loss": 0.4063,
      "step": 278
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019958005085776836,
      "loss": 0.4481,
      "step": 279
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001995769249633492,
      "loss": 0.5287,
      "step": 280
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001995737875028806,
      "loss": 0.7435,
      "step": 281
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019957063847672693,
      "loss": 0.4257,
      "step": 282
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001995674778852541,
      "loss": 0.5351,
      "step": 283
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019956430572882906,
      "loss": 0.539,
      "step": 284
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019956112200782036,
      "loss": 0.4973,
      "step": 285
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019955792672259778,
      "loss": 0.4631,
      "step": 286
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001995547198735324,
      "loss": 0.5286,
      "step": 287
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019955150146099676,
      "loss": 0.4874,
      "step": 288
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019954827148536467,
      "loss": 0.4117,
      "step": 289
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019954502994701132,
      "loss": 0.4981,
      "step": 290
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019954177684631318,
      "loss": 0.4683,
      "step": 291
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019953851218364814,
      "loss": 0.4435,
      "step": 292
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001995352359593954,
      "loss": 0.9429,
      "step": 293
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019953194817393541,
      "loss": 0.5293,
      "step": 294
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019952864882765017,
      "loss": 0.5297,
      "step": 295
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019952533792092286,
      "loss": 0.4196,
      "step": 296
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019952201545413804,
      "loss": 0.4448,
      "step": 297
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019951868142768163,
      "loss": 0.4726,
      "step": 298
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019951533584194086,
      "loss": 0.4884,
      "step": 299
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019951197869730436,
      "loss": 0.3992,
      "step": 300
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019950860999416205,
      "loss": 0.2992,
      "step": 301
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019950522973290522,
      "loss": 0.3934,
      "step": 302
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001995018379139265,
      "loss": 0.568,
      "step": 303
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019949843453761983,
      "loss": 0.3944,
      "step": 304
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019949501960438054,
      "loss": 0.3959,
      "step": 305
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019949159311460527,
      "loss": 0.5346,
      "step": 306
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019948815506869204,
      "loss": 0.4367,
      "step": 307
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019948470546704016,
      "loss": 0.5085,
      "step": 308
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001994812443100503,
      "loss": 0.3277,
      "step": 309
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019947777159812452,
      "loss": 0.5946,
      "step": 310
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.0001994742873316661,
      "loss": 0.57,
      "step": 311
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019947079151107984,
      "loss": 0.5171,
      "step": 312
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019946728413677175,
      "loss": 0.3905,
      "step": 313
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001994637652091492,
      "loss": 0.4382,
      "step": 314
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019946023472862092,
      "loss": 0.4955,
      "step": 315
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019945669269559702,
      "loss": 0.4098,
      "step": 316
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019945313911048888,
      "loss": 0.5335,
      "step": 317
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019944957397370928,
      "loss": 0.4289,
      "step": 318
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019944599728567228,
      "loss": 0.4759,
      "step": 319
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019944240904679335,
      "loss": 0.4845,
      "step": 320
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019943880925748925,
      "loss": 0.5239,
      "step": 321
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019943519791817813,
      "loss": 0.5245,
      "step": 322
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019943157502927945,
      "loss": 0.5723,
      "step": 323
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019942794059121398,
      "loss": 0.413,
      "step": 324
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001994242946044039,
      "loss": 0.4031,
      "step": 325
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001994206370692727,
      "loss": 0.3496,
      "step": 326
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019941696798624517,
      "loss": 0.4433,
      "step": 327
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019941328735574752,
      "loss": 0.4978,
      "step": 328
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019940959517820728,
      "loss": 0.4069,
      "step": 329
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019940589145405325,
      "loss": 0.2685,
      "step": 330
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019940217618371567,
      "loss": 0.5005,
      "step": 331
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019939844936762603,
      "loss": 0.4049,
      "step": 332
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019939471100621726,
      "loss": 0.3551,
      "step": 333
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019939096109992352,
      "loss": 0.5653,
      "step": 334
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019938719964918041,
      "loss": 0.5415,
      "step": 335
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019938342665442484,
      "loss": 0.4397,
      "step": 336
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.000199379642116095,
      "loss": 0.4598,
      "step": 337
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019937584603463054,
      "loss": 0.4695,
      "step": 338
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001993720384104723,
      "loss": 0.4652,
      "step": 339
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001993682192440626,
      "loss": 0.4917,
      "step": 340
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019936438853584504,
      "loss": 0.5436,
      "step": 341
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019936054628626456,
      "loss": 0.4429,
      "step": 342
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019935669249576741,
      "loss": 0.6908,
      "step": 343
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019935282716480125,
      "loss": 0.6403,
      "step": 344
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019934895029381506,
      "loss": 0.3833,
      "step": 345
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019934506188325912,
      "loss": 0.529,
      "step": 346
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019934116193358507,
      "loss": 0.3884,
      "step": 347
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019933725044524593,
      "loss": 0.3232,
      "step": 348
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00019933332741869597,
      "loss": 0.4554,
      "step": 349
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001993293928543909,
      "loss": 0.3563,
      "step": 350
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019932544675278774,
      "loss": 0.4566,
      "step": 351
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001993214891143448,
      "loss": 0.5977,
      "step": 352
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019931751993952176,
      "loss": 0.3993,
      "step": 353
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019931353922877972,
      "loss": 0.4005,
      "step": 354
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019930954698258097,
      "loss": 0.4295,
      "step": 355
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019930554320138924,
      "loss": 0.4768,
      "step": 356
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001993015278856696,
      "loss": 0.6435,
      "step": 357
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0001992975010358884,
      "loss": 0.3118,
      "step": 358
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019929346265251335,
      "loss": 0.2813,
      "step": 359
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019928941273601362,
      "loss": 0.6655,
      "step": 360
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0001992853512868595,
      "loss": 0.3739,
      "step": 361
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019928127830552282,
      "loss": 0.4101,
      "step": 362
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019927719379247657,
      "loss": 0.4865,
      "step": 363
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019927309774819528,
      "loss": 0.3554,
      "step": 364
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019926899017315463,
      "loss": 0.3452,
      "step": 365
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019926487106783176,
      "loss": 0.6231,
      "step": 366
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001992607404327051,
      "loss": 0.5088,
      "step": 367
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019925659826825447,
      "loss": 0.4252,
      "step": 368
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019925244457496093,
      "loss": 0.4643,
      "step": 369
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019924827935330697,
      "loss": 0.5473,
      "step": 370
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001992441026037764,
      "loss": 0.5066,
      "step": 371
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.0001992399143268543,
      "loss": 0.3966,
      "step": 372
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019923571452302725,
      "loss": 0.3844,
      "step": 373
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019923150319278298,
      "loss": 0.5569,
      "step": 374
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019922728033661065,
      "loss": 0.4612,
      "step": 375
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019922304595500082,
      "loss": 0.5429,
      "step": 376
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019921880004844522,
      "loss": 0.4139,
      "step": 377
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0001992145426174371,
      "loss": 0.594,
      "step": 378
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019921027366247092,
      "loss": 0.5506,
      "step": 379
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019920599318404256,
      "loss": 0.3213,
      "step": 380
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0001992017011826492,
      "loss": 0.5904,
      "step": 381
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019919739765878939,
      "loss": 0.5192,
      "step": 382
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019919308261296293,
      "loss": 0.3534,
      "step": 383
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019918875604567105,
      "loss": 0.4387,
      "step": 384
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0001991844179574163,
      "loss": 0.6316,
      "step": 385
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019918006834870254,
      "loss": 0.4937,
      "step": 386
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.000199175707220035,
      "loss": 0.5105,
      "step": 387
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019917133457192024,
      "loss": 0.4127,
      "step": 388
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0001991669504048661,
      "loss": 0.6069,
      "step": 389
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019916255471938188,
      "loss": 0.3679,
      "step": 390
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019915814751597812,
      "loss": 0.401,
      "step": 391
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019915372879516672,
      "loss": 0.3537,
      "step": 392
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019914929855746093,
      "loss": 0.4484,
      "step": 393
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019914485680337527,
      "loss": 0.4842,
      "step": 394
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019914040353342574,
      "loss": 0.3667,
      "step": 395
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0001991359387481296,
      "loss": 0.3484,
      "step": 396
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019913146244800541,
      "loss": 0.323,
      "step": 397
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019912697463357306,
      "loss": 0.4554,
      "step": 398
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019912247530535387,
      "loss": 0.5663,
      "step": 399
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019911796446387046,
      "loss": 0.5668,
      "step": 400
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019911344210964672,
      "loss": 0.5918,
      "step": 401
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019910890824320797,
      "loss": 0.2257,
      "step": 402
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019910436286508083,
      "loss": 0.5382,
      "step": 403
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019909980597579324,
      "loss": 0.4374,
      "step": 404
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019909523757587446,
      "loss": 0.3998,
      "step": 405
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0001990906576658552,
      "loss": 0.4526,
      "step": 406
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019908606624626733,
      "loss": 0.5144,
      "step": 407
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019908146331764422,
      "loss": 0.541,
      "step": 408
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019907684888052047,
      "loss": 0.6822,
      "step": 409
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019907222293543209,
      "loss": 0.4046,
      "step": 410
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019906758548291636,
      "loss": 0.4929,
      "step": 411
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019906293652351197,
      "loss": 0.3962,
      "step": 412
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019905827605775885,
      "loss": 0.432,
      "step": 413
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019905360408619833,
      "loss": 0.4594,
      "step": 414
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0001990489206093731,
      "loss": 0.3956,
      "step": 415
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019904422562782719,
      "loss": 0.639,
      "step": 416
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019903951914210585,
      "loss": 0.3293,
      "step": 417
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019903480115275582,
      "loss": 0.5015,
      "step": 418
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019903007166032502,
      "loss": 0.4892,
      "step": 419
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00019902533066536285,
      "loss": 0.4328,
      "step": 420
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00019902057816841996,
      "loss": 0.3849,
      "step": 421
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001990158141700484,
      "loss": 0.3825,
      "step": 422
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00019901103867080147,
      "loss": 0.6273,
      "step": 423
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00019900625167123385,
      "loss": 0.4414,
      "step": 424
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001990014531719016,
      "loss": 0.4716,
      "step": 425
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001989966431733621,
      "loss": 0.4584,
      "step": 426
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0001989918216761739,
      "loss": 0.5656,
      "step": 427
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00019898698868089722,
      "loss": 0.5669,
      "step": 428
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0001989821441880933,
      "loss": 0.4384,
      "step": 429
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00019897728819832483,
      "loss": 0.352,
      "step": 430
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.0001989724207121559,
      "loss": 0.3306,
      "step": 431
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00019896754173015185,
      "loss": 0.4437,
      "step": 432
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00019896265125287938,
      "loss": 0.5659,
      "step": 433
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00019895774928090654,
      "loss": 0.3009,
      "step": 434
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001989528358148027,
      "loss": 0.4718,
      "step": 435
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00019894791085513857,
      "loss": 0.6008,
      "step": 436
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00019894297440248618,
      "loss": 0.508,
      "step": 437
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001989380264574189,
      "loss": 0.4314,
      "step": 438
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001989330670205115,
      "loss": 0.3694,
      "step": 439
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00019892809609233997,
      "loss": 0.3637,
      "step": 440
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00019892311367348167,
      "loss": 0.472,
      "step": 441
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.0001989181197645154,
      "loss": 0.5284,
      "step": 442
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019891311436602118,
      "loss": 0.3974,
      "step": 443
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019890809747858036,
      "loss": 0.408,
      "step": 444
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0001989030691027757,
      "loss": 0.5062,
      "step": 445
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019889802923919126,
      "loss": 0.4252,
      "step": 446
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019889297788841238,
      "loss": 0.4821,
      "step": 447
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019888791505102583,
      "loss": 0.4383,
      "step": 448
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019888284072761968,
      "loss": 0.5106,
      "step": 449
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019887775491878327,
      "loss": 0.3167,
      "step": 450
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0001988726576251074,
      "loss": 0.4538,
      "step": 451
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00019886754884718404,
      "loss": 0.5705,
      "step": 452
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00019886242858560663,
      "loss": 0.4679,
      "step": 453
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00019885729684096992,
      "loss": 0.4496,
      "step": 454
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00019885215361386993,
      "loss": 0.4231,
      "step": 455
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00019884699890490408,
      "loss": 0.5019,
      "step": 456
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0001988418327146711,
      "loss": 0.4275,
      "step": 457
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00019883665504377105,
      "loss": 0.5317,
      "step": 458
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001988314658928053,
      "loss": 0.422,
      "step": 459
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00019882626526237664,
      "loss": 0.4602,
      "step": 460
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00019882105315308907,
      "loss": 0.3006,
      "step": 461
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.000198815829565548,
      "loss": 0.5431,
      "step": 462
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001988105945003602,
      "loss": 0.3506,
      "step": 463
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001988053479581337,
      "loss": 0.4162,
      "step": 464
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00019880008993947788,
      "loss": 0.4948,
      "step": 465
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00019879482044500352,
      "loss": 0.4552,
      "step": 466
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00019878953947532265,
      "loss": 0.4246,
      "step": 467
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00019878424703104867,
      "loss": 0.3214,
      "step": 468
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00019877894311279626,
      "loss": 0.5128,
      "step": 469
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00019877362772118158,
      "loss": 0.4842,
      "step": 470
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00019876830085682192,
      "loss": 0.2497,
      "step": 471
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0001987629625203361,
      "loss": 0.4015,
      "step": 472
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00019875761271234409,
      "loss": 0.4122,
      "step": 473
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00019875225143346732,
      "loss": 0.5834,
      "step": 474
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0001987468786843285,
      "loss": 0.6073,
      "step": 475
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00019874149446555176,
      "loss": 0.4439,
      "step": 476
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0001987360987777624,
      "loss": 0.3859,
      "step": 477
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00019873069162158714,
      "loss": 0.5788,
      "step": 478
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00019872527299765407,
      "loss": 0.437,
      "step": 479
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00019871984290659257,
      "loss": 0.4549,
      "step": 480
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0001987144013490333,
      "loss": 0.3508,
      "step": 481
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0001987089483256084,
      "loss": 0.4045,
      "step": 482
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00019870348383695121,
      "loss": 0.304,
      "step": 483
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0001986980078836964,
      "loss": 0.3786,
      "step": 484
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00019869252046648007,
      "loss": 0.433,
      "step": 485
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00019868702158593954,
      "loss": 0.423,
      "step": 486
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00019868151124271356,
      "loss": 0.3812,
      "step": 487
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00019867598943744218,
      "loss": 0.3025,
      "step": 488
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0001986704561707667,
      "loss": 0.3424,
      "step": 489
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0001986649114433299,
      "loss": 0.4141,
      "step": 490
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0001986593552557757,
      "loss": 0.4925,
      "step": 491
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0001986537876087496,
      "loss": 0.4464,
      "step": 492
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0001986482085028982,
      "loss": 0.4524,
      "step": 493
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00019864261793886955,
      "loss": 0.4066,
      "step": 494
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00019863701591731298,
      "loss": 0.4235,
      "step": 495
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.0001986314024388792,
      "loss": 0.3453,
      "step": 496
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00019862577750422026,
      "loss": 0.4502,
      "step": 497
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00019862014111398946,
      "loss": 0.4076,
      "step": 498
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00019861449326884147,
      "loss": 0.4082,
      "step": 499
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00019860883396943232,
      "loss": 0.3837,
      "step": 500
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00019860316321641935,
      "loss": 0.3554,
      "step": 501
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00019859748101046123,
      "loss": 0.4545,
      "step": 502
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00019859178735221795,
      "loss": 0.5174,
      "step": 503
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00019858608224235081,
      "loss": 0.343,
      "step": 504
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00019858036568152256,
      "loss": 0.5002,
      "step": 505
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001985746376703971,
      "loss": 0.4377,
      "step": 506
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00019856889820963976,
      "loss": 0.4197,
      "step": 507
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00019856314729991726,
      "loss": 0.4253,
      "step": 508
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00019855738494189748,
      "loss": 0.6028,
      "step": 509
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001985516111362498,
      "loss": 0.5472,
      "step": 510
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00019854582588364484,
      "loss": 0.4237,
      "step": 511
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019854002918475455,
      "loss": 0.3112,
      "step": 512
    },
    {
      "epoch": 0.67,
      "eval_loss": 0.43665099143981934,
      "eval_runtime": 126.0856,
      "eval_samples_per_second": 1.015,
      "eval_steps_per_second": 0.341,
      "step": 512
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019853422104025226,
      "loss": 0.4999,
      "step": 513
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019852840145081254,
      "loss": 0.471,
      "step": 514
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0001985225704171114,
      "loss": 0.4006,
      "step": 515
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019851672793982616,
      "loss": 0.4797,
      "step": 516
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019851087401963536,
      "loss": 0.4435,
      "step": 517
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019850500865721893,
      "loss": 0.4752,
      "step": 518
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019849913185325822,
      "loss": 0.248,
      "step": 519
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00019849324360843578,
      "loss": 0.397,
      "step": 520
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00019848734392343556,
      "loss": 0.4828,
      "step": 521
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00019848143279894282,
      "loss": 0.6112,
      "step": 522
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00019847551023564414,
      "loss": 0.5375,
      "step": 523
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00019846957623422743,
      "loss": 0.5776,
      "step": 524
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00019846363079538198,
      "loss": 0.386,
      "step": 525
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0001984576739197983,
      "loss": 0.5265,
      "step": 526
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019845170560816832,
      "loss": 0.4189,
      "step": 527
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0001984457258611853,
      "loss": 0.5289,
      "step": 528
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019843973467954373,
      "loss": 0.3956,
      "step": 529
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019843373206393955,
      "loss": 0.4328,
      "step": 530
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019842771801507,
      "loss": 0.4069,
      "step": 531
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019842169253363355,
      "loss": 0.6731,
      "step": 532
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019841565562033012,
      "loss": 0.4303,
      "step": 533
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019840960727586092,
      "loss": 0.4138,
      "step": 534
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00019840354750092846,
      "loss": 0.4702,
      "step": 535
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00019839747629623653,
