{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.999050933248972,
  "global_step": 2370,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 2.105263157894737e-07,
      "loss": 1.5856,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.210526315789474e-07,
      "loss": 1.573,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.315789473684211e-07,
      "loss": 1.4315,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.421052631578948e-07,
      "loss": 1.3999,
      "step": 4
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.0526315789473685e-06,
      "loss": 1.424,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.2631578947368422e-06,
      "loss": 1.3637,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.4736842105263159e-06,
      "loss": 1.3418,
      "step": 7
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.6842105263157895e-06,
      "loss": 1.4105,
      "step": 8
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.8947368421052634e-06,
      "loss": 1.3645,
      "step": 9
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.105263157894737e-06,
      "loss": 1.3461,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.3157894736842105e-06,
      "loss": 1.3618,
      "step": 11
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.5263157894736844e-06,
      "loss": 1.3328,
      "step": 12
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.7368421052631583e-06,
      "loss": 1.2787,
      "step": 13
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.9473684210526317e-06,
      "loss": 1.2735,
      "step": 14
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.157894736842105e-06,
      "loss": 1.2918,
      "step": 15
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.368421052631579e-06,
      "loss": 1.2733,
      "step": 16
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.578947368421053e-06,
      "loss": 1.2679,
      "step": 17
    },
    {
      "epoch": 0.02,
      "learning_rate": 3.789473684210527e-06,
      "loss": 1.2451,
      "step": 18
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.2595,
      "step": 19
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.210526315789474e-06,
      "loss": 1.2567,
      "step": 20
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.4210526315789476e-06,
      "loss": 1.2327,
      "step": 21
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.631578947368421e-06,
      "loss": 1.2265,
      "step": 22
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.842105263157895e-06,
      "loss": 1.2121,
      "step": 23
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.052631578947369e-06,
      "loss": 1.2397,
      "step": 24
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.263157894736842e-06,
      "loss": 1.2098,
      "step": 25
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.4736842105263165e-06,
      "loss": 1.1875,
      "step": 26
    },
    {
      "epoch": 0.03,
      "learning_rate": 5.68421052631579e-06,
      "loss": 1.1877,
      "step": 27
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.8947368421052634e-06,
      "loss": 1.1919,
      "step": 28
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.105263157894738e-06,
      "loss": 1.1708,
      "step": 29
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.31578947368421e-06,
      "loss": 1.1785,
      "step": 30
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.526315789473685e-06,
      "loss": 1.15,
      "step": 31
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.736842105263158e-06,
      "loss": 1.1659,
      "step": 32
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.947368421052632e-06,
      "loss": 1.1557,
      "step": 33
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.157894736842106e-06,
      "loss": 1.1617,
      "step": 34
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.368421052631579e-06,
      "loss": 1.1419,
      "step": 35
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.578947368421054e-06,
      "loss": 1.1169,
      "step": 36
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.789473684210526e-06,
      "loss": 1.1212,
      "step": 37
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.1322,
      "step": 38
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.210526315789475e-06,
      "loss": 1.0998,
      "step": 39
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.421052631578948e-06,
      "loss": 1.1113,
      "step": 40
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.631578947368422e-06,
      "loss": 1.1033,
      "step": 41
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.842105263157895e-06,
      "loss": 1.0894,
      "step": 42
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.05263157894737e-06,
      "loss": 1.1111,
      "step": 43
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.263157894736842e-06,
      "loss": 1.1085,
      "step": 44
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.473684210526315e-06,
      "loss": 1.0887,
      "step": 45
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.68421052631579e-06,
      "loss": 1.115,
      "step": 46
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.894736842105264e-06,
      "loss": 1.0923,
      "step": 47
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.0105263157894738e-05,
      "loss": 1.0964,
      "step": 48
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.0315789473684213e-05,
      "loss": 1.078,
      "step": 49
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.0526315789473684e-05,
      "loss": 1.0953,
      "step": 50
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.073684210526316e-05,
      "loss": 1.1186,
      "step": 51
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.0947368421052633e-05,
      "loss": 1.0751,
      "step": 52
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.1157894736842105e-05,
      "loss": 1.0513,
      "step": 53
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.136842105263158e-05,
      "loss": 1.0871,
      "step": 54
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.1578947368421053e-05,
      "loss": 1.0493,
      "step": 55
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.1789473684210527e-05,
      "loss": 1.0712,
      "step": 56
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.2e-05,
      "loss": 1.0591,
      "step": 57
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.2210526315789475e-05,
      "loss": 1.0672,
      "step": 58
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.2421052631578949e-05,
      "loss": 1.0666,
      "step": 59
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.263157894736842e-05,
      "loss": 1.0666,
      "step": 60
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.2842105263157896e-05,
      "loss": 1.0562,
      "step": 61
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.305263157894737e-05,
      "loss": 1.0614,
      "step": 62
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.3263157894736843e-05,
      "loss": 1.0575,
      "step": 63
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.3473684210526316e-05,
      "loss": 1.0871,
      "step": 64
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.3684210526315791e-05,
      "loss": 1.0483,
      "step": 65
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.3894736842105265e-05,
      "loss": 1.0593,
      "step": 66
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.4105263157894738e-05,
      "loss": 1.0558,
      "step": 67
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.4315789473684212e-05,
      "loss": 1.033,
      "step": 68
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.4526315789473687e-05,
      "loss": 1.0284,
      "step": 69
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.4736842105263159e-05,