      "loss": 0.6149,
      "step": 536
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00019839139366249043,
      "loss": 0.4623,
      "step": 537
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00019838529960039658,
      "loss": 0.3893,
      "step": 538
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00019837919411066287,
      "loss": 0.4094,
      "step": 539
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0001983730771939984,
      "loss": 0.5875,
      "step": 540
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00019836694885111373,
      "loss": 0.3896,
      "step": 541
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00019836080908272065,
      "loss": 0.4023,
      "step": 542
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00019835465788953233,
      "loss": 0.4272,
      "step": 543
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0001983484952722632,
      "loss": 0.286,
      "step": 544
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00019834232123162909,
      "loss": 0.3807,
      "step": 545
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00019833613576834708,
      "loss": 0.3156,
      "step": 546
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00019832993888313572,
      "loss": 0.6188,
      "step": 547
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00019832373057671466,
      "loss": 0.3001,
      "step": 548
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00019831751084980513,
      "loss": 0.4459,
      "step": 549
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.0001983112797031295,
      "loss": 0.4228,
      "step": 550
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00019830503713741152,
      "loss": 0.4117,
      "step": 551
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00019829878315337634,
      "loss": 0.2905,
      "step": 552
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00019829251775175028,
      "loss": 0.3158,
      "step": 553
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00019828624093326113,
      "loss": 0.4535,
      "step": 554
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.000198279952698638,
      "loss": 0.3953,
      "step": 555
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00019827365304861115,
      "loss": 0.3113,
      "step": 556
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00019826734198391245,
      "loss": 0.5263,
      "step": 557
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00019826101950527484,
      "loss": 0.324,
      "step": 558
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00019825468561343273,
      "loss": 0.4191,
      "step": 559
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00019824834030912177,
      "loss": 0.5083,
      "step": 560
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00019824198359307902,
      "loss": 0.4819,
      "step": 561
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00019823561546604282,
      "loss": 0.3715,
      "step": 562
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00019822923592875284,
      "loss": 0.5131,
      "step": 563
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001982228449819501,
      "loss": 0.3625,
      "step": 564
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00019821644262637685,
      "loss": 0.4485,
      "step": 565
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019821002886277677,
      "loss": 0.4093,
      "step": 566
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019820360369189485,
      "loss": 0.3596,
      "step": 567
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019819716711447737,
      "loss": 0.3697,
      "step": 568
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019819071913127197,
      "loss": 0.397,
      "step": 569
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019818425974302758,
      "loss": 0.5212,
      "step": 570
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019817778895049446,
      "loss": 0.4664,
      "step": 571
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019817130675442421,
      "loss": 0.2758,
      "step": 572
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00019816481315556976,
      "loss": 0.4397,
      "step": 573
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00019815830815468534,
      "loss": 0.4317,
      "step": 574
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00019815179175252653,
      "loss": 0.4341,
      "step": 575
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00019814526394985023,
      "loss": 0.2899,
      "step": 576
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00019813872474741464,
      "loss": 0.3975,
      "step": 577
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001981321741459793,
      "loss": 0.4867,
      "step": 578
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00019812561214630509,
      "loss": 0.3514,
      "step": 579
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00019811903874915419,
      "loss": 0.4726,
      "step": 580
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019811245395529007,
      "loss": 0.3546,
      "step": 581
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019810585776547767,
      "loss": 0.4236,
      "step": 582
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019809925018048305,
      "loss": 0.4237,
      "step": 583
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019809263120107375,
      "loss": 0.5809,
      "step": 584
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019808600082801855,
      "loss": 0.2929,
      "step": 585
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019807935906208758,
      "loss": 0.5829,
      "step": 586
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019807270590405234,
      "loss": 0.4853,
      "step": 587
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00019806604135468554,
      "loss": 0.3673,
      "step": 588
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00019805936541476132,
      "loss": 0.6119,
      "step": 589
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00019805267808505509,
      "loss": 0.3506,
      "step": 590
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0001980459793663436,
      "loss": 0.4122,
      "step": 591
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00019803926925940493,
      "loss": 0.4036,
      "step": 592
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00019803254776501846,
      "loss": 0.3933,
      "step": 593
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.0001980258148839649,
      "loss": 0.4893,
      "step": 594
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00019801907061702627,
      "loss": 0.3551,
      "step": 595
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00019801231496498596,
      "loss": 0.6133,
      "step": 596
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00019800554792862867,
      "loss": 0.3696,
      "step": 597
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00019799876950874038,
      "loss": 0.3762,
      "step": 598
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00019799197970610838,
      "loss": 0.3658,
      "step": 599
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00019798517852152135,
      "loss": 0.284,
      "step": 600
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0001979783659557693,
      "loss": 0.4446,
      "step": 601
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00019797154200964343,
      "loss": 0.3505,
      "step": 602
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00019796470668393646,
      "loss": 0.395,
      "step": 603
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00019795785997944225,
      "loss": 0.386,
      "step": 604
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0001979510018969561,
      "loss": 0.3277,
      "step": 605
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00019794413243727455,
      "loss": 0.2902,
      "step": 606
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00019793725160119555,
      "loss": 0.6051,
      "step": 607
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00019793035938951828,
      "loss": 0.4136,
      "step": 608
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0001979234558030433,
      "loss": 0.3152,
      "step": 609
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0001979165408425725,
      "loss": 0.4189,
      "step": 610
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.000197909614508909,
      "loss": 0.3629,
      "step": 611
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00019790267680285735,
      "loss": 0.4443,
      "step": 612
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0001978957277252234,
      "loss": 0.4289,
      "step": 613
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00019788876727681425,
      "loss": 0.4259,
      "step": 614
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00019788179545843842,
      "loss": 0.3419,
      "step": 615
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00019787481227090563,
      "loss": 0.4069,
      "step": 616
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00019786781771502707,
      "loss": 0.326,
      "step": 617
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00019786081179161514,
      "loss": 0.4719,
      "step": 618
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00019785379450148357,
      "loss": 0.3694,
      "step": 619
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00019784676584544745,
      "loss": 0.4971,
      "step": 620
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00019783972582432314,
      "loss": 0.4496,
      "step": 621
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00019783267443892842,
      "loss": 0.3679,
      "step": 622
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00019782561169008224,
      "loss": 0.5644,
      "step": 623
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00019781853757860503,
      "loss": 0.3394,
      "step": 624
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0001978114521053184,
      "loss": 0.3588,
      "step": 625
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0001978043552710454,
      "loss": 0.4644,
      "step": 626
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0001977972470766103,
      "loss": 0.4633,
      "step": 627
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00019779012752283869,
      "loss": 0.4794,
      "step": 628
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00019778299661055764,
      "loss": 0.4752,
      "step": 629
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00019777585434059528,
      "loss": 0.341,
      "step": 630
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00019776870071378132,
      "loss": 0.4361,
      "step": 631
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00019776153573094658,
      "loss": 0.3016,
      "step": 632
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00019775435939292332,
      "loss": 0.3691,
      "step": 633
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00019774717170054507,
      "loss": 0.245,
      "step": 634
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00019773997265464676,
      "loss": 0.5932,
      "step": 635
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00019773276225606447,
      "loss": 0.4991,
      "step": 636
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0001977255405056358,
      "loss": 0.2882,
      "step": 637
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00019771830740419947,
      "loss": 0.4743,
      "step": 638
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00019771106295259568,
      "loss": 0.5216,
      "step": 639
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00019770380715166593,
      "loss": 0.4873,
      "step": 640
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00019769654000225287,
      "loss": 0.4264,
      "step": 641
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00019768926150520074,
      "loss": 0.4799,
      "step": 642
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019768197166135482,
      "loss": 0.5423,
      "step": 643
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019767467047156193,
      "loss": 0.4226,
      "step": 644
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019766735793667005,
      "loss": 0.4955,
      "step": 645
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0001976600340575286,
      "loss": 0.2959,
      "step": 646
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019765269883498826,
      "loss": 0.4561,
      "step": 647
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019764535226990098,
      "loss": 0.2649,
      "step": 648
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019763799436312013,
      "loss": 0.2536,
      "step": 649
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00019763062511550031,
      "loss": 0.3922,
      "step": 650
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00019762324452789748,
      "loss": 0.3382,
      "step": 651
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00019761585260116892,
      "loss": 0.4749,
      "step": 652
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00019760844933617325,
      "loss": 0.2675,
      "step": 653
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00019760103473377027,
      "loss": 0.4435,
      "step": 654
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001975936087948213,
      "loss": 0.343,
      "step": 655
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00019758617152018887,
      "loss": 0.4403,
      "step": 656
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00019757872291073676,
      "loss": 0.4972,
      "step": 657
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001975712629673302,
      "loss": 0.4281,
      "step": 658
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00019756379169083568,
      "loss": 0.3155,
      "step": 659
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00019755630908212094,
      "loss": 0.3099,
      "step": 660
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001975488151420552,
      "loss": 0.3569,
      "step": 661
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001975413098715088,
      "loss": 0.4285,
      "step": 662
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00019753379327135356,
      "loss": 0.4585,
      "step": 663
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001975262653424625,
      "loss": 0.4746,
      "step": 664
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00019751872608571,
      "loss": 0.4333,
      "step": 665
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00019751117550197183,
      "loss": 0.559,
      "step": 666
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00019750361359212491,
      "loss": 0.4815,
      "step": 667
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00019749604035704764,
      "loss": 0.5804,
      "step": 668
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00019748845579761968,
      "loss": 0.2635,
      "step": 669
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00019748085991472188,
      "loss": 0.4177,
      "step": 670
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00019747325270923663,
      "loss": 0.5958,
      "step": 671
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0001974656341820475,
      "loss": 0.382,
      "step": 672
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00019745800433403933,
      "loss": 0.432,
      "step": 673
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001974503631660984,
      "loss": 0.3672,
      "step": 674
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00019744271067911228,
      "loss": 0.4354,
      "step": 675
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00019743504687396972,
      "loss": 0.3247,
      "step": 676
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00019742737175156095,
      "loss": 0.3646,
      "step": 677
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00019741968531277745,
      "loss": 0.4159,
      "step": 678
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.000197411987558512,
      "loss": 0.36,
      "step": 679
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00019740427848965872,
      "loss": 0.2744,
      "step": 680
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019739655810711306,
      "loss": 0.449,
      "step": 681
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019738882641177167,
      "loss": 0.3903,
      "step": 682
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019738108340453268,
      "loss": 0.3458,
      "step": 683
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019737332908629544,
      "loss": 0.3794,
      "step": 684
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019736556345796062,
      "loss": 0.3408,
      "step": 685
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019735778652043017,
      "loss": 0.3885,
      "step": 686
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019734999827460747,
      "loss": 0.4461,
      "step": 687
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00019734219872139712,
      "loss": 0.2602,
      "step": 688
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00019733438786170504,
      "loss": 0.4512,
      "step": 689
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00019732656569643845,
      "loss": 0.3648,
      "step": 690
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00019731873222650597,
      "loss": 0.5536,
      "step": 691
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00019731088745281742,
      "loss": 0.413,
      "step": 692
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.000197303031376284,
      "loss": 0.4932,
      "step": 693
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00019729516399781823,
      "loss": 0.4581,
      "step": 694
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00019728728531833386,
      "loss": 0.5025,
      "step": 695
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00019727939533874608,
      "loss": 0.3666,
      "step": 696
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00019727149405997132,
      "loss": 0.3298,
      "step": 697
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0001972635814829273,
      "loss": 0.3449,
      "step": 698
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0001972556576085331,
      "loss": 0.4254,
      "step": 699
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00019724772243770908,
      "loss": 0.4643,
      "step": 700
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00019723977597137696,
      "loss": 0.3006,
      "step": 701
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00019723181821045967,
      "loss": 0.4121,
      "step": 702
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00019722384915588155,
      "loss": 0.3859,
      "step": 703
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00019721586880856826,
      "loss": 0.3633,
      "step": 704
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001972078771694467,
      "loss": 0.3501,
      "step": 705
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00019719987423944509,
      "loss": 0.5538,
      "step": 706
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.000197191860019493,
      "loss": 0.4472,
      "step": 707
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00019718383451052134,
      "loss": 0.4672,
      "step": 708
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00019717579771346224,
      "loss": 0.3612,
      "step": 709
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001971677496292492,
      "loss": 0.443,
      "step": 710
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00019715969025881704,
      "loss": 0.2191,
      "step": 711
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00019715161960310185,
      "loss": 0.2183,
      "step": 712
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00019714353766304106,
      "loss": 0.2933,
      "step": 713
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00019713544443957342,
      "loss": 0.4959,
      "step": 714
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00019712733993363892,
      "loss": 0.4614,
      "step": 715
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00019711922414617896,
      "loss": 0.3833,
      "step": 716
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0001971110970781362,
      "loss": 0.3107,
      "step": 717
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00019710295873045462,
      "loss": 0.3033,
      "step": 718
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00019709480910407945,
      "loss": 0.3975,
      "step": 719
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00019708664819995735,
      "loss": 0.3954,
      "step": 720
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0001970784760190362,
      "loss": 0.4631,
      "step": 721
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00019707029256226522,
      "loss": 0.2907,
      "step": 722
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0001970620978305949,
      "loss": 0.4039,
      "step": 723
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00019705389182497713,
      "loss": 0.4677,
      "step": 724
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00019704567454636502,
      "loss": 0.3156,
      "step": 725
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00019703744599571302,
      "loss": 0.3899,
      "step": 726
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0001970292061739769,
      "loss": 0.3481,
      "step": 727
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00019702095508211375,
      "loss": 0.4538,
      "step": 728
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00019701269272108193,
      "loss": 0.293,
      "step": 729
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00019700441909184112,
      "loss": 0.335,
      "step": 730
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00019699613419535232,
      "loss": 0.4475,
      "step": 731
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00019698783803257785,
      "loss": 0.262,
      "step": 732
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00019697953060448134,
      "loss": 0.4223,
      "step": 733
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.0001969712119120277,
      "loss": 0.371,
      "step": 734
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00019696288195618312,
      "loss": 0.4543,
      "step": 735
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00019695454073791514,
      "loss": 0.3358,
      "step": 736
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0001969461882581927,
      "loss": 0.3762,
      "step": 737
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0001969378245179859,
      "loss": 0.311,
      "step": 738
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00019692944951826618,
      "loss": 0.3588,
      "step": 739
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00019692106326000633,
      "loss": 0.336,
      "step": 740
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00019691266574418047,
      "loss": 0.3955,
      "step": 741
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00019690425697176396,
      "loss": 0.3129,
      "step": 742
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0001968958369437335,
      "loss": 0.4329,
      "step": 743
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.000196887405661067,
      "loss": 0.4609,
      "step": 744
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00019687896312474393,
      "loss": 0.4475,
      "step": 745
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0001968705093357448,
      "loss": 0.3866,
      "step": 746
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00019686204429505155,
      "loss": 0.5112,
      "step": 747
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00019685356800364743,
      "loss": 0.3883,
      "step": 748
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00019684508046251696,
      "loss": 0.5395,
      "step": 749
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00019683658167264601,
      "loss": 0.3663,
      "step": 750
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0001968280716350217,
      "loss": 0.3175,
      "step": 751
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0001968195503506325,
      "loss": 0.2949,
      "step": 752
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00019681101782046822,
      "loss": 0.4723,
      "step": 753
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00019680247404551984,
      "loss": 0.4848,
      "step": 754
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00019679391902677976,
      "loss": 0.3346,
      "step": 755
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00019678535276524173,
      "loss": 0.3134,
      "step": 756
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00019677677526190067,
      "loss": 0.5254,
      "step": 757
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00019676818651775287,
      "loss": 0.2757,
      "step": 758
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00019675958653379597,
      "loss": 0.4769,
      "step": 759
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00019675097531102886,
      "loss": 0.3951,
      "step": 760
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00019674235285045173,
      "loss": 0.3977,
      "step": 761
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00019673371915306613,
      "loss": 0.2992,
      "step": 762
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00019672507421987482,
      "loss": 0.378,
      "step": 763
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00019671641805188203,
      "loss": 0.4027,
      "step": 764
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00019670775065009307,
      "loss": 0.4151,
      "step": 765
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00019669907201551478,
      "loss": 0.3575,
      "step": 766
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001966903821491551,
      "loss": 0.3742,
      "step": 767
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00019668168105202346,
      "loss": 0.4871,
      "step": 768
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.39378172159194946,
      "eval_runtime": 126.0443,
      "eval_samples_per_second": 1.016,
      "eval_steps_per_second": 0.341,
      "step": 768
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00019667296872513047,
      "loss": 0.4258,
      "step": 769
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001966642451694881,
      "loss": 0.4159,
      "step": 770
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00019665551038610958,
      "loss": 0.3253,
      "step": 771
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00019664676437600954,
      "loss": 0.3057,
      "step": 772
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00019663800714020372,
      "loss": 0.2249,
      "step": 773
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0001966292386797094,
      "loss": 0.3716,
      "step": 774
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00019662045899554502,
      "loss": 0.3536,
      "step": 775
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00019661166808873035,
      "loss": 0.3721,
      "step": 776
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00019660286596028646,
      "loss": 0.3523,
      "step": 777
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00019659405261123576,
      "loss": 0.3123,
      "step": 778
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00019658522804260193,
      "loss": 0.2925,
      "step": 779
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00019657639225540994,
      "loss": 0.3948,
      "step": 780
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0001965675452506861,
      "loss": 0.251,
      "step": 781
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00019655868702945803,
      "loss": 0.2469,
      "step": 782
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00019654981759275458,
      "loss": 0.279,
      "step": 783
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00019654093694160602,
      "loss": 0.3305,
      "step": 784
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00019653204507704376,
      "loss": 0.3095,
      "step": 785
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00019652314200010069,
      "loss": 0.4169,
      "step": 786
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00019651422771181087,
      "loss": 0.3093,
      "step": 787
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00019650530221320976,
      "loss": 0.3405,
      "step": 788
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.000196496365505334,
      "loss": 0.2096,
      "step": 789
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00019648741758922168,
      "loss": 0.3772,
      "step": 790
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0001964784584659121,
      "loss": 0.2245,
      "step": 791
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00019646948813644585,
      "loss": 0.3697,
      "step": 792
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00019646050660186487,
      "loss": 0.3763,
      "step": 793
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00019645151386321236,
      "loss": 0.3277,
      "step": 794
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00019644250992153288,
      "loss": 0.3424,
      "step": 795
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00019643349477787222,
      "loss": 0.3075,
      "step": 796
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00019642446843327752,
      "loss": 0.4018,
      "step": 797
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00019641543088879727,
      "loss": 0.2499,
      "step": 798
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00019640638214548107,
      "loss": 0.1815,
      "step": 799
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00019639732220438005,
      "loss": 0.4432,
      "step": 800
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0001963882510665465,
      "loss": 0.2972,
      "step": 801
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00019637916873303408,
      "loss": 0.2499,
      "step": 802
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00019637007520489767,
      "loss": 0.2781,
      "step": 803
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00019636097048319353,
      "loss": 0.3808,
      "step": 804
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001963518545689792,
      "loss": 0.3636,
      "step": 805
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00019634272746331348,
      "loss": 0.2766,
      "step": 806
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00019633358916725655,
      "loss": 0.3313,
      "step": 807
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001963244396818698,
      "loss": 0.3362,
      "step": 808
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.000196315279008216,
      "loss": 0.2669,
      "step": 809
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001963061071473591,
      "loss": 0.2483,
      "step": 810
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001962969241003645,
      "loss": 0.4369,
      "step": 811
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0001962877298682988,
      "loss": 0.4452,
      "step": 812
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00019627852445222991,
      "loss": 0.3319,
      "step": 813
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00019626930785322712,
      "loss": 0.2258,
      "step": 814
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00019626008007236095,
      "loss": 0.2945,
      "step": 815
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00019625084111070314,
      "loss": 0.3458,
      "step": 816
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00019624159096932688,
      "loss": 0.3135,
      "step": 817
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0001962323296493066,
      "loss": 0.2852,
      "step": 818
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00019622305715171794,
      "loss": 0.2731,
      "step": 819
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00019621377347763802,
      "loss": 0.2995,
      "step": 820
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00019620447862814507,
      "loss": 0.4437,
      "step": 821
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00019619517260431878,
      "loss": 0.2496,
      "step": 822
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00019618585540724003,
      "loss": 0.4478,
      "step": 823
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00019617652703799102,
      "loss": 0.322,
      "step": 824
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00019616718749765528,
      "loss": 0.3728,
      "step": 825
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00019615783678731758,
      "loss": 0.272,
      "step": 826
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00019614847490806407,
      "loss": 0.466,
      "step": 827
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00019613910186098212,
      "loss": 0.4062,
      "step": 828
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.0001961297176471604,
      "loss": 0.5759,
      "step": 829
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00019612032226768897,
      "loss": 0.3071,
      "step": 830
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00019611091572365908,
      "loss": 0.3356,
      "step": 831
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00019610149801616331,
      "loss": 0.2425,
      "step": 832
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00019609206914629557,
      "loss": 0.3167,
      "step": 833
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00019608262911515104,
      "loss": 0.4198,
      "step": 834
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00019607317792382618,
      "loss": 0.2592,
      "step": 835
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00019606371557341875,
      "loss": 0.2562,
      "step": 836
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00019605424206502788,
      "loss": 0.3915,
      "step": 837
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00019604475739975387,
      "loss": 0.3362,
      "step": 838
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00019603526157869846,
      "loss": 0.2693,
      "step": 839
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00019602575460296451,
      "loss": 0.3727,
      "step": 840
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00019601623647365633,
      "loss": 0.1955,
      "step": 841
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00019600670719187944,
      "loss": 0.3112,
      "step": 842
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00019599716675874077,
      "loss": 0.2826,
      "step": 843
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00019598761517534836,
      "loss": 0.2735,
      "step": 844
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00019597805244281168,
      "loss": 0.1933,
      "step": 845
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00019596847856224143,
      "loss": 0.2888,
      "step": 846
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0001959588935347497,
      "loss": 0.2562,
      "step": 847
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00019594929736144976,
      "loss": 0.3926,
      "step": 848
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00019593969004345622,
      "loss": 0.4679,
      "step": 849
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00019593007158188503,
      "loss": 0.4017,
      "step": 850
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00019592044197785336,
      "loss": 0.25,
      "step": 851
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00019591080123247968,
      "loss": 0.2956,
      "step": 852
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00019590114934688385,
      "loss": 0.2996,
      "step": 853
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00019589148632218693,
      "loss": 0.3169,
      "step": 854
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00019588181215951126,
      "loss": 0.3477,
      "step": 855
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00019587212685998057,
      "loss": 0.4385,
      "step": 856
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00019586243042471975,
      "loss": 0.2828,
      "step": 857
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00019585272285485515,
      "loss": 0.2466,
      "step": 858
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00019584300415151424,
      "loss": 0.4562,
      "step": 859
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00019583327431582593,
      "loss": 0.3024,
      "step": 860
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00019582353334892034,
      "loss": 0.2594,
      "step": 861
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00019581378125192888,
      "loss": 0.341,
      "step": 862
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00019580401802598432,
      "loss": 0.3997,
      "step": 863
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00019579424367222061,
      "loss": 0.3042,
      "step": 864
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0001957844581917731,
      "loss": 0.3772,
      "step": 865
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00019577466158577842,
      "loss": 0.4009,
      "step": 866
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0001957648538553744,
      "loss": 0.3383,
      "step": 867
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00019575503500170028,
      "loss": 0.3132,
      "step": 868
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0001957452050258965,
      "loss": 0.3554,
      "step": 869
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00019573536392910486,
      "loss": 0.39,
      "step": 870
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00019572551171246843,
      "loss": 0.3972,
      "step": 871
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00019571564837713153,
      "loss": 0.3257,
      "step": 872
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0001957057739242398,
      "loss": 0.391,
      "step": 873
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00019569588835494024,
      "loss": 0.2231,
      "step": 874
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.000195685991670381,
      "loss": 0.2611,
      "step": 875
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00019567608387171165,
      "loss": 0.3928,
      "step": 876
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00019566616496008297,
      "loss": 0.2767,
      "step": 877
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0001956562349366471,
      "loss": 0.3578,
      "step": 878
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00019564629380255735,
      "loss": 0.2069,
      "step": 879
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0001956363415589685,
      "loss": 0.3666,
      "step": 880
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00019562637820703648,
      "loss": 0.2841,
      "step": 881
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00019561640374791855,
      "loss": 0.2757,
      "step": 882
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00019560641818277324,
      "loss": 0.3934,
      "step": 883
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0001955964215127604,
      "loss": 0.2819,
      "step": 884
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00019558641373904122,
      "loss": 0.2897,
      "step": 885
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00019557639486277807,
      "loss": 0.3542,
      "step": 886
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00019556636488513466,
      "loss": 0.3016,
      "step": 887
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00019555632380727602,
      "loss": 0.3754,
      "step": 888
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0001955462716303684,
      "loss": 0.3327,
      "step": 889
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00019553620835557943,
      "loss": 0.3017,
      "step": 890
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00019552613398407794,
      "loss": 0.3822,
      "step": 891
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00019551604851703408,
      "loss": 0.2675,
      "step": 892
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00019550595195561932,
      "loss": 0.327,
      "step": 893
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00019549584430100637,
      "loss": 0.3951,
      "step": 894
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00019548572555436931,
      "loss": 0.3737,
      "step": 895
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.0001954755957168834,
      "loss": 0.4491,
      "step": 896
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00019546545478972527,
      "loss": 0.3841,
      "step": 897
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00019545530277407277,
      "loss": 0.2815,
      "step": 898
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0001954451396711051,
      "loss": 0.2377,
      "step": 899
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00019543496548200272,
      "loss": 0.3446,
      "step": 900
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0001954247802079474,
      "loss": 0.2723,
      "step": 901
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00019541458385012215,
      "loss": 0.3397,
      "step": 902
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00019540437640971133,
      "loss": 0.2867,
      "step": 903
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00019539415788790047,
      "loss": 0.3577,
      "step": 904
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0001953839282858766,
      "loss": 0.4501,
      "step": 905
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00019537368760482783,
      "loss": 0.3251,
      "step": 906
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0001953634358459436,
      "loss": 0.2571,
      "step": 907
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00019535317301041475,
      "loss": 0.4466,
      "step": 908
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00019534289909943332,
      "loss": 0.3144,
      "step": 909
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00019533261411419257,
      "loss": 0.3949,
      "step": 910
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00019532231805588718,
      "loss": 0.3376,
      "step": 911
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00019531201092571308,
      "loss": 0.2345,
      "step": 912
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00019530169272486737,
      "loss": 0.2388,
      "step": 913
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00019529136345454863,
      "loss": 0.3143,
      "step": 914
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00019528102311595658,
      "loss": 0.3764,
      "step": 915
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00019527067171029225,
      "loss": 0.2668,
      "step": 916
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00019526030923875802,
      "loss": 0.3442,
      "step": 917
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0001952499357025575,
      "loss": 0.3139,
      "step": 918
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00019523955110289554,
      "loss": 0.2875,
      "step": 919
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00019522915544097842,
      "loss": 0.2848,
      "step": 920
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00019521874871801355,
      "loss": 0.2545,
      "step": 921
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001952083309352097,
      "loss": 0.2817,
      "step": 922
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00019519790209377696,
      "loss": 0.3223,
      "step": 923
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001951874621949266,
      "loss": 0.2461,
      "step": 924
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00019517701123987128,
      "loss": 0.2298,
      "step": 925
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00019516654922982485,
      "loss": 0.3617,
      "step": 926
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00019515607616600253,
      "loss": 0.3293,
      "step": 927
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00019514559204962078,
      "loss": 0.3168,
      "step": 928
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00019513509688189737,
      "loss": 0.2559,
      "step": 929
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001951245906640513,
      "loss": 0.2614,
      "step": 930
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00019511407339730289,
      "loss": 0.1547,
      "step": 931
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00019510354508287377,
      "loss": 0.2791,
      "step": 932
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00019509300572198676,
      "loss": 0.3189,
      "step": 933
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001950824553158661,
      "loss": 0.238,
      "step": 934
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00019507189386573724,
      "loss": 0.2883,
      "step": 935
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00019506132137282685,
      "loss": 0.2701,
      "step": 936
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.000195050737838363,
      "loss": 0.3448,
      "step": 937
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00019504014326357494,
      "loss": 0.3287,
      "step": 938
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00019502953764969326,
      "loss": 0.3484,
      "step": 939
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00019501892099794987,
      "loss": 0.3451,
      "step": 940
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00019500829330957787,
      "loss": 0.2606,
      "step": 941
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0001949976545858117,
      "loss": 0.405,
      "step": 942
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00019498700482788704,
      "loss": 0.2153,
      "step": 943
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00019497634403704096,
      "loss": 0.3551,
      "step": 944
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00019496567221451164,
      "loss": 0.2599,
      "step": 945
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001949549893615387,
      "loss": 0.2572,
      "step": 946
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001949442954793629,
      "loss": 0.3406,
      "step": 947
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00019493359056922641,
      "loss": 0.2875,
      "step": 948
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00019492287463237262,
      "loss": 0.4013,
      "step": 949
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00019491214767004621,
      "loss": 0.4147,
      "step": 950
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00019490140968349314,
      "loss": 0.326,
      "step": 951
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00019489066067396063,
      "loss": 0.2902,
      "step": 952
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00019487990064269716,
      "loss": 0.4147,
      "step": 953
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0001948691295909526,
      "loss": 0.3967,
      "step": 954
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.000194858347519978,
      "loss": 0.3386,
      "step": 955
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00019484755443102572,
      "loss": 0.3928,
      "step": 956
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0001948367503253494,
      "loss": 0.2456,
      "step": 957
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00019482593520420393,
      "loss": 0.379,
      "step": 958
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00019481510906884555,
      "loss": 0.2581,
      "step": 959
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00019480427192053172,
      "loss": 0.2874,
      "step": 960
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00019479342376052116,
      "loss": 0.3278,
      "step": 961
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00019478256459007394,
      "loss": 0.2587,
      "step": 962
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001947716944104514,
      "loss": 0.3273,
      "step": 963
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001947608132229161,
      "loss": 0.3379,
      "step": 964
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00019474992102873188,
      "loss": 0.3002,
      "step": 965
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00019473901782916396,
      "loss": 0.3688,
      "step": 966
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00019472810362547872,
      "loss": 0.3492,
      "step": 967
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00019471717841894383,
      "loss": 0.3043,
      "step": 968
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00019470624221082835,
      "loss": 0.3182,
      "step": 969
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00019469529500240252,
      "loss": 0.3463,
      "step": 970
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00019468433679493784,
      "loss": 0.3291,
      "step": 971
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00019467336758970718,
      "loss": 0.3005,
      "step": 972
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00019466238738798458,
      "loss": 0.2085,
      "step": 973
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00019465139619104547,
      "loss": 0.2524,
      "step": 974
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00019464039400016644,
      "loss": 0.3744,
      "step": 975
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00019462938081662548,
      "loss": 0.2755,
      "step": 976
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0001946183566417017,
      "loss": 0.3592,
      "step": 977
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0001946073214766757,
      "loss": 0.2044,
      "step": 978
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0001945962753228291,
      "loss": 0.1866,
      "step": 979
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00019458521818144507,
      "loss": 0.3281,
      "step": 980
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00019457415005380782,
      "loss": 0.333,
      "step": 981
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00019456307094120297,
      "loss": 0.3888,
      "step": 982
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00019455198084491736,
      "loss": 0.2315,
      "step": 983
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00019454087976623912,
      "loss": 0.3785,
      "step": 984
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00019452976770645774,
      "loss": 0.2856,
      "step": 985
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00019451864466686383,
      "loss": 0.2846,
      "step": 986
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00019450751064874932,
      "loss": 0.2623,
      "step": 987
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00019449636565340757,
      "loss": 0.3268,
      "step": 988
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00019448520968213298,
      "loss": 0.2488,
      "step": 989
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00019447404273622136,
      "loss": 0.3303,
      "step": 990
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00019446286481696984,
      "loss": 0.2962,
      "step": 991
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00019445167592567665,
      "loss": 0.2631,
      "step": 992
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0001944404760636415,
      "loss": 0.2803,
      "step": 993
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00019442926523216523,
      "loss": 0.3368,
      "step": 994
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00019441804343254998,
      "loss": 0.3139,
      "step": 995
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0001944068106660992,
      "loss": 0.2348,
      "step": 996
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00019439556693411764,
      "loss": 0.4491,
      "step": 997
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00019438431223791123,
      "loss": 0.3262,
      "step": 998
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00019437304657878723,
      "loss": 0.2067,
      "step": 999
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00019436176995805417,
      "loss": 0.2212,
      "step": 1000
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00019435048237702188,
      "loss": 0.3714,
      "step": 1001
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0001943391838370014,
      "loss": 0.2876,
      "step": 1002
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00019432787433930506,
      "loss": 0.2204,
      "step": 1003
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00019431655388524655,
      "loss": 0.4956,
      "step": 1004
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00019430522247614068,
      "loss": 0.3084,
      "step": 1005
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00019429388011330367,
      "loss": 0.3344,
      "step": 1006
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00019428252679805296,
      "loss": 0.2964,
      "step": 1007
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0001942711625317072,
      "loss": 0.2835,
      "step": 1008
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00019425978731558644,
      "loss": 0.305,
      "step": 1009
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0001942484011510119,
      "loss": 0.3776,
      "step": 1010
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00019423700403930612,
      "loss": 0.2737,
      "step": 1011
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00019422559598179285,
      "loss": 0.2516,
      "step": 1012
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00019421417697979725,
      "loss": 0.2163,
      "step": 1013
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00019420274703464555,
      "loss": 0.2015,
      "step": 1014
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00019419130614766546,
      "loss": 0.2409,
      "step": 1015
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00019417985432018577,
      "loss": 0.246,
      "step": 1016
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0001941683915535367,
      "loss": 0.2794,
      "step": 1017
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0001941569178490496,
      "loss": 0.319,
      "step": 1018
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00019414543320805727,
      "loss": 0.3483,
      "step": 1019
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00019413393763189357,
      "loss": 0.2229,
      "step": 1020
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0001941224311218938,
      "loss": 0.3954,
      "step": 1021
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00019411091367939443,
      "loss": 0.2752,
      "step": 1022
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00019409938530573327,
      "loss": 0.2538,
      "step": 1023
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00019408784600224928,
      "loss": 0.2262,
      "step": 1024
    },
    {
      "epoch": 1.33,
      "eval_loss": 0.3685576915740967,
      "eval_runtime": 126.0989,
      "eval_samples_per_second": 1.015,
      "eval_steps_per_second": 0.341,
      "step": 1024
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00019407629577028286,
      "loss": 0.2382,
      "step": 1025
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00019406473461117552,
      "loss": 0.3778,
      "step": 1026
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00019405316252627019,
      "loss": 0.2761,
      "step": 1027
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00019404157951691096,
      "loss": 0.3233,
      "step": 1028
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00019402998558444315,
      "loss": 0.3098,
      "step": 1029
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00019401838073021354,
      "loss": 0.2829,
      "step": 1030
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0001940067649555699,
      "loss": 0.3732,
      "step": 1031
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00019399513826186157,
      "loss": 0.3161,
      "step": 1032
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00019398350065043894,
      "loss": 0.2405,
      "step": 1033
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0001939718521226538,
      "loss": 0.2629,
      "step": 1034
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00019396019267985907,
      "loss": 0.3626,
      "step": 1035
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00019394852232340905,
      "loss": 0.2626,
      "step": 1036
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001939368410546593,
      "loss": 0.3953,
      "step": 1037
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00019392514887496656,
      "loss": 0.3477,
      "step": 1038
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00019391344578568895,
      "loss": 0.2979,
      "step": 1039
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001939017317881858,
      "loss": 0.3475,
      "step": 1040
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00019389000688381774,
      "loss": 0.3291,
      "step": 1041
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00019387827107394655,
      "loss": 0.2084,
      "step": 1042
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00019386652435993544,
      "loss": 0.3352,
      "step": 1043
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00019385476674314882,
      "loss": 0.2507,
      "step": 1044
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0001938429982249523,
      "loss": 0.2267,
      "step": 1045
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00019383121880671287,
      "loss": 0.2711,
      "step": 1046
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0001938194284897987,
      "loss": 0.2541,
      "step": 1047
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0001938076272755793,
      "loss": 0.2554,
      "step": 1048
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00019379581516542534,
      "loss": 0.334,
      "step": 1049
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0001937839921607089,
      "loss": 0.1763,
      "step": 1050
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00019377215826280316,
      "loss": 0.2836,
      "step": 1051
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00019376031347308268,
      "loss": 0.3321,
      "step": 1052
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00019374845779292324,
      "loss": 0.2917,
      "step": 1053
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00019373659122370198,
      "loss": 0.2535,
      "step": 1054
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00019372471376679715,
      "loss": 0.3794,
      "step": 1055
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00019371282542358836,
      "loss": 0.3111,
      "step": 1056
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00019370092619545645,
      "loss": 0.3477,
      "step": 1057
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00019368901608378354,
      "loss": 0.219,
      "step": 1058
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00019367709508995304,
      "loss": 0.4126,
      "step": 1059
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00019366516321534956,
      "loss": 0.3947,
      "step": 1060
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00019365322046135903,
      "loss": 0.3675,
      "step": 1061
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00019364126682936862,
      "loss": 0.3851,
      "step": 1062
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00019362930232076677,
      "loss": 0.3431,
      "step": 1063
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0001936173269369432,
      "loss": 0.3519,
      "step": 1064
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0001936053406792888,
      "loss": 0.2915,
      "step": 1065
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00019359334354919586,
      "loss": 0.3292,
      "step": 1066
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0001935813355480579,
      "loss": 0.4476,
      "step": 1067
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00019356931667726957,
      "loss": 0.3128,
      "step": 1068
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.000193557286938227,
      "loss": 0.2429,
      "step": 1069
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00019354524633232738,
      "loss": 0.2881,
      "step": 1070
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00019353319486096928,
      "loss": 0.2746,
      "step": 1071
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0001935211325255525,
      "loss": 0.239,
      "step": 1072
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00019350905932747812,
      "loss": 0.2141,
      "step": 1073
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00019349697526814844,
      "loss": 0.2655,
      "step": 1074
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00019348488034896704,
      "loss": 0.25,
      "step": 1075
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00019347277457133882,
      "loss": 0.3764,
      "step": 1076
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00019346065793666986,
      "loss": 0.3347,
      "step": 1077
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00019344853044636748,
      "loss": 0.3458,
      "step": 1078
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0001934363921018404,
      "loss": 0.359,
      "step": 1079
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00019342424290449842,
      "loss": 0.1712,
      "step": 1080
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00019341208285575276,
      "loss": 0.298,
      "step": 1081
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00019339991195701584,
      "loss": 0.2617,
      "step": 1082
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00019338773020970127,
      "loss": 0.288,
      "step": 1083
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.000193375537615224,
      "loss": 0.1474,
      "step": 1084
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0001933633341750003,
      "loss": 0.3504,
      "step": 1085
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00019335111989044753,
      "loss": 0.2316,
      "step": 1086
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00019333889476298445,
      "loss": 0.2308,
      "step": 1087
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.000193326658794031,
      "loss": 0.2105,
      "step": 1088
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0001933144119850084,
      "loss": 0.3845,
      "step": 1089
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00019330215433733923,
      "loss": 0.3707,
      "step": 1090
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00019328988585244714,
      "loss": 0.3099,
      "step": 1091
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0001932776065317572,
      "loss": 0.3548,
      "step": 1092
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00019326531637669562,
      "loss": 0.3725,
      "step": 1093
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00019325301538869,
      "loss": 0.2787,
      "step": 1094
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00019324070356916906,
      "loss": 0.2073,
      "step": 1095
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.0001932283809195629,
      "loss": 0.3209,
      "step": 1096
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00019321604744130272,
      "loss": 0.2468,
      "step": 1097
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00019320370313582113,
      "loss": 0.334,
      "step": 1098
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00019319134800455202,
      "loss": 0.3051,
      "step": 1099
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00019317898204893033,
      "loss": 0.2047,
      "step": 1100
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00019316660527039252,
      "loss": 0.2682,
      "step": 1101
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00019315421767037604,
      "loss": 0.3717,
      "step": 1102
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00019314181925031987,
      "loss": 0.2567,
      "step": 1103
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00019312941001166403,
      "loss": 0.2155,
      "step": 1104
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00019311698995584986,
      "loss": 0.3092,
      "step": 1105
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00019310455908432,
      "loss": 0.3117,
      "step": 1106
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00019309211739851836,
      "loss": 0.281,
      "step": 1107
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00019307966489989,
      "loss": 0.2761,
      "step": 1108
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00019306720158988134,
      "loss": 0.3296,
      "step": 1109
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00019305472746994,
      "loss": 0.291,
      "step": 1110
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0001930422425415149,
      "loss": 0.2714,
      "step": 1111
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00019302974680605618,
      "loss": 0.2245,
      "step": 1112
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00019301724026501518,
      "loss": 0.486,
      "step": 1113
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00019300472291984466,
      "loss": 0.2837,
      "step": 1114
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00019299219477199845,
      "loss": 0.3475,
      "step": 1115
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0001929796558229318,
      "loss": 0.2475,
      "step": 1116
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00019296710607410105,
      "loss": 0.1813,
      "step": 1117
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00019295454552696392,
      "loss": 0.3507,
      "step": 1118
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0001929419741829793,
      "loss": 0.3116,
      "step": 1119
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00019292939204360745,
      "loss": 0.2863,
      "step": 1120
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00019291679911030975,
      "loss": 0.3303,
      "step": 1121
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0001929041953845489,
      "loss": 0.2179,
      "step": 1122
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00019289158086778892,
      "loss": 0.2529,
      "step": 1123
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0001928789555614949,
      "loss": 0.3153,
      "step": 1124
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00019286631946713334,
      "loss": 0.292,
      "step": 1125
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00019285367258617194,
      "loss": 0.3219,
      "step": 1126
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001928410149200797,
      "loss": 0.2845,
      "step": 1127
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001928283464703268,
      "loss": 0.3506,
      "step": 1128
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00019281566723838468,
      "loss": 0.2702,
      "step": 1129
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001928029772257261,
      "loss": 0.2335,
      "step": 1130
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00019279027643382497,
      "loss": 0.2799,
      "step": 1131
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001927775648641566,
      "loss": 0.3922,
      "step": 1132
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00019276484251819738,
      "loss": 0.3061,
      "step": 1133
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001927521093974251,
      "loss": 0.3252,
      "step": 1134
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00019273936550331868,
      "loss": 0.3119,
      "step": 1135
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0001927266108373584,
      "loss": 0.4087,
      "step": 1136
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0001927138454010257,
      "loss": 0.3443,
      "step": 1137
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00019270106919580334,
      "loss": 0.3264,
      "step": 1138
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00019268828222317522,
      "loss": 0.164,
      "step": 1139
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0001926754844846267,
      "loss": 0.3195,
      "step": 1140
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00019266267598164416,
      "loss": 0.2899,
      "step": 1141
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00019264985671571538,
      "loss": 0.3437,
      "step": 1142
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00019263702668832931,
      "loss": 0.3296,
      "step": 1143
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00019262418590097622,
      "loss": 0.2888,
      "step": 1144
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00019261133435514758,
      "loss": 0.1965,
      "step": 1145
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00019259847205233612,
      "loss": 0.3131,
      "step": 1146
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0001925855989940358,
      "loss": 0.2681,
      "step": 1147
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00019257271518174186,
      "loss": 0.2894,
      "step": 1148
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0001925598206169508,
      "loss": 0.3103,
      "step": 1149
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00019254691530116031,
      "loss": 0.2432,
      "step": 1150
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00019253399923586942,
      "loss": 0.2876,
      "step": 1151
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00019252107242257832,
      "loss": 0.2255,
      "step": 1152
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00019250813486278848,
      "loss": 0.2662,
      "step": 1153
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00019249518655800263,
      "loss": 0.2769,
      "step": 1154
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00019248222750972477,
      "loss": 0.32,
      "step": 1155
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00019246925771946005,
      "loss": 0.3298,
      "step": 1156
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.000192456277188715,
      "loss": 0.3429,
      "step": 1157
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.0001924432859189973,
      "loss": 0.3646,
      "step": 1158
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00019243028391181593,
      "loss": 0.3323,
      "step": 1159
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00019241727116868108,
      "loss": 0.2312,
      "step": 1160
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00019240424769110422,
      "loss": 0.3179,
      "step": 1161
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00019239121348059804,
      "loss": 0.2549,
      "step": 1162
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00019237816853867653,
      "loss": 0.2725,
      "step": 1163
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00019236511286685478,
      "loss": 0.3727,
      "step": 1164
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00019235204646664934,
      "loss": 0.3319,
      "step": 1165
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00019233896933957786,
      "loss": 0.3974,
      "step": 1166
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00019232588148715925,
      "loss": 0.2445,
      "step": 1167
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00019231278291091374,
      "loss": 0.3839,
      "step": 1168
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0001922996736123627,
      "loss": 0.3148,
      "step": 1169
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00019228655359302882,
      "loss": 0.3836,
      "step": 1170
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00019227342285443603,
      "loss": 0.4034,
      "step": 1171
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00019226028139810946,
      "loss": 0.2433,
      "step": 1172
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00019224712922557557,
      "loss": 0.2529,
      "step": 1173
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00019223396633836196,
      "loss": 0.3506,
      "step": 1174
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00019222079273799753,
      "loss": 0.2858,
      "step": 1175
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00019220760842601244,
      "loss": 0.3681,
      "step": 1176
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00019219441340393803,
      "loss": 0.35,
      "step": 1177
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00019218120767330697,
      "loss": 0.1697,
      "step": 1178
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00019216799123565313,
      "loss": 0.2607,
      "step": 1179
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0001921547640925116,
      "loss": 0.3,
      "step": 1180
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0001921415262454188,
      "loss": 0.3574,
      "step": 1181
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0001921282776959122,
      "loss": 0.3801,
      "step": 1182
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0001921150184455308,
      "loss": 0.4554,
      "step": 1183
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00019210174849581456,
      "loss": 0.437,
      "step": 1184
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0001920884678483049,
      "loss": 0.3121,
      "step": 1185
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00019207517650454432,
      "loss": 0.254,
      "step": 1186
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0001920618744660767,
      "loss": 0.2993,
      "step": 1187
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00019204856173444705,
      "loss": 0.2969,
      "step": 1188
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00019203523831120168,
      "loss": 0.3467,
      "step": 1189
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00019202190419788815,
      "loss": 0.171,
      "step": 1190
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00019200855939605522,
      "loss": 0.1801,
      "step": 1191
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00019199520390725293,
      "loss": 0.3069,
      "step": 1192
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00019198183773303255,
      "loss": 0.2364,
      "step": 1193
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00019196846087494657,
      "loss": 0.3313,
      "step": 1194
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00019195507333454875,
      "loss": 0.2404,
      "step": 1195
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00019194167511339406,
      "loss": 0.4415,
      "step": 1196
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00019192826621303874,
      "loss": 0.3399,
      "step": 1197
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00019191484663504027,
      "loss": 0.256,
      "step": 1198
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00019190141638095735,
      "loss": 0.2957,
      "step": 1199
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00019188797545234988,
      "loss": 0.3057,
      "step": 1200
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0001918745238507792,
      "loss": 0.3376,
      "step": 1201
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00019186106157780753,
      "loss": 0.3544,
      "step": 1202
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00019184758863499872,
      "loss": 0.3666,
      "step": 1203
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00019183410502391756,
      "loss": 0.2379,
      "step": 1204
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00019182061074613028,
      "loss": 0.2148,
      "step": 1205
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0001918071058032042,
      "loss": 0.3268,
      "step": 1206
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00019179359019670796,
      "loss": 0.3154,
      "step": 1207
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00019178006392821145,
      "loss": 0.3143,
      "step": 1208
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0001917665269992858,
      "loss": 0.3019,
      "step": 1209
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00019175297941150326,
      "loss": 0.345,
      "step": 1210
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0001917394211664375,
      "loss": 0.3269,
      "step": 1211
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00019172585226566322,
      "loss": 0.3177,
      "step": 1212
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0001917122727107566,
      "loss": 0.3129,
      "step": 1213
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00019169868250329487,
      "loss": 0.3017,
      "step": 1214
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00019168508164485654,
      "loss": 0.3077,
      "step": 1215
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00019167147013702144,
      "loss": 0.3319,
      "step": 1216
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00019165784798137053,
      "loss": 0.1942,
      "step": 1217
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.000191644215179486,
      "loss": 0.3434,
      "step": 1218
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00019163057173295143,
      "loss": 0.2991,
      "step": 1219
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00019161691764335145,
      "loss": 0.2241,
      "step": 1220
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0001916032529122721,
      "loss": 0.3782,
      "step": 1221
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0001915895775413004,
      "loss": 0.3771,
      "step": 1222
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00019157589153202494,
      "loss": 0.2274,
      "step": 1223
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00019156219488603534,
      "loss": 0.4142,
      "step": 1224
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0001915484876049224,
      "loss": 0.2122,
      "step": 1225
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00019153476969027832,
      "loss": 0.2436,
      "step": 1226
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00019152104114369647,
      "loss": 0.2954,
      "step": 1227
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00019150730196677143,
      "loss": 0.1918,
      "step": 1228
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00019149355216109903,
      "loss": 0.3016,
      "step": 1229
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00019147979172827632,
      "loss": 0.3195,
      "step": 1230
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00019146602066990164,
      "loss": 0.3555,
      "step": 1231
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00019145223898757446,
      "loss": 0.2394,
      "step": 1232
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.0001914384466828956,
      "loss": 0.3479,
      "step": 1233
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00019142464375746708,
      "loss": 0.2667,
      "step": 1234
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00019141083021289207,
      "loss": 0.4288,
      "step": 1235
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00019139700605077509,
      "loss": 0.2508,
      "step": 1236
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00019138317127272186,
      "loss": 0.3234,
      "step": 1237
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00019136932588033925,
      "loss": 0.3002,
      "step": 1238
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00019135546987523547,
      "loss": 0.3368,
      "step": 1239
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00019134160325901992,
      "loss": 0.4084,
      "step": 1240
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00019132772603330323,
      "loss": 0.3808,
      "step": 1241
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00019131383819969726,
      "loss": 0.1211,
      "step": 1242
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0001912999397598151,
      "loss": 0.244,
      "step": 1243
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00019128603071527113,
      "loss": 0.2467,
      "step": 1244
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00019127211106768086,
      "loss": 0.2399,
      "step": 1245
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0001912581808186611,
      "loss": 0.2606,
      "step": 1246
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00019124423996982987,
      "loss": 0.2742,
      "step": 1247
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00019123028852280642,
      "loss": 0.2349,
      "step": 1248
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00019121632647921128,
      "loss": 0.3226,
      "step": 1249
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00019120235384066612,
      "loss": 0.2733,
      "step": 1250
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00019118837060879388,
      "loss": 0.353,
      "step": 1251
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0001911743767852188,
      "loss": 0.3002,
      "step": 1252
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00019116037237156626,
      "loss": 0.3115,
      "step": 1253
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00019114635736946286,
      "loss": 0.2753,
      "step": 1254
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.0001911323317805365,
      "loss": 0.3478,
      "step": 1255
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00019111829560641627,
      "loss": 0.2323,
      "step": 1256
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00019110424884873255,
      "loss": 0.3511,
      "step": 1257
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00019109019150911682,
      "loss": 0.2498,
      "step": 1258
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00019107612358920193,
      "loss": 0.3677,
      "step": 1259
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00019106204509062184,
      "loss": 0.1988,
      "step": 1260
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00019104795601501185,
      "loss": 0.2792,
      "step": 1261
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0001910338563640084,
      "loss": 0.2375,
      "step": 1262
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0001910197461392492,
      "loss": 0.2665,
      "step": 1263
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00019100562534237316,
      "loss": 0.3094,
      "step": 1264
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00019099149397502045,
      "loss": 0.3336,
      "step": 1265
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0001909773520388325,
      "loss": 0.3321,
      "step": 1266
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.00019096319953545185,
      "loss": 0.2783,
      "step": 1267
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0001909490364665224,
      "loss": 0.3253,
      "step": 1268
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.00019093486283368915,
      "loss": 0.2892,
      "step": 1269
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.00019092067863859846,
      "loss": 0.2823,
      "step": 1270
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0001909064838828978,
      "loss": 0.3611,
      "step": 1271
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.00019089227856823598,
      "loss": 0.2887,
      "step": 1272
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0001908780626962629,
      "loss": 0.3133,
      "step": 1273
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00019086383626862983,
      "loss": 0.2241,
      "step": 1274
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00019084959928698914,
      "loss": 0.346,
      "step": 1275
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00019083535175299453,
      "loss": 0.404,
      "step": 1276
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00019082109366830083,
      "loss": 0.3876,
      "step": 1277
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00019080682503456417,
      "loss": 0.2603,
      "step": 1278
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0001907925458534419,
      "loss": 0.3152,
      "step": 1279
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00019077825612659255,
      "loss": 0.3053,
      "step": 1280
    },
    {
      "epoch": 1.66,
      "eval_loss": 0.3257058262825012,
      "eval_runtime": 126.1404,
      "eval_samples_per_second": 1.015,
      "eval_steps_per_second": 0.341,
      "step": 1280
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00019076395585567588,
      "loss": 0.2848,
      "step": 1281
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00019074964504235293,
      "loss": 0.2577,
      "step": 1282
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00019073532368828593,
      "loss": 0.3397,
      "step": 1283
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0001907209917951383,
      "loss": 0.2922,
      "step": 1284
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00019070664936457474,
      "loss": 0.2523,
      "step": 1285
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00019069229639826114,
      "loss": 0.3602,
      "step": 1286
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0001906779328978646,
      "loss": 0.3506,
      "step": 1287
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.0001906635588650535,
      "loss": 0.2798,
      "step": 1288
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00019064917430149745,
      "loss": 0.2589,
      "step": 1289
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00019063477920886718,
      "loss": 0.2804,
      "step": 1290
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00019062037358883474,
      "loss": 0.334,
      "step": 1291
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00019060595744307333,
      "loss": 0.2549,
      "step": 1292
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00019059153077325745,
      "loss": 0.2608,
      "step": 1293
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00019057709358106278,
      "loss": 0.3043,
      "step": 1294
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00019056264586816624,
      "loss": 0.2438,
      "step": 1295
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00019054818763624592,
      "loss": 0.2682,
      "step": 1296
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00019053371888698122,
      "loss": 0.333,
      "step": 1297
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00019051923962205267,
      "loss": 0.3476,
      "step": 1298
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0001905047498431421,
      "loss": 0.3281,
      "step": 1299
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00019049024955193248,
      "loss": 0.2256,
      "step": 1300
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00019047573875010808,
      "loss": 0.2789,
      "step": 1301
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00019046121743935434,
      "loss": 0.3995,
      "step": 1302
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00019044668562135796,
      "loss": 0.2473,
      "step": 1303
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00019043214329780684,
      "loss": 0.2712,
      "step": 1304
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00019041759047039007,
      "loss": 0.4017,
      "step": 1305
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00019040302714079803,
      "loss": 0.3127,
      "step": 1306
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0001903884533107222,
      "loss": 0.2442,
      "step": 1307
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00019037386898185544,
      "loss": 0.2669,
      "step": 1308
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00019035927415589175,
      "loss": 0.2689,
      "step": 1309
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0001903446688345263,
      "loss": 0.3205,
      "step": 1310
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00019033005301945553,
      "loss": 0.1659,
      "step": 1311
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.00019031542671237709,
      "loss": 0.2421,
      "step": 1312
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.00019030078991498988,
      "loss": 0.3052,
      "step": 1313
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.00019028614262899397,
      "loss": 0.3252,
      "step": 1314
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.0001902714848560907,
      "loss": 0.303,
      "step": 1315
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.00019025681659798257,
      "loss": 0.4066,
      "step": 1316
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.00019024213785637337,
      "loss": 0.3821,
      "step": 1317
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.000190227448632968,
      "loss": 0.2653,
      "step": 1318
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.00019021274892947265,
      "loss": 0.1936,
      "step": 1319
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00019019803874759477,
      "loss": 0.1962,
      "step": 1320
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00019018331808904292,
      "loss": 0.3207,
      "step": 1321
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00019016858695552696,
      "loss": 0.1277,
      "step": 1322
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00019015384534875796,
      "loss": 0.2242,
      "step": 1323
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00019013909327044814,
      "loss": 0.3564,
      "step": 1324
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.000190124330722311,
      "loss": 0.3023,
      "step": 1325
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00019010955770606128,
      "loss": 0.1655,
      "step": 1326
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00019009477422341484,
      "loss": 0.276,
      "step": 1327
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00019007998027608883,
      "loss": 0.2892,
      "step": 1328
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00019006517586580158,
      "loss": 0.29,
      "step": 1329
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00019005036099427265,
      "loss": 0.2187,
      "step": 1330
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00019003553566322286,
      "loss": 0.2178,
      "step": 1331
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0001900206998743741,
      "loss": 0.3623,
      "step": 1332
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0001900058536294497,
      "loss": 0.3068,
      "step": 1333
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00018999099693017404,
      "loss": 0.2663,
      "step": 1334
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0001899761297782727,
      "loss": 0.2623,
      "step": 1335
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.00018996125217547255,
      "loss": 0.1923,
      "step": 1336
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0001899463641235017,
      "loss": 0.2725,
      "step": 1337
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.00018993146562408937,
      "loss": 0.2273,
      "step": 1338
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0001899165566789661,
      "loss": 0.2597,
      "step": 1339
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0001899016372898635,
      "loss": 0.2557,
      "step": 1340
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.00018988670745851463,
      "loss": 0.2154,
      "step": 1341
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.0001898717671866535,
      "loss": 0.2913,
      "step": 1342
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00018985681647601548,
      "loss": 0.2064,
      "step": 1343
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0001898418553283371,
      "loss": 0.2476,
      "step": 1344
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00018982688374535623,
      "loss": 0.2842,
      "step": 1345
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00018981190172881173,
      "loss": 0.2392,
      "step": 1346
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00018979690928044385,
      "loss": 0.3484,
      "step": 1347
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00018978190640199397,
      "loss": 0.2955,
      "step": 1348
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0001897668930952047,
      "loss": 0.3465,
      "step": 1349
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00018975186936181988,
      "loss": 0.3952,
      "step": 1350
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.00018973683520358454,
      "loss": 0.3037,
      "step": 1351
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.00018972179062224493,
      "loss": 0.3306,
      "step": 1352
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0001897067356195485,
      "loss": 0.2537,
      "step": 1353
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.0001896916701972439,
      "loss": 0.2602,
      "step": 1354
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.00018967659435708105,
      "loss": 0.3817,
      "step": 1355
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.000189661508100811,
      "loss": 0.267,
      "step": 1356
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.00018964641143018606,
      "loss": 0.3266,
      "step": 1357
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.00018963130434695974,
      "loss": 0.4192,
      "step": 1358
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.00018961618685288677,
      "loss": 0.2269,
      "step": 1359
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.00018960105894972308,
      "loss": 0.2297,
      "step": 1360
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.00018958592063922572,
      "loss": 0.2811,
      "step": 1361
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.00018957077192315313,
      "loss": 0.3755,
      "step": 1362
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.00018955561280326483,
      "loss": 0.2225,
      "step": 1363
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0001895404432813216,
      "loss": 0.2326,
      "step": 1364
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0001895252633590854,
      "loss": 0.3409,
      "step": 1365
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00018951007303831934,
      "loss": 0.2308,
      "step": 1366
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00018949487232078788,
      "loss": 0.345,
      "step": 1367
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00018947966120825662,
      "loss": 0.1538,
      "step": 1368
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00018946443970249233,
      "loss": 0.2724,
      "step": 1369
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.000189449207805263,
      "loss": 0.3015,
      "step": 1370
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00018943396551833786,
      "loss": 0.2707,
      "step": 1371
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00018941871284348734,
      "loss": 0.261,
      "step": 1372
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0001894034497824831,
      "loss": 0.2795,
      "step": 1373
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00018938817633709794,
      "loss": 0.2273,
      "step": 1374
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00018937289250910583,
      "loss": 0.3615,
      "step": 1375
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00018935759830028212,
      "loss": 0.2291,
      "step": 1376
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00018934229371240325,
      "loss": 0.3049,
      "step": 1377
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0001893269787472468,
      "loss": 0.2803,
      "step": 1378
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0001893116534065917,
      "loss": 0.208,
      "step": 1379
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.000189296317692218,
      "loss": 0.1992,
      "step": 1380
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00018928097160590698,
      "loss": 0.3837,
      "step": 1381
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0001892656151494411,
      "loss": 0.3224,
      "step": 1382
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00018925024832460406,
      "loss": 0.2423,
      "step": 1383
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00018923487113318074,
      "loss": 0.3356,
      "step": 1384
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00018921948357695722,
      "loss": 0.2908,
      "step": 1385
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.0001892040856577208,
      "loss": 0.2742,
      "step": 1386
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00018918867737726,
      "loss": 0.2531,
      "step": 1387
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00018917325873736446,
      "loss": 0.2981,
      "step": 1388
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00018915782973982515,
      "loss": 0.2435,
      "step": 1389
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00018914239038643414,
      "loss": 0.2498,
      "step": 1390
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00018912694067898476,
      "loss": 0.289,
      "step": 1391
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0001891114806192715,
      "loss": 0.1783,
      "step": 1392
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00018909601020909012,
      "loss": 0.3221,
      "step": 1393
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00018908052945023748,
      "loss": 0.2271,
      "step": 1394
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00018906503834451174,
      "loss": 0.2807,
      "step": 1395
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.0001890495368937122,
      "loss": 0.2764,
      "step": 1396
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00018903402509963941,
      "loss": 0.2495,
      "step": 1397
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00018901850296409507,
      "loss": 0.3514,
      "step": 1398
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0001890029704888821,
      "loss": 0.2085,
      "step": 1399
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00018898742767580465,
      "loss": 0.1708,
      "step": 1400
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00018897187452666803,
      "loss": 0.2994,
      "step": 1401
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00018895631104327878,
      "loss": 0.2363,
      "step": 1402
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00018894073722744464,
      "loss": 0.3174,
      "step": 1403
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00018892515308097446,
      "loss": 0.2667,
      "step": 1404
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00018890955860567847,
      "loss": 0.295,
      "step": 1405
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00018889395380336797,
      "loss": 0.245,
      "step": 1406
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00018887833867585544,
      "loss": 0.1857,
      "step": 1407
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00018886271322495466,
      "loss": 0.1815,
      "step": 1408
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00018884707745248056,
      "loss": 0.1643,
      "step": 1409
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00018883143136024918,
      "loss": 0.2801,
      "step": 1410
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00018881577495007799,
      "loss": 0.1796,
      "step": 1411
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00018880010822378536,
      "loss": 0.214,
      "step": 1412
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0001887844311831911,
      "loss": 0.233,
      "step": 1413
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.0001887687438301161,
      "loss": 0.2752,
      "step": 1414
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00018875304616638248,
      "loss": 0.316,
      "step": 1415
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00018873733819381355,
      "loss": 0.2955,
      "step": 1416
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00018872161991423384,
      "loss": 0.3319,
      "step": 1417
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00018870589132946903,
      "loss": 0.3265,
      "step": 1418
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00018869015244134602,
      "loss": 0.3156,
      "step": 1419
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.000188674403251693,
      "loss": 0.2772,
      "step": 1420
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.00018865864376233914,
      "loss": 0.2706,
      "step": 1421
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.000188642873975115,
      "loss": 0.2608,
      "step": 1422
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.00018862709389185227,
      "loss": 0.2339,
      "step": 1423
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0001886113035143838,
      "loss": 0.2442,
      "step": 1424
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.00018859550284454376,
      "loss": 0.2842,
      "step": 1425
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.00018857969188416736,
      "loss": 0.3515,
      "step": 1426
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.00018856387063509106,
      "loss": 0.2177,
      "step": 1427
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.00018854803909915257,
      "loss": 0.3266,
      "step": 1428
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0001885321972781907,
      "loss": 0.3692,
      "step": 1429
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0001885163451740456,
      "loss": 0.2915,
      "step": 1430
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.00018850048278855843,
      "loss": 0.1637,
      "step": 1431
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0001884846101235717,
      "loss": 0.1503,
      "step": 1432
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.00018846872718092898,
      "loss": 0.3421,
      "step": 1433
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.00018845283396247515,
      "loss": 0.3319,
      "step": 1434
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00018843693047005625,
      "loss": 0.3489,
      "step": 1435
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0001884210167055195,
      "loss": 0.2843,
      "step": 1436
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.0001884050926707133,
      "loss": 0.2905,
      "step": 1437
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00018838915836748722,
      "loss": 0.261,
      "step": 1438
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00018837321379769212,
      "loss": 0.2275,
      "step": 1439
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00018835725896317995,
      "loss": 0.2968,
      "step": 1440
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00018834129386580392,
      "loss": 0.4893,
      "step": 1441
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00018832531850741842,
      "loss": 0.2556,
      "step": 1442
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00018830933288987897,
      "loss": 0.2214,
      "step": 1443
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00018829333701504236,
      "loss": 0.2305,
      "step": 1444
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00018827733088476657,
      "loss": 0.2766,
      "step": 1445
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00018826131450091067,
      "loss": 0.3543,
      "step": 1446
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00018824528786533503,
      "loss": 0.3815,
      "step": 1447
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.0001882292509799012,
      "loss": 0.2006,
      "step": 1448
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00018821320384647188,
      "loss": 0.385,
      "step": 1449
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00018819714646691095,
      "loss": 0.2266,
      "step": 1450
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.00018818107884308353,
      "loss": 0.19,
      "step": 1451
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0001881650009768559,
      "loss": 0.3216,
      "step": 1452
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.00018814891287009553,
      "loss": 0.4322,
      "step": 1453
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.00018813281452467108,
      "loss": 0.2276,
      "step": 1454
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.00018811670594245242,
      "loss": 0.2911,
      "step": 1455
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.00018810058712531057,
      "loss": 0.2621,
      "step": 1456
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0001880844580751178,
      "loss": 0.2503,
      "step": 1457
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00018806831879374748,
      "loss": 0.2535,
      "step": 1458
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00018805216928307426,
      "loss": 0.206,
      "step": 1459
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0001880360095449739,
      "loss": 0.3362,
      "step": 1460
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00018801983958132343,
      "loss": 0.25,
      "step": 1461
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00018800365939400097,
      "loss": 0.3135,
      "step": 1462
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0001879874689848859,
      "loss": 0.2447,
      "step": 1463
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00018797126835585882,
      "loss": 0.3153,
      "step": 1464
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0001879550575088014,
      "loss": 0.3696,
      "step": 1465
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0001879388364455965,
      "loss": 0.3259,
      "step": 1466
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00018792260516812837,
      "loss": 0.3434,
      "step": 1467
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00018790636367828226,
      "loss": 0.3168,
      "step": 1468
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.0001878901119779446,
      "loss": 0.2786,
      "step": 1469
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00018787385006900306,
      "loss": 0.2728,
      "step": 1470
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00018785757795334655,
      "loss": 0.2303,
      "step": 1471
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00018784129563286506,
      "loss": 0.2849,
      "step": 1472
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00018782500310944982,
      "loss": 0.4088,
      "step": 1473
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00018780870038499322,
      "loss": 0.3093,
      "step": 1474
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00018779238746138892,
      "loss": 0.3192,
      "step": 1475
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00018777606434053164,
      "loss": 0.2671,
      "step": 1476
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00018775973102431735,
      "loss": 0.3268,
      "step": 1477
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00018774338751464319,
      "loss": 0.2133,
      "step": 1478
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0001877270338134075,
      "loss": 0.1731,
      "step": 1479
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.0001877106699225098,
      "loss": 0.2849,
      "step": 1480
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0001876942958438508,
      "loss": 0.2706,
      "step": 1481
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0001876779115793323,
      "loss": 0.2535,
      "step": 1482
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0001876615171308575,
      "loss": 0.2635,
      "step": 1483
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0001876451125003305,
      "loss": 0.1989,
      "step": 1484
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.00018762869768965685,
      "loss": 0.2229,
      "step": 1485
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.0001876122727007431,
      "loss": 0.2294,
      "step": 1486
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.00018759583753549705,
      "loss": 0.2833,
      "step": 1487
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.00018757939219582768,
      "loss": 0.2894,
      "step": 1488
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00018756293668364515,
      "loss": 0.2355,
      "step": 1489
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00018754647100086081,
      "loss": 0.39,
      "step": 1490
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00018752999514938713,
      "loss": 0.181,
      "step": 1491
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00018751350913113788,
      "loss": 0.1916,
      "step": 1492
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0001874970129480279,
      "loss": 0.2571,
      "step": 1493
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00018748050660197328,
      "loss": 0.3147,
      "step": 1494
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.0001874639900948912,
      "loss": 0.3191,
      "step": 1495
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00018744746342870018,
      "loss": 0.2968,
      "step": 1496
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00018743092660531975,
      "loss": 0.228,
      "step": 1497
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00018741437962667073,
      "loss": 0.2624,
      "step": 1498
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00018739782249467508,
      "loss": 0.3671,
      "step": 1499
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00018738125521125594,
      "loss": 0.4026,
      "step": 1500
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00018736467777833758,
      "loss": 0.3615,
      "step": 1501
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0001873480901978456,
      "loss": 0.2445,
      "step": 1502
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00018733149247170662,
      "loss": 0.3337,
      "step": 1503
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00018731488460184848,
      "loss": 0.2327,
      "step": 1504
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00018729826659020024,
      "loss": 0.4135,
      "step": 1505
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00018728163843869213,
      "loss": 0.2073,
      "step": 1506
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0001872650001492555,
      "loss": 0.303,
      "step": 1507
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00018724835172382297,
      "loss": 0.196,
      "step": 1508
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00018723169316432823,
      "loss": 0.3536,
      "step": 1509
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00018721502447270624,
      "loss": 0.2573,
      "step": 1510
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0001871983456508931,
      "loss": 0.243,
      "step": 1511
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0001871816567008261,
      "loss": 0.2735,
      "step": 1512
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00018716495762444364,
      "loss": 0.24,
      "step": 1513
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00018714824842368537,
      "loss": 0.2188,
      "step": 1514
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00018713152910049217,
      "loss": 0.2548,
      "step": 1515
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00018711479965680588,
      "loss": 0.3372,
      "step": 1516
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.0001870980600945698,
      "loss": 0.3962,
      "step": 1517
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00018708131041572816,
      "loss": 0.2462,
      "step": 1518
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00018706455062222652,
      "loss": 0.2494,
      "step": 1519
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.00018704778071601153,
      "loss": 0.2707,
      "step": 1520
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0001870310006990311,
      "loss": 0.2443,
      "step": 1521
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0001870142105732342,
      "loss": 0.1915,
      "step": 1522
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0001869974103405711,
      "loss": 0.2649,
      "step": 1523
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0001869806000029931,
      "loss": 0.1946,
      "step": 1524
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.00018696377956245276,
      "loss": 0.3691,
      "step": 1525
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0001869469490209039,
      "loss": 0.1585,
      "step": 1526
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00018693010838030132,
      "loss": 0.2439,
      "step": 1527
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00018691325764260113,
      "loss": 0.2272,
      "step": 1528
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00018689639680976057,
      "loss": 0.3215,
      "step": 1529
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0001868795258837381,
      "loss": 0.2571,
      "step": 1530
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00018686264486649323,
      "loss": 0.2036,
      "step": 1531
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0001868457537599868,
      "loss": 0.26,
      "step": 1532
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00018682885256618068,
      "loss": 0.2675,
      "step": 1533
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00018681194128703803,
      "loss": 0.2373,
      "step": 1534
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00018679501992452307,
      "loss": 0.3953,
      "step": 1535
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0001867780884806013,
      "loss": 0.2438,
      "step": 1536
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.30171331763267517,
      "eval_runtime": 126.0264,
      "eval_samples_per_second": 1.016,
      "eval_steps_per_second": 0.341,
      "step": 1536
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00018676114695723932,
      "loss": 0.4045,
      "step": 1537
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00018674419535640493,
      "loss": 0.2427,
      "step": 1538
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00018672723368006705,
      "loss": 0.2534,
      "step": 1539
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00018671026193019588,
      "loss": 0.1234,
      "step": 1540
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00018669328010876266,
      "loss": 0.3334,
      "step": 1541
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0001866762882177399,
      "loss": 0.1799,
      "step": 1542
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0001866592862591012,
      "loss": 0.2233,
      "step": 1543
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.00018664227423482142,
      "loss": 0.1853,
      "step": 1544
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0001866252521468765,
      "loss": 0.3073,
      "step": 1545
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0001866082199972436,
      "loss": 0.1835,
      "step": 1546
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.00018659117778790105,
      "loss": 0.2118,
      "step": 1547
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0001865741255208283,
      "loss": 0.1749,
      "step": 1548
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.00018655706319800606,
      "loss": 0.1897,
      "step": 1549
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.0001865399908214161,
      "loss": 0.2241,
      "step": 1550
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.0001865229083930414,
      "loss": 0.2073,
      "step": 1551
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.00018650581591486616,
      "loss": 0.248,
      "step": 1552
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.00018648871338887573,
      "loss": 0.1317,
      "step": 1553
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.00018647160081705652,
      "loss": 0.1367,
      "step": 1554
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.00018645447820139622,
      "loss": 0.3073,
      "step": 1555
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.00018643734554388367,
      "loss": 0.1581,
      "step": 1556
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.00018642020284650884,
      "loss": 0.1755,
      "step": 1557
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.00018640305011126293,
      "loss": 0.2127,
      "step": 1558
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0001863858873401382,
      "loss": 0.2084,
      "step": 1559
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.00018636871453512818,
      "loss": 0.2127,
      "step": 1560
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0001863515316982275,
      "loss": 0.2437,
      "step": 1561
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.00018633433883143204,
      "loss": 0.2642,
      "step": 1562
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0001863171359367387,
      "loss": 0.1924,
      "step": 1563
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.00018629992301614565,
      "loss": 0.2045,
      "step": 1564
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.00018628270007165222,
      "loss": 0.1442,
      "step": 1565
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.00018626546710525892,
      "loss": 0.2106,
      "step": 1566
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.00018624822411896736,
      "loss": 0.2154,
      "step": 1567
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.00018623097111478032,
      "loss": 0.184,
      "step": 1568
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0001862137080947018,
      "loss": 0.1698,
      "step": 1569
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.00018619643506073694,
      "loss": 0.1918,
      "step": 1570
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.000186179152014892,
      "loss": 0.247,
      "step": 1571
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.0001861618589591745,
      "loss": 0.1742,
      "step": 1572
    },
    {
      "epoch": 2.04,
      "learning_rate": 0.00018614455589559297,
      "loss": 0.2158,
      "step": 1573
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00018612724282615726,
      "loss": 0.1842,
      "step": 1574
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00018610991975287833,
      "loss": 0.1903,
      "step": 1575
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00018609258667776822,
      "loss": 0.2079,
      "step": 1576
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00018607524360284028,
      "loss": 0.1808,
      "step": 1577
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00018605789053010888,
      "loss": 0.2533,
      "step": 1578
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.00018604052746158965,
      "loss": 0.1759,
      "step": 1579
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0001860231543992993,
      "loss": 0.2405,
      "step": 1580
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00018600577134525578,
      "loss": 0.1832,
      "step": 1581
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00018598837830147816,
      "loss": 0.2517,
      "step": 1582
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00018597097526998671,
      "loss": 0.1962,
      "step": 1583
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.0001859535622528027,
      "loss": 0.1718,
      "step": 1584
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00018593613925194883,
      "loss": 0.2459,
      "step": 1585
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00018591870626944877,
      "loss": 0.2002,
      "step": 1586
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00018590126330732734,
      "loss": 0.1496,
      "step": 1587
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00018588381036761066,
      "loss": 0.1906,
      "step": 1588
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00018586634745232583,
      "loss": 0.1875,
      "step": 1589
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00018584887456350128,
      "loss": 0.1832,
      "step": 1590
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00018583139170316645,
      "loss": 0.114,
      "step": 1591
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00018581389887335206,
      "loss": 0.2391,
      "step": 1592
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00018579639607608993,
      "loss": 0.1527,
      "step": 1593
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00018577888331341302,
      "loss": 0.2136,
      "step": 1594
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00018576136058735547,
      "loss": 0.1953,
      "step": 1595
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00018574382789995263,
      "loss": 0.175,
      "step": 1596
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.00018572628525324089,
      "loss": 0.2264,
      "step": 1597
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0001857087326492579,
      "loss": 0.195,
      "step": 1598
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0001856911700900424,
      "loss": 0.1818,
      "step": 1599
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.00018567359757763434,
      "loss": 0.2553,
      "step": 1600
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.0001856560151140748,
      "loss": 0.1678,
      "step": 1601
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.00018563842270140598,
      "loss": 0.2156,
      "step": 1602
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.00018562082034167133,
      "loss": 0.2093,
      "step": 1603
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.00018560320803691536,
      "loss": 0.2245,
      "step": 1604
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.00018558558578918378,
      "loss": 0.1548,
      "step": 1605
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.00018556795360052345,
      "loss": 0.1417,
      "step": 1606
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.00018555031147298237,
      "loss": 0.196,
      "step": 1607
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.00018553265940860972,
      "loss": 0.2274,
      "step": 1608
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0001855149974094558,
      "loss": 0.2272,
      "step": 1609
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.00018549732547757213,
      "loss": 0.2089,
      "step": 1610
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.0001854796436150113,
      "loss": 0.2665,
      "step": 1611
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00018546195182382713,
      "loss": 0.2262,
      "step": 1612
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00018544425010607452,
      "loss": 0.2003,
      "step": 1613
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00018542653846380955,
      "loss": 0.2251,
      "step": 1614
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00018540881689908947,
      "loss": 0.2232,
      "step": 1615
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.0001853910854139727,
      "loss": 0.185,
      "step": 1616
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00018537334401051876,
      "loss": 0.1574,
      "step": 1617
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00018535559269078838,
      "loss": 0.3544,
      "step": 1618
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00018533783145684333,
      "loss": 0.142,
      "step": 1619
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0001853200603107467,
      "loss": 0.2075,
      "step": 1620
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00018530227925456263,
      "loss": 0.2051,
      "step": 1621
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.0001852844882903564,
      "loss": 0.213,
      "step": 1622
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00018526668742019443,
      "loss": 0.1631,
      "step": 1623
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00018524887664614437,
      "loss": 0.2869,
      "step": 1624
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00018523105597027497,
      "loss": 0.2442,
      "step": 1625
    },
    {
      "epoch": 2.11,
      "learning_rate": 0.00018521322539465614,
      "loss": 0.1539,
      "step": 1626
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00018519538492135892,
      "loss": 0.2605,
      "step": 1627
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00018517753455245552,
      "loss": 0.1869,
      "step": 1628
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00018515967429001928,
      "loss": 0.3312,
      "step": 1629
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00018514180413612476,
      "loss": 0.1993,
      "step": 1630
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00018512392409284755,
      "loss": 0.1681,
      "step": 1631
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00018510603416226447,
      "loss": 0.1727,
      "step": 1632
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00018508813434645348,
      "loss": 0.1439,
      "step": 1633
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00018507022464749368,
      "loss": 0.2009,
      "step": 1634
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.0001850523050674653,
      "loss": 0.2099,
      "step": 1635
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00018503437560844974,
      "loss": 0.1971,
      "step": 1636
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.0001850164362725295,
      "loss": 0.2468,
      "step": 1637
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00018499848706178838,
      "loss": 0.2148,
      "step": 1638
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00018498052797831112,
      "loss": 0.1785,
      "step": 1639
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00018496255902418368,
      "loss": 0.1991,
      "step": 1640
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.0001849445802014933,
      "loss": 0.2498,
      "step": 1641
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00018492659151232813,
      "loss": 0.1991,
      "step": 1642
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00018490859295877766,
      "loss": 0.2326,
      "step": 1643
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00018489058454293248,
      "loss": 0.2403,
      "step": 1644
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00018487256626688426,
      "loss": 0.1884,
      "step": 1645
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00018485453813272583,
      "loss": 0.2443,
      "step": 1646
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00018483650014255125,
      "loss": 0.2203,
      "step": 1647
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00018481845229845564,
      "loss": 0.1742,
      "step": 1648
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.0001848003946025353,
      "loss": 0.1088,
      "step": 1649
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00018478232705688766,
      "loss": 0.1367,
      "step": 1650
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.00018476424966361129,
      "loss": 0.2282,
      "step": 1651
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0001847461624248059,
      "loss": 0.2517,
      "step": 1652
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.00018472806534257241,
      "loss": 0.2158,
      "step": 1653
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0001847099584190128,
      "loss": 0.225,
      "step": 1654
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0001846918416562302,
      "loss": 0.1728,
      "step": 1655
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.00018467371505632895,
      "loss": 0.173,
      "step": 1656
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.0001846555786214145,
      "loss": 0.1927,
      "step": 1657
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00018463743235359336,
      "loss": 0.2207,
      "step": 1658
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.0001846192762549733,
      "loss": 0.2068,
      "step": 1659
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00018460111032766321,
      "loss": 0.2096,
      "step": 1660
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00018458293457377303,
      "loss": 0.2498,
      "step": 1661
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00018456474899541397,
      "loss": 0.1832,
      "step": 1662
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00018454655359469827,
      "loss": 0.2119,
      "step": 1663
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00018452834837373944,
      "loss": 0.1874,
      "step": 1664
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00018451013333465195,
      "loss": 0.1257,
      "step": 1665
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.00018449190847955155,
      "loss": 0.2162,
      "step": 1666
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.00018447367381055514,
      "loss": 0.1814,
      "step": 1667
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.00018445542932978067,
      "loss": 0.1905,
      "step": 1668
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.00018443717503934724,
      "loss": 0.1778,
      "step": 1669
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.00018441891094137514,
      "loss": 0.1585,
      "step": 1670
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.00018440063703798588,
      "loss": 0.2189,
      "step": 1671
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.00018438235333130182,
      "loss": 0.1793,
      "step": 1672
    },
    {
      "epoch": 2.17,
      "learning_rate": 0.0001843640598234468,
      "loss": 0.1189,
      "step": 1673
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.00018434575651654563,
      "loss": 0.1894,
      "step": 1674
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.00018432744341272418,
      "loss": 0.2054,
      "step": 1675
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.0001843091205141096,
      "loss": 0.1508,
      "step": 1676
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.0001842907878228302,
      "loss": 0.2787,
      "step": 1677
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.00018427244534101526,
      "loss": 0.2339,
      "step": 1678
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.00018425409307079536,
      "loss": 0.2128,
      "step": 1679
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.00018423573101430208,
      "loss": 0.246,
      "step": 1680
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.00018421735917366827,
      "loss": 0.194,
      "step": 1681
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.00018419897755102786,
      "loss": 0.2445,
      "step": 1682
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.00018418058614851586,
      "loss": 0.1869,
      "step": 1683
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0001841621849682685,
      "loss": 0.2737,
      "step": 1684
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.00018414377401242309,
      "loss": 0.2893,
      "step": 1685
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.00018412535328311814,
      "loss": 0.214,
      "step": 1686
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.0001841069227824932,
      "loss": 0.1567,
      "step": 1687
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.00018408848251268904,
      "loss": 0.1519,
      "step": 1688
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0001840700324758475,
      "loss": 0.1939,
      "step": 1689
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00018405157267411165,
      "loss": 0.3227,
      "step": 1690
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00018403310310962558,
      "loss": 0.1531,
      "step": 1691
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00018401462378453458,
      "loss": 0.1496,
      "step": 1692
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00018399613470098506,
      "loss": 0.1725,
      "step": 1693
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00018397763586112454,
      "loss": 0.1432,
      "step": 1694
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00018395912726710173,
      "loss": 0.2055,
      "step": 1695
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.0001839406089210664,
      "loss": 0.3052,
      "step": 1696
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.00018392208082516955,
      "loss": 0.2126,
      "step": 1697
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0001839035429815632,
      "loss": 0.2005,
      "step": 1698
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.00018388499539240057,
      "loss": 0.2297,
      "step": 1699
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.00018386643805983602,
      "loss": 0.2293,
      "step": 1700
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.00018384787098602495,
      "loss": 0.1635,
      "step": 1701
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.00018382929417312406,
      "loss": 0.2389,
      "step": 1702
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.00018381070762329103,
      "loss": 0.3057,
      "step": 1703
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00018379211133868467,
      "loss": 0.2323,
      "step": 1704
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0001837735053214651,
      "loss": 0.1949,
      "step": 1705
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00018375488957379337,
      "loss": 0.1149,
      "step": 1706
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0001837362640978317,
      "loss": 0.2569,
      "step": 1707
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00018371762889574352,
      "loss": 0.1956,
      "step": 1708
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.0001836989839696934,
      "loss": 0.2374,
      "step": 1709
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00018368032932184684,
      "loss": 0.1851,
      "step": 1710
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00018366166495437073,
      "loss": 0.2362,
      "step": 1711
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.00018364299086943295,
      "loss": 0.1565,
      "step": 1712
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0001836243070692025,
      "loss": 0.1515,
      "step": 1713
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.0001836056135558496,
      "loss": 0.1774,
      "step": 1714
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.00018358691033154543,
      "loss": 0.2357,
      "step": 1715
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.00018356819739846246,
      "loss": 0.2099,
      "step": 1716
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.00018354947475877431,
      "loss": 0.2094,
      "step": 1717
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.00018353074241465552,
      "loss": 0.2436,
      "step": 1718
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.00018351200036828197,
      "loss": 0.253,
      "step": 1719
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00018349324862183057,
      "loss": 0.109,
      "step": 1720
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00018347448717747936,
      "loss": 0.2923,
      "step": 1721
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00018345571603740752,
      "loss": 0.1554,
      "step": 1722
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00018343693520379536,
      "loss": 0.1754,
      "step": 1723
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0001834181446788243,
      "loss": 0.3069,
      "step": 1724
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.0001833993444646769,
      "loss": 0.1148,
      "step": 1725
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00018338053456353688,
      "loss": 0.2048,
      "step": 1726
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00018336171497758893,
      "loss": 0.2549,
      "step": 1727
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00018334288570901911,
      "loss": 0.1548,
      "step": 1728
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00018332404676001443,
      "loss": 0.2534,
      "step": 1729
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00018330519813276305,
      "loss": 0.1079,
      "step": 1730
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00018328633982945427,
      "loss": 0.1764,
      "step": 1731
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00018326747185227858,
      "loss": 0.155,
      "step": 1732
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00018324859420342743,
      "loss": 0.1984,
      "step": 1733
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00018322970688509357,
      "loss": 0.1837,
      "step": 1734
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0001832108098994708,
      "loss": 0.255,
      "step": 1735
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.000183191903248754,
      "loss": 0.1547,
      "step": 1736
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.00018317298693513922,
      "loss": 0.2091,
      "step": 1737
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.00018315406096082367,
      "loss": 0.1516,
      "step": 1738
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.00018313512532800558,
      "loss": 0.214,
      "step": 1739
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.0001831161800388844,
      "loss": 0.1774,
      "step": 1740
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.00018309722509566066,
      "loss": 0.2001,
      "step": 1741
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.00018307826050053598,
      "loss": 0.1877,
      "step": 1742
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.00018305928625571315,
      "loss": 0.2431,
      "step": 1743
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.00018304030236339608,
      "loss": 0.1703,
      "step": 1744
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.0001830213088257898,
      "loss": 0.2493,
      "step": 1745
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.00018300230564510038,
      "loss": 0.179,
      "step": 1746
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.00018298329282353515,
      "loss": 0.2308,
      "step": 1747
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.00018296427036330242,
      "loss": 0.2079,
      "step": 1748
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.00018294523826661174,
      "loss": 0.2389,
      "step": 1749
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.00018292619653567367,
      "loss": 0.1804,
      "step": 1750
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.00018290714517270004,
      "loss": 0.1826,
      "step": 1751
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.00018288808417990358,
      "loss": 0.3082,
      "step": 1752
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.00018286901355949833,
      "loss": 0.2244,
      "step": 1753
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.00018284993331369934,
      "loss": 0.1951,
      "step": 1754
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.00018283084344472287,
      "loss": 0.3064,
      "step": 1755
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.0001828117439547862,
      "loss": 0.2476,
      "step": 1756
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.00018279263484610782,
      "loss": 0.1869,
      "step": 1757
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.00018277351612090725,
      "loss": 0.2664,
      "step": 1758
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.00018275438778140516,
      "loss": 0.1358,
      "step": 1759
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.00018273524982982336,
      "loss": 0.192,
      "step": 1760
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.00018271610226838476,
      "loss": 0.2055,
      "step": 1761
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.00018269694509931336,
      "loss": 0.1595,
      "step": 1762
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.00018267777832483436,
      "loss": 0.2304,
      "step": 1763
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.00018265860194717398,
      "loss": 0.2907,
      "step": 1764
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.00018263941596855958,
      "loss": 0.1074,
      "step": 1765
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0001826202203912197,
      "loss": 0.1579,
      "step": 1766
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.00018260101521738384,
      "loss": 0.2168,
      "step": 1767
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.00018258180044928283,
      "loss": 0.2133,
      "step": 1768
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0001825625760891484,
      "loss": 0.2551,
      "step": 1769
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.00018254334213921362,
      "loss": 0.1861,
      "step": 1770
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.00018252409860171245,
      "loss": 0.2485,
      "step": 1771
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0001825048454788801,
      "loss": 0.1539,
      "step": 1772
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.00018248558277295286,
      "loss": 0.2283,
      "step": 1773
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0001824663104861681,
      "loss": 0.1856,
      "step": 1774
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.00018244702862076437,
      "loss": 0.1655,
      "step": 1775
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0001824277371789813,
      "loss": 0.1568,
      "step": 1776
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0001824084361630596,
      "loss": 0.1965,
      "step": 1777
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.00018238912557524107,
      "loss": 0.1539,
      "step": 1778
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0001823698054177688,
      "loss": 0.1512,
      "step": 1779
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.00018235047569288673,
      "loss": 0.1847,
      "step": 1780
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.00018233113640284015,
      "loss": 0.2177,
      "step": 1781
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0001823117875498753,
      "loss": 0.31,
      "step": 1782
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0001822924291362396,
      "loss": 0.2443,
      "step": 1783
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.00018227306116418154,
      "loss": 0.2303,
      "step": 1784
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.00018225368363595078,
      "loss": 0.2963,
      "step": 1785
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.00018223429655379807,
      "loss": 0.1226,
      "step": 1786
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.00018221489991997522,
      "loss": 0.2136,
      "step": 1787
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.0001821954937367352,
      "loss": 0.2258,
      "step": 1788
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0001821760780063321,
      "loss": 0.205,
      "step": 1789
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.00018215665273102103,
      "loss": 0.169,
      "step": 1790
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.00018213721791305833,
      "loss": 0.1537,
      "step": 1791
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.00018211777355470136,
      "loss": 0.2556,
      "step": 1792
    },
    {
      "epoch": 2.33,
      "eval_loss": 0.2978043258190155,
      "eval_runtime": 126.0528,
      "eval_samples_per_second": 1.015,
      "eval_steps_per_second": 0.341,
      "step": 1792
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.00018209831965820868,
      "loss": 0.1731,
      "step": 1793
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0001820788562258398,
      "loss": 0.1776,
      "step": 1794
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.00018205938325985548,
      "loss": 0.2769,
      "step": 1795
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.00018203990076251758,
      "loss": 0.2337,
      "step": 1796
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.00018202040873608897,
      "loss": 0.2645,
      "step": 1797
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.00018200090718283374,
      "loss": 0.1306,
      "step": 1798
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.00018198139610501696,
      "loss": 0.1442,
      "step": 1799
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.00018196187550490497,
      "loss": 0.2014,
      "step": 1800
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.00018194234538476507,
      "loss": 0.1715,
      "step": 1801
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.00018192280574686568,
      "loss": 0.1736,
      "step": 1802
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.00018190325659347647,
      "loss": 0.2138,
      "step": 1803
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00018188369792686803,
      "loss": 0.1184,
      "step": 1804
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00018186412974931218,
      "loss": 0.2659,
      "step": 1805
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00018184455206308176,
      "loss": 0.2465,
      "step": 1806
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.0001818249648704508,
      "loss": 0.1994,
      "step": 1807
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00018180536817369433,
      "loss": 0.2462,
      "step": 1808
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00018178576197508864,
      "loss": 0.1849,
      "step": 1809
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00018176614627691092,
      "loss": 0.2728,
      "step": 1810
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00018174652108143965,
      "loss": 0.2842,
      "step": 1811
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0001817268863909543,
      "loss": 0.1774,
      "step": 1812
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00018170724220773548,
      "loss": 0.2536,
      "step": 1813
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0001816875885340649,
      "loss": 0.2398,
      "step": 1814
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00018166792537222537,
      "loss": 0.1743,
      "step": 1815
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0001816482527245008,
      "loss": 0.2857,
      "step": 1816
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0001816285705931762,
      "loss": 0.1834,
      "step": 1817
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.0001816088789805377,
      "loss": 0.1102,
      "step": 1818
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00018158917788887254,
      "loss": 0.2211,
      "step": 1819
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.000181569467320469,
      "loss": 0.2596,
      "step": 1820
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0001815497472776165,
      "loss": 0.1578,
      "step": 1821
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.0001815300177626056,
      "loss": 0.1897,
      "step": 1822
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.00018151027877772791,
      "loss": 0.2598,
      "step": 1823
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.00018149053032527612,
      "loss": 0.1365,
      "step": 1824
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.00018147077240754408,
      "loss": 0.2498,
      "step": 1825
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.00018145100502682667,
      "loss": 0.131,
      "step": 1826
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00018143122818541996,
      "loss": 0.1627,
      "step": 1827
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00018141144188562103,
      "loss": 0.1777,
      "step": 1828
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00018139164612972813,
      "loss": 0.2598,
      "step": 1829
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00018137184092004055,
      "loss": 0.2107,
      "step": 1830
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.0001813520262588587,
      "loss": 0.1415,
      "step": 1831
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00018133220214848407,
      "loss": 0.1971,
      "step": 1832
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00018131236859121933,
      "loss": 0.2707,
      "step": 1833
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00018129252558936817,
      "loss": 0.2415,
      "step": 1834
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.00018127267314523536,
      "loss": 0.2554,
      "step": 1835
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0001812528112611268,
      "loss": 0.2513,
      "step": 1836
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0001812329399393495,
      "loss": 0.279,
      "step": 1837
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.00018121305918221157,
      "loss": 0.2353,
      "step": 1838
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.00018119316899202216,
      "loss": 0.1953,
      "step": 1839
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0001811732693710916,
      "loss": 0.1805,
      "step": 1840
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0001811533603217312,
      "loss": 0.2581,
      "step": 1841
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.0001811334418462535,
      "loss": 0.257,
      "step": 1842
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00018111351394697206,
      "loss": 0.203,
      "step": 1843
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.0001810935766262015,
      "loss": 0.1835,
      "step": 1844
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.0001810736298862576,
      "loss": 0.1881,
      "step": 1845
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00018105367372945726,
      "loss": 0.3353,
      "step": 1846
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00018103370815811834,
      "loss": 0.2238,
      "step": 1847
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00018101373317455996,
      "loss": 0.1838,
      "step": 1848
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00018099374878110218,
      "loss": 0.217,
      "step": 1849
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.00018097375498006627,
      "loss": 0.2876,
      "step": 1850
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.00018095375177377455,
      "loss": 0.2539,
      "step": 1851
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.00018093373916455043,
      "loss": 0.245,
      "step": 1852
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.0001809137171547184,
      "loss": 0.1553,
      "step": 1853
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.000180893685746604,
      "loss": 0.2691,
      "step": 1854
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.00018087364494253402,
      "loss": 0.2159,
      "step": 1855
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.0001808535947448362,
      "loss": 0.2365,
      "step": 1856
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.00018083353515583936,
      "loss": 0.231,
      "step": 1857
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00018081346617787353,
      "loss": 0.3324,
      "step": 1858
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.0001807933878132697,
      "loss": 0.1902,
      "step": 1859
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00018077330006436006,
      "loss": 0.2294,
      "step": 1860
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00018075320293347785,
      "loss": 0.223,
      "step": 1861
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00018073309642295727,
      "loss": 0.2148,
      "step": 1862
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00018071298053513388,
      "loss": 0.1496,
      "step": 1863
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00018069285527234411,
      "loss": 0.2058,
      "step": 1864
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00018067272063692556,
      "loss": 0.261,
      "step": 1865
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.00018065257663121687,
      "loss": 0.2195,
      "step": 1866
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.00018063242325755787,
      "loss": 0.2069,
      "step": 1867
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.00018061226051828934,
      "loss": 0.1952,
      "step": 1868
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0001805920884157533,
      "loss": 0.1546,
      "step": 1869
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.0001805719069522927,
      "loss": 0.2045,
      "step": 1870
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.00018055171613025174,
      "loss": 0.2794,
      "step": 1871
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.00018053151595197553,
      "loss": 0.1438,
      "step": 1872
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.00018051130641981043,
      "loss": 0.1942,
      "step": 1873
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.00018049108753610376,
      "loss": 0.2217,
      "step": 1874
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.00018047085930320402,
      "loss": 0.2013,
      "step": 1875
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.00018045062172346077,
      "loss": 0.2259,
      "step": 1876
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0001804303747992246,
      "loss": 0.2015,
      "step": 1877
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.00018041011853284726,
      "loss": 0.2322,
      "step": 1878
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.00018038985292668152,
      "loss": 0.2386,
      "step": 1879
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.00018036957798308134,
      "loss": 0.1981,
      "step": 1880
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00018034929370440162,
      "loss": 0.1149,
      "step": 1881
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00018032900009299846,
      "loss": 0.2586,
      "step": 1882
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00018030869715122895,
      "loss": 0.1959,
      "step": 1883
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00018028838488145138,
      "loss": 0.1697,
      "step": 1884
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00018026806328602502,
      "loss": 0.2643,
      "step": 1885
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.0001802477323673103,
      "loss": 0.1416,
      "step": 1886
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00018022739212766861,
      "loss": 0.1998,
      "step": 1887
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00018020704256946265,
      "loss": 0.1216,
      "step": 1888
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0001801866836950559,
      "loss": 0.1875,
      "step": 1889
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0001801663155068132,
      "loss": 0.2273,
      "step": 1890
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.00018014593800710026,
      "loss": 0.1715,
      "step": 1891
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.00018012555119828404,
      "loss": 0.1523,
      "step": 1892
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0001801051550827325,
      "loss": 0.2986,
      "step": 1893
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.00018008474966281466,
      "loss": 0.1799,
      "step": 1894
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.00018006433494090066,
      "loss": 0.1538,
      "step": 1895
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.00018004391091936168,
      "loss": 0.2668,
      "step": 1896
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0001800234776005701,
      "loss": 0.2841,
      "step": 1897
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.00018000303498689918,
      "loss": 0.2697,
      "step": 1898
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0001799825830807234,
      "loss": 0.3179,
      "step": 1899
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.0001799621218844183,
      "loss": 0.2528,
      "step": 1900
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.00017994165140036055,
      "loss": 0.2547,
      "step": 1901
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.00017992117163092772,
      "loss": 0.2736,
      "step": 1902
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.00017990068257849864,
      "loss": 0.1665,
      "step": 1903
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.00017988018424545314,
      "loss": 0.1677,
      "step": 1904
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.00017985967663417215,
      "loss": 0.2202,
      "step": 1905
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.00017983915974703764,
      "loss": 0.1926,
      "step": 1906
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.00017981863358643274,
      "loss": 0.1309,
      "step": 1907
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.00017979809815474153,
      "loss": 0.2037,
      "step": 1908
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0001797775534543493,
      "loss": 0.1317,
      "step": 1909
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.00017975699948764233,
      "loss": 0.1783,
      "step": 1910
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.000179736436257008,
      "loss": 0.1985,
      "step": 1911
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00017971586376483477,
      "loss": 0.1585,
      "step": 1912
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00017969528201351219,
      "loss": 0.2064,
      "step": 1913
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00017967469100543085,
      "loss": 0.2569,
      "step": 1914
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00017965409074298245,
      "loss": 0.2017,
      "step": 1915
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00017963348122855975,
      "loss": 0.2093,
      "step": 1916
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00017961286246455656,
      "loss": 0.2214,
      "step": 1917
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00017959223445336778,
      "loss": 0.2013,
      "step": 1918
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00017957159719738945,
      "loss": 0.2482,
      "step": 1919
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.0001795509506990186,
      "loss": 0.2725,
      "step": 1920
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.00017953029496065335,
      "loss": 0.1986,
      "step": 1921
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.0001795096299846929,
      "loss": 0.2836,
      "step": 1922
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.00017948895577353752,
      "loss": 0.1762,
      "step": 1923
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.00017946827232958858,
      "loss": 0.2074,
      "step": 1924
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.00017944757965524853,
      "loss": 0.2635,
      "step": 1925
    },
    {
      "epoch": 2.5,
      "learning_rate": 0.0001794268777529208,
      "loss": 0.1095,
      "step": 1926
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00017940616662500997,
      "loss": 0.1585,
      "step": 1927
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00017938544627392174,
      "loss": 0.1445,
      "step": 1928
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00017936471670206271,
      "loss": 0.1487,
      "step": 1929
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00017934397791184075,
      "loss": 0.1718,
      "step": 1930
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0001793232299056647,
      "loss": 0.1983,
      "step": 1931
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.0001793024726859444,
      "loss": 0.2173,
      "step": 1932
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00017928170625509094,
      "loss": 0.1526,
      "step": 1933
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00017926093061551633,
      "loss": 0.2006,
      "step": 1934
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0001792401457696337,
      "loss": 0.1876,
      "step": 1935
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.00017921935171985725,
      "loss": 0.2101,
      "step": 1936
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.00017919854846860228,
      "loss": 0.1989,
      "step": 1937
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0001791777360182851,
      "loss": 0.1382,
      "step": 1938
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.00017915691437132313,
      "loss": 0.2648,
      "step": 1939
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0001791360835301348,
      "loss": 0.1615,
      "step": 1940
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.0001791152434971397,
      "loss": 0.2802,
      "step": 1941
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.00017909439427475846,
      "loss": 0.133,
      "step": 1942
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0001790735358654127,
      "loss": 0.2264,
      "step": 1943
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0001790526682715252,
      "loss": 0.1874,
      "step": 1944
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0001790317914955197,
      "loss": 0.184,
      "step": 1945
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0001790109055398212,
      "loss": 0.1595,
      "step": 1946
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.00017899001040685555,
      "loss": 0.1823,
      "step": 1947
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.00017896910609904978,
      "loss": 0.2416,
      "step": 1948
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.00017894819261883197,
      "loss": 0.2473,
      "step": 1949
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.0001789272699686313,
      "loss": 0.2587,
      "step": 1950
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00017890633815087793,
      "loss": 0.1501,
      "step": 1951
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00017888539716800312,
      "loss": 0.3171,
      "step": 1952
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.0001788644470224392,
      "loss": 0.3347,
      "step": 1953
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00017884348771661962,
      "loss": 0.371,
      "step": 1954
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00017882251925297883,
      "loss": 0.1413,
      "step": 1955
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00017880154163395234,
      "loss": 0.2189,
      "step": 1956
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.00017878055486197675,
      "loss": 0.3184,
      "step": 1957
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0001787595589394897,
      "loss": 0.2856,
      "step": 1958
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0001787385538689299,
      "loss": 0.1721,
      "step": 1959
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.00017871753965273716,
      "loss": 0.2154,
      "step": 1960
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0001786965162933523,
      "loss": 0.145,
      "step": 1961
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.00017867548379321723,
      "loss": 0.2302,
      "step": 1962
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0001786544421547749,
      "loss": 0.2575,
      "step": 1963
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.00017863339138046938,
      "loss": 0.2503,
      "step": 1964
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.00017861233147274569,
      "loss": 0.1718,
      "step": 1965
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.00017859126243405005,
      "loss": 0.2619,
      "step": 1966
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.0001785701842668296,
      "loss": 0.1512,
      "step": 1967
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.00017854909697353265,
      "loss": 0.1957,
      "step": 1968
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.00017852800055660853,
      "loss": 0.2477,
      "step": 1969
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.00017850689501850765,
      "loss": 0.2151,
      "step": 1970
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.00017848578036168137,
      "loss": 0.2089,
      "step": 1971
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.0001784646565885823,
      "loss": 0.2333,
      "step": 1972
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.00017844352370166396,
      "loss": 0.143,
      "step": 1973
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00017842238170338095,
      "loss": 0.1975,
      "step": 1974
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00017840123059618902,
      "loss": 0.1595,
      "step": 1975
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00017838007038254485,
      "loss": 0.1623,
      "step": 1976
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00017835890106490627,
      "loss": 0.1674,
      "step": 1977
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.0001783377226457321,
      "loss": 0.2307,
      "step": 1978
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00017831653512748234,
      "loss": 0.2007,
      "step": 1979
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.0001782953385126179,
      "loss": 0.2285,
      "step": 1980
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00017827413280360077,
      "loss": 0.2971,
      "step": 1981
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.0001782529180028941,
      "loss": 0.1728,
      "step": 1982
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00017823169411296198,
      "loss": 0.1637,
      "step": 1983
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00017821046113626965,
      "loss": 0.17,
      "step": 1984
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00017818921907528334,
      "loss": 0.2228,
      "step": 1985
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00017816796793247036,
      "loss": 0.164,
      "step": 1986
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00017814670771029904,
      "loss": 0.1329,
      "step": 1987
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.00017812543841123882,
      "loss": 0.1583,
      "step": 1988
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0001781041600377602,
      "loss": 0.2057,
      "step": 1989
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.00017808287259233466,
      "loss": 0.1103,
      "step": 1990
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0001780615760774348,
      "loss": 0.2168,
      "step": 1991
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.00017804027049553425,
      "loss": 0.2324,
      "step": 1992
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.00017801895584910766,
      "loss": 0.1858,
      "step": 1993
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0001779976321406308,
      "loss": 0.1564,
      "step": 1994
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.00017797629937258046,
      "loss": 0.1393,
      "step": 1995
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.0001779549575474345,
      "loss": 0.252,
      "step": 1996
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00017793360666767174,
      "loss": 0.249,
      "step": 1997
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.0001779122467357722,
      "loss": 0.165,
      "step": 1998
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00017789087775421685,
      "loss": 0.1776,
      "step": 1999
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00017786949972548773,
      "loss": 0.1032,
      "step": 2000
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00017784811265206797,
      "loss": 0.2124,
      "step": 2001
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00017782671653644169,
      "loss": 0.1608,
      "step": 2002
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00017780531138109407,
      "loss": 0.2037,
      "step": 2003
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.00017778389718851138,
      "loss": 0.396,
      "step": 2004
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.00017776247396118094,
      "loss": 0.3557,
      "step": 2005
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.0001777410417015911,
      "loss": 0.2828,
      "step": 2006
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.00017771960041223123,
      "loss": 0.1668,
      "step": 2007
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.00017769815009559178,
      "loss": 0.2073,
      "step": 2008
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.00017767669075416427,
      "loss": 0.2035,
      "step": 2009
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.00017765522239044122,
      "loss": 0.1915,
      "step": 2010
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.0001776337450069162,
      "loss": 0.2128,
      "step": 2011
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.0001776122586060839,
      "loss": 0.1459,
      "step": 2012
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.00017759076319043998,
      "loss": 0.1584,
      "step": 2013
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.00017756925876248118,
      "loss": 0.2191,
      "step": 2014
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.00017754774532470526,
      "loss": 0.1926,
      "step": 2015
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.00017752622287961107,
      "loss": 0.1782,
      "step": 2016
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.0001775046914296985,
      "loss": 0.1705,
      "step": 2017
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.0001774831509774684,
      "loss": 0.1786,
      "step": 2018
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.00017746160152542283,
      "loss": 0.1697,
      "step": 2019
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.00017744004307606472,
      "loss": 0.2698,
      "step": 2020
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.00017741847563189814,
      "loss": 0.2492,
      "step": 2021
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.00017739689919542823,
      "loss": 0.2277,
      "step": 2022
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.0001773753137691611,
      "loss": 0.1194,
      "step": 2023
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.00017735371935560394,
      "loss": 0.1916,
      "step": 2024
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.000177332115957265,
      "loss": 0.2522,
      "step": 2025
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.00017731050357665354,
      "loss": 0.1384,
      "step": 2026
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.00017728888221627987,
      "loss": 0.0788,
      "step": 2027
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0001772672518786554,
      "loss": 0.1951,
      "step": 2028
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.00017724561256629248,
      "loss": 0.2268,
      "step": 2029
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.00017722396428170457,
      "loss": 0.1674,
      "step": 2030
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0001772023070274062,
      "loss": 0.2005,
      "step": 2031
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.00017718064080591284,
      "loss": 0.2412,
      "step": 2032
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0001771589656197411,
      "loss": 0.2005,
      "step": 2033
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.00017713728147140858,
      "loss": 0.1724,
      "step": 2034
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00017711558836343394,
      "loss": 0.1319,
      "step": 2035
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.0001770938862983369,
      "loss": 0.2192,
      "step": 2036
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00017707217527863814,
      "loss": 0.1772,
      "step": 2037
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00017705045530685948,
      "loss": 0.1366,
      "step": 2038
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00017702872638552375,
      "loss": 0.2691,
      "step": 2039
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00017700698851715475,
      "loss": 0.204,
      "step": 2040
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00017698524170427744,
      "loss": 0.2358,
      "step": 2041
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.00017696348594941767,
      "loss": 0.2283,
      "step": 2042
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0001769417212551025,
      "loss": 0.1942,
      "step": 2043
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0001769199476238599,
      "loss": 0.2006,
      "step": 2044
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00017689816505821892,
      "loss": 0.2855,
      "step": 2045
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00017687637356070965,
      "loss": 0.2071,
      "step": 2046
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.0001768545731338632,
      "loss": 0.2588,
      "step": 2047
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00017683276378021174,
      "loss": 0.271,
      "step": 2048
    },
    {
      "epoch": 2.66,
      "eval_loss": 0.2892937958240509,
      "eval_runtime": 126.0286,
      "eval_samples_per_second": 1.016,
      "eval_steps_per_second": 0.341,
      "step": 2048
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00017681094550228848,
      "loss": 0.2261,
      "step": 2049
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00017678911830262763,
      "loss": 0.1987,
      "step": 2050
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00017676728218376448,
      "loss": 0.1919,
      "step": 2051
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00017674543714823535,
      "loss": 0.1234,
      "step": 2052
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00017672358319857756,
      "loss": 0.2166,
      "step": 2053
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00017670172033732947,
      "loss": 0.1756,
      "step": 2054
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00017667984856703056,
      "loss": 0.2801,
      "step": 2055
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0001766579678902212,
      "loss": 0.248,
      "step": 2056
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.00017663607830944292,
      "loss": 0.1942,
      "step": 2057
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00017661417982723824,
      "loss": 0.2191,
      "step": 2058
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00017659227244615067,
      "loss": 0.263,
      "step": 2059
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00017657035616872483,
      "loss": 0.2926,
      "step": 2060
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0001765484309975063,
      "loss": 0.1678,
      "step": 2061
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00017652649693504176,
      "loss": 0.2045,
      "step": 2062
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00017650455398387891,
      "loss": 0.2263,
      "step": 2063
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0001764826021465664,
      "loss": 0.191,
      "step": 2064
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00017646064142565407,
      "loss": 0.1409,
      "step": 2065
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00017643867182369266,
      "loss": 0.2766,
      "step": 2066
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00017641669334323394,
      "loss": 0.1524,
      "step": 2067
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00017639470598683076,
      "loss": 0.2093,
      "step": 2068
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00017637270975703708,
      "loss": 0.1775,
      "step": 2069
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0001763507046564077,
      "loss": 0.2091,
      "step": 2070
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0001763286906874986,
      "loss": 0.1971,
      "step": 2071
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00017630666785286675,
      "loss": 0.2471,
      "step": 2072
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.0001762846361550701,
      "loss": 0.2106,
      "step": 2073
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.00017626259559666777,
      "loss": 0.2559,
      "step": 2074
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.0001762405461802197,
      "loss": 0.1857,
      "step": 2075
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.00017621848790828705,
      "loss": 0.1556,
      "step": 2076
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.0001761964207834319,
      "loss": 0.1652,
      "step": 2077
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.0001761743448082174,
      "loss": 0.1897,
      "step": 2078
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.00017615225998520768,
      "loss": 0.2233,
      "step": 2079
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.00017613016631696797,
      "loss": 0.1655,
      "step": 2080
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0001761080638060645,
      "loss": 0.2169,
      "step": 2081
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0001760859524550645,
      "loss": 0.2573,
      "step": 2082
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.00017606383226653624,
      "loss": 0.2041,
      "step": 2083
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.00017604170324304902,
      "loss": 0.2545,
      "step": 2084
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.0001760195653871732,
      "loss": 0.1869,
      "step": 2085
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.00017599741870148012,
      "loss": 0.1274,
      "step": 2086
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.00017597526318854212,
      "loss": 0.1502,
      "step": 2087
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.00017595309885093266,
      "loss": 0.1991,
      "step": 2088
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.00017593092569122612,
      "loss": 0.2035,
      "step": 2089
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.00017590874371199803,
      "loss": 0.1691,
      "step": 2090
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.0001758865529158248,
      "loss": 0.1921,
      "step": 2091
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.00017586435330528395,
      "loss": 0.2289,
      "step": 2092
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.000175842144882954,
      "loss": 0.2726,
      "step": 2093
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.00017581992765141452,
      "loss": 0.2778,
      "step": 2094
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.00017579770161324607,
      "loss": 0.195,
      "step": 2095
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.00017577546677103025,
      "loss": 0.2027,
      "step": 2096
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00017575322312734972,
      "loss": 0.1797,
      "step": 2097
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00017573097068478806,
      "loss": 0.2266,
      "step": 2098
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00017570870944592993,
      "loss": 0.2276,
      "step": 2099
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00017568643941336107,
      "loss": 0.2023,
      "step": 2100
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00017566416058966815,
      "loss": 0.2261,
      "step": 2101
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.0001756418729774389,
      "loss": 0.2037,
      "step": 2102
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00017561957657926208,
      "loss": 0.1994,
      "step": 2103
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00017559727139772747,
      "loss": 0.1561,
      "step": 2104
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00017557495743542585,
      "loss": 0.132,
      "step": 2105
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00017555263469494898,
      "loss": 0.1897,
      "step": 2106
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00017553030317888978,
      "loss": 0.1894,
      "step": 2107
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.000175507962889842,
      "loss": 0.1734,
      "step": 2108
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.0001754856138304006,
      "loss": 0.2016,
      "step": 2109
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.0001754632560031614,
      "loss": 0.1596,
      "step": 2110
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00017544088941072136,
      "loss": 0.2621,
      "step": 2111
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00017541851405567833,
      "loss": 0.1874,
      "step": 2112
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.0001753961299406313,
      "loss": 0.1477,
      "step": 2113
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00017537373706818022,
      "loss": 0.1244,
      "step": 2114
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00017535133544092606,
      "loss": 0.1275,
      "step": 2115
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.0001753289250614708,
      "loss": 0.1416,
      "step": 2116
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00017530650593241748,
      "loss": 0.1973,
      "step": 2117
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.0001752840780563701,
      "loss": 0.3105,
      "step": 2118
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00017526164143593368,
      "loss": 0.2674,
      "step": 2119
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00017523919607371436,
      "loss": 0.1889,
      "step": 2120
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.0001752167419723191,
      "loss": 0.2015,
      "step": 2121
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00017519427913435605,
      "loss": 0.1657,
      "step": 2122
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00017517180756243433,
      "loss": 0.131,
      "step": 2123
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00017514932725916404,
      "loss": 0.1973,
      "step": 2124
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00017512683822715625,
      "loss": 0.1405,
      "step": 2125
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00017510434046902318,
      "loss": 0.1522,
      "step": 2126
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.00017508183398737796,
      "loss": 0.1548,
      "step": 2127
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.00017505931878483474,
      "loss": 0.2502,
      "step": 2128
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.00017503679486400876,
      "loss": 0.2235,
      "step": 2129
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.00017501426222751615,
      "loss": 0.1299,
      "step": 2130
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.00017499172087797414,
      "loss": 0.2065,
      "step": 2131
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.000174969170818001,
      "loss": 0.1202,
      "step": 2132
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.0001749466120502159,
      "loss": 0.2818,
      "step": 2133
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.0001749240445772391,
      "loss": 0.1546,
      "step": 2134
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00017490146840169185,
      "loss": 0.271,
      "step": 2135
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00017487888352619646,
      "loss": 0.2052,
      "step": 2136
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00017485628995337615,
      "loss": 0.1933,
      "step": 2137
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00017483368768585523,
      "loss": 0.1903,
      "step": 2138
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.000174811076726259,
      "loss": 0.1978,
      "step": 2139
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00017478845707721377,
      "loss": 0.2103,
      "step": 2140
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00017476582874134683,
      "loss": 0.1169,
      "step": 2141
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00017474319172128654,
      "loss": 0.0871,
      "step": 2142
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.0001747205460196622,
      "loss": 0.2559,
      "step": 2143
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00017469789163910418,
      "loss": 0.1737,
      "step": 2144
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.0001746752285822438,
      "loss": 0.1898,
      "step": 2145
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00017465255685171342,
      "loss": 0.1499,
      "step": 2146
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00017462987645014644,
      "loss": 0.1593,
      "step": 2147
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.0001746071873801772,
      "loss": 0.1993,
      "step": 2148
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00017458448964444108,
      "loss": 0.1268,
      "step": 2149
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00017456178324557452,
      "loss": 0.2577,
      "step": 2150
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0001745390681862148,
      "loss": 0.2465,
      "step": 2151
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0001745163444690004,
      "loss": 0.2686,
      "step": 2152
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.0001744936120965707,
      "loss": 0.2443,
      "step": 2153
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00017447087107156612,
      "loss": 0.1644,
      "step": 2154
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00017444812139662808,
      "loss": 0.2542,
      "step": 2155
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.000174425363074399,
      "loss": 0.172,
      "step": 2156
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00017440259610752222,
      "loss": 0.1822,
      "step": 2157
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.0001743798204986423,
      "loss": 0.1984,
      "step": 2158
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00017435703625040457,
      "loss": 0.2425,
      "step": 2159
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00017433424336545553,
      "loss": 0.2062,
      "step": 2160
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00017431144184644254,
      "loss": 0.1843,
      "step": 2161
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00017428863169601414,
      "loss": 0.2402,
      "step": 2162
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00017426581291681966,
      "loss": 0.2193,
      "step": 2163
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00017424298551150963,
      "loss": 0.147,
      "step": 2164
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00017422014948273543,
      "loss": 0.2407,
      "step": 2165
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.0001741973048331496,
      "loss": 0.1949,
      "step": 2166
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00017417445156540554,
      "loss": 0.1414,
      "step": 2167
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00017415158968215764,
      "loss": 0.2046,
      "step": 2168
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00017412871918606144,
      "loss": 0.1413,
      "step": 2169
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00017410584007977335,
      "loss": 0.1467,
      "step": 2170
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00017408295236595085,
      "loss": 0.2108,
      "step": 2171
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00017406005604725234,
      "loss": 0.1902,
      "step": 2172
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00017403715112633732,
      "loss": 0.1502,
      "step": 2173
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.00017401423760586625,
      "loss": 0.1554,
      "step": 2174
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.00017399131548850053,
      "loss": 0.1801,
      "step": 2175
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0001739683847769026,
      "loss": 0.1972,
      "step": 2176
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.00017394544547373597,
      "loss": 0.1778,
      "step": 2177
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.00017392249758166505,
      "loss": 0.2468,
      "step": 2178
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.00017389954110335528,
      "loss": 0.1634,
      "step": 2179
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0001738765760414731,
      "loss": 0.2443,
      "step": 2180
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00017385360239868594,
      "loss": 0.1597,
      "step": 2181
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00017383062017766222,
      "loss": 0.1668,
      "step": 2182
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00017380762938107145,
      "loss": 0.1797,
      "step": 2183
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.0001737846300115839,
      "loss": 0.1941,
      "step": 2184
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00017376162207187115,
      "loss": 0.1833,
      "step": 2185
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00017373860556460554,
      "loss": 0.2082,
      "step": 2186
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00017371558049246048,
      "loss": 0.1937,
      "step": 2187
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00017369254685811037,
      "loss": 0.2277,
      "step": 2188
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00017366950466423065,
      "loss": 0.1935,
      "step": 2189
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00017364645391349765,
      "loss": 0.191,
      "step": 2190
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00017362339460858883,
      "loss": 0.1312,
      "step": 2191
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00017360032675218254,
      "loss": 0.1366,
      "step": 2192
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00017357725034695815,
      "loss": 0.2018,
      "step": 2193
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00017355416539559606,
      "loss": 0.1258,
      "step": 2194
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00017353107190077758,
      "loss": 0.2611,
      "step": 2195
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.0001735079698651851,
      "loss": 0.323,
      "step": 2196
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.00017348485929150192,
      "loss": 0.1913,
      "step": 2197
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.00017346174018241248,
      "loss": 0.2127,
      "step": 2198
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.000173438612540602,
      "loss": 0.2409,
      "step": 2199
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0001734154763687568,
      "loss": 0.2083,
      "step": 2200
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0001733923316695643,
      "loss": 0.1723,
      "step": 2201
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.00017336917844571273,
      "loss": 0.1318,
      "step": 2202
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.0001733460166998913,
      "loss": 0.1883,
      "step": 2203
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.00017332284643479045,
      "loss": 0.3081,
      "step": 2204
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.00017329966765310134,
      "loss": 0.2408,
      "step": 2205
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.00017327648035751628,
      "loss": 0.2301,
      "step": 2206
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.00017325328455072852,
      "loss": 0.1243,
      "step": 2207
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.00017323008023543225,
      "loss": 0.1157,
      "step": 2208
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.00017320686741432274,
      "loss": 0.1833,
      "step": 2209
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.00017318364609009618,
      "loss": 0.2826,
      "step": 2210
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0001731604162654498,
      "loss": 0.148,
      "step": 2211
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.0001731371779430817,
      "loss": 0.2009,
      "step": 2212
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.0001731139311256912,
      "loss": 0.2452,
      "step": 2213
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00017309067581597836,
      "loss": 0.1737,
      "step": 2214
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00017306741201664433,
      "loss": 0.1559,
      "step": 2215
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00017304413973039128,
      "loss": 0.1314,
      "step": 2216
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00017302085895992233,
      "loss": 0.2144,
      "step": 2217
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00017299756970794156,
      "loss": 0.2484,
      "step": 2218
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.00017297427197715407,
      "loss": 0.3195,
      "step": 2219
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00017295096577026598,
      "loss": 0.1495,
      "step": 2220
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00017292765108998425,
      "loss": 0.1206,
      "step": 2221
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00017290432793901704,
      "loss": 0.1458,
      "step": 2222
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.0001728809963200733,
      "loss": 0.233,
      "step": 2223
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00017285765623586304,
      "loss": 0.2075,
      "step": 2224
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00017283430768909734,
      "loss": 0.1893,
      "step": 2225
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00017281095068248808,
      "loss": 0.138,
      "step": 2226
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.00017278758521874826,
      "loss": 0.1672,
      "step": 2227
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.00017276421130059183,
      "loss": 0.2068,
      "step": 2228
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0001727408289307337,
      "loss": 0.1339,
      "step": 2229
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0001727174381118898,
      "loss": 0.1832,
      "step": 2230
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.000172694038846777,
      "loss": 0.2141,
      "step": 2231
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.00017267063113811317,
      "loss": 0.2831,
      "step": 2232
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0001726472149886172,
      "loss": 0.1652,
      "step": 2233
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.0001726237904010088,
      "loss": 0.1955,
      "step": 2234
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00017260035737800893,
      "loss": 0.2618,
      "step": 2235
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.0001725769159223393,
      "loss": 0.2219,
      "step": 2236
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00017255346603672265,
      "loss": 0.1344,
      "step": 2237
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00017253000772388282,
      "loss": 0.2023,
      "step": 2238
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00017250654098654447,
      "loss": 0.1684,
      "step": 2239
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00017248306582743332,
      "loss": 0.1259,
      "step": 2240
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.0001724595822492761,
      "loss": 0.1468,
      "step": 2241
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.0001724360902548004,
      "loss": 0.2214,
      "step": 2242
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.0001724125898467349,
      "loss": 0.2355,
      "step": 2243
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.0001723890810278092,
      "loss": 0.1852,
      "step": 2244
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.0001723655638007539,
      "loss": 0.166,
      "step": 2245
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.0001723420381683006,
      "loss": 0.2227,
      "step": 2246
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.00017231850413318181,
      "loss": 0.1642,
      "step": 2247
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.00017229496169813104,
      "loss": 0.2268,
      "step": 2248
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.00017227141086588287,
      "loss": 0.2199,
      "step": 2249
    },
    {
      "epoch": 2.92,
      "learning_rate": 0.00017224785163917264,
      "loss": 0.1681,
      "step": 2250
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.00017222428402073693,
      "loss": 0.1722,
      "step": 2251
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.00017220070801331308,
      "loss": 0.1519,
      "step": 2252
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.00017217712361963954,
      "loss": 0.1977,
      "step": 2253
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.00017215353084245564,
      "loss": 0.1992,
      "step": 2254
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.00017212992968450178,
      "loss": 0.207,
      "step": 2255
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0001721063201485192,
      "loss": 0.2786,
      "step": 2256
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.00017208270223725028,
      "loss": 0.356,
      "step": 2257
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.00017205907595343824,
      "loss": 0.1813,
      "step": 2258
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0001720354412998273,
      "loss": 0.2467,
      "step": 2259
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0001720117982791627,
      "loss": 0.1958,
      "step": 2260
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.00017198814689419064,
      "loss": 0.2133,
      "step": 2261
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.0001719644871476582,
      "loss": 0.2212,
      "step": 2262
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.00017194081904231358,
      "loss": 0.1409,
      "step": 2263
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.00017191714258090582,
      "loss": 0.3216,
      "step": 2264
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.00017189345776618504,
      "loss": 0.1847,
      "step": 2265
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00017186976460090226,
      "loss": 0.1886,
      "step": 2266
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00017184606308780947,
      "loss": 0.2472,
      "step": 2267
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00017182235322965963,
      "loss": 0.1039,
      "step": 2268
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.0001717986350292067,
      "loss": 0.1208,
      "step": 2269
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.0001717749084892056,
      "loss": 0.2703,
      "step": 2270
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00017175117361241225,
      "loss": 0.1026,
      "step": 2271
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00017172743040158342,
      "loss": 0.1021,
      "step": 2272
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.00017170367885947703,
      "loss": 0.1675,
      "step": 2273
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.00017167991898885178,
      "loss": 0.2324,
      "step": 2274
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.00017165615079246745,
      "loss": 0.1915,
      "step": 2275
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.00017163237427308476,
      "loss": 0.1495,
      "step": 2276
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0001716085894334654,
      "loss": 0.2358,
      "step": 2277
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.00017158479627637206,
      "loss": 0.2538,
      "step": 2278
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.0001715609948045683,
      "loss": 0.1639,
      "step": 2279
    },
    {
      "epoch": 2.96,
      "learning_rate": 0.00017153718502081874,
      "loss": 0.1062,
      "step": 2280
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.00017151336692788896,
      "loss": 0.2224,
      "step": 2281
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0001714895405285454,
      "loss": 0.2107,
      "step": 2282
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0001714657058255556,
      "loss": 0.1335,
      "step": 2283
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.000171441862821688,
      "loss": 0.1085,
      "step": 2284
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.000171418011519712,
      "loss": 0.136,
      "step": 2285
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.00017139415192239795,
      "loss": 0.161,
      "step": 2286
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.00017137028403251724,
      "loss": 0.2021,
      "step": 2287
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.00017134640785284214,
      "loss": 0.1534,
      "step": 2288
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.00017132252338614593,
      "loss": 0.2178,
      "step": 2289
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.00017129863063520282,
      "loss": 0.3059,
      "step": 2290
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.00017127472960278798,
      "loss": 0.184,
      "step": 2291
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.00017125082029167762,
      "loss": 0.1533,
      "step": 2292
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0001712269027046488,
      "loss": 0.1915,
      "step": 2293
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.00017120297684447956,
      "loss": 0.1555,
      "step": 2294
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.00017117904271394906,
      "loss": 0.1638,
      "step": 2295
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.00017115510031583716,
      "loss": 0.2048,
      "step": 2296
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.0001711311496529249,
      "loss": 0.2216,
      "step": 2297
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.0001711071907279942,
      "loss": 0.1468,
      "step": 2298
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.00017108322354382786,
      "loss": 0.1922,
      "step": 2299
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.00017105924810320973,
      "loss": 0.191,
      "step": 2300
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.00017103526440892468,
      "loss": 0.1456,
      "step": 2301
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.00017101127246375842,
      "loss": 0.2829,
      "step": 2302
    },
    {
      "epoch": 2.99,
      "learning_rate": 0.00017098727227049762,
      "loss": 0.22,
      "step": 2303
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.00017096326383192999,
      "loss": 0.2229,
      "step": 2304
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.2669857442378998,
      "eval_runtime": 126.0505,
      "eval_samples_per_second": 1.015,
      "eval_steps_per_second": 0.341,
      "step": 2304
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.00017093924715084414,
      "loss": 0.1813,
      "step": 2305
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.00017091522223002968,
      "loss": 0.1929,
      "step": 2306
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0001708911890722771,
      "loss": 0.1259,
      "step": 2307
    }
  ],
  "logging_steps": 1,
  "max_steps": 9228,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 12,
  "save_steps": 500,
  "total_flos": 3.4538838739963085e+18,
  "train_batch_size": 3,
  "trial_name": null,
  "trial_params": null
}