      "loss": 1.0432,
      "step": 70
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.4947368421052632e-05,
      "loss": 1.0385,
      "step": 71
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.5157894736842107e-05,
      "loss": 1.0325,
      "step": 72
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.536842105263158e-05,
      "loss": 1.0387,
      "step": 73
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.5578947368421052e-05,
      "loss": 1.0428,
      "step": 74
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.578947368421053e-05,
      "loss": 1.0494,
      "step": 75
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.0484,
      "step": 76
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.6210526315789473e-05,
      "loss": 1.0109,
      "step": 77
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.642105263157895e-05,
      "loss": 1.0118,
      "step": 78
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.6631578947368423e-05,
      "loss": 1.0488,
      "step": 79
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.6842105263157896e-05,
      "loss": 1.0156,
      "step": 80
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.705263157894737e-05,
      "loss": 1.011,
      "step": 81
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.7263157894736843e-05,
      "loss": 1.0459,
      "step": 82
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.7473684210526317e-05,
      "loss": 1.0371,
      "step": 83
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.768421052631579e-05,
      "loss": 1.0154,
      "step": 84
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.7894736842105264e-05,
      "loss": 1.0281,
      "step": 85
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.810526315789474e-05,
      "loss": 1.027,
      "step": 86
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.831578947368421e-05,
      "loss": 1.0242,
      "step": 87
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.8526315789473684e-05,
      "loss": 1.0328,
      "step": 88
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.873684210526316e-05,
      "loss": 1.0159,
      "step": 89
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.894736842105263e-05,
      "loss": 0.998,
      "step": 90
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9157894736842108e-05,
      "loss": 1.0505,
      "step": 91
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.936842105263158e-05,
      "loss": 1.0051,
      "step": 92
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9578947368421055e-05,
      "loss": 0.9971,
      "step": 93
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9789473684210528e-05,
      "loss": 1.0066,
      "step": 94
    },
    {
      "epoch": 0.12,
      "learning_rate": 2e-05,
      "loss": 1.027,
      "step": 95
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9999990465303388e-05,
      "loss": 1.0247,
      "step": 96
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9999961861231728e-05,
      "loss": 0.9942,
      "step": 97
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9999914187839568e-05,
      "loss": 1.0084,
      "step": 98
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9999847445217817e-05,
      "loss": 1.0151,
      "step": 99
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9999761633493754e-05,
      "loss": 1.0025,
      "step": 100
    },
    {
      "epoch": 0.13,
      "eval_loss": 1.0003799200057983,
      "eval_runtime": 183.8122,
      "eval_samples_per_second": 11.229,
      "eval_steps_per_second": 0.092,
      "step": 100
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.999965675283101e-05,
      "loss": 0.9928,
      "step": 101
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.999953280342959e-05,
      "loss": 0.9875,
      "step": 102
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9999389785525857e-05,
      "loss": 0.9989,
      "step": 103
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9999227699392537e-05,
      "loss": 1.019,
      "step": 104
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.999904654533872e-05,
      "loss": 0.9953,
      "step": 105
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9998846323709855e-05,
      "loss": 0.992,
      "step": 106
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.999862703488775e-05,
      "loss": 0.9932,
      "step": 107
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9998388679290583e-05,
      "loss": 0.981,
      "step": 108
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9998131257372878e-05,
      "loss": 0.9758,
      "step": 109
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.999785476962552e-05,
      "loss": 1.0076,
      "step": 110
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.999755921657576e-05,
      "loss": 1.013,
      "step": 111
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9997244598787197e-05,
      "loss": 0.9909,
      "step": 112
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9996910916859788e-05,
      "loss": 0.9722,
      "step": 113
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.999655817142985e-05,
      "loss": 0.9891,
      "step": 114
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9996186363170037e-05,
      "loss": 0.9938,
      "step": 115
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9995795492789368e-05,
      "loss": 0.9917,
      "step": 116
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9995385561033215e-05,
      "loss": 0.9778,
      "step": 117
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9994956568683285e-05,
      "loss": 0.9826,
      "step": 118
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9994508516557647e-05,
      "loss": 0.9938,
      "step": 119
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9994041405510705e-05,
      "loss": 0.9863,
      "step": 120
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9993555236433216e-05,
      "loss": 0.9885,
      "step": 121
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.999305001025227e-05,
      "loss": 0.9866,
      "step": 122
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9992525727931303e-05,
      "loss": 1.0213,
      "step": 123
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.999198239047009e-05,
      "loss": 0.9699,
      "step": 124
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.999141999890475e-05,
      "loss": 0.988,
      "step": 125
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9990838554307718e-05,
      "loss": 0.9707,
      "step": 126
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9990238057787777e-05,
      "loss": 0.9622,
      "step": 127
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9989618510490042e-05,
      "loss": 0.981,
      "step": 128
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.998897991359595e-05,
      "loss": 1.002,
      "step": 129
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.998832226832327e-05,
      "loss": 0.945,
      "step": 130
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.998764557592608e-05,
      "loss": 0.9872,
      "step": 131
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9986949837694803e-05,
      "loss": 0.9975,
      "step": 132
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9986235054956166e-05,
      "loss": 0.9723,
      "step": 133
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9985501229073213e-05,
      "loss": 0.9641,
      "step": 134
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9984748361445306e-05,
      "loss": 0.9653,
      "step": 135
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9983976453508123e-05,
      "loss": 0.9681,
      "step": 136
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9983185506733643e-05,
      "loss": 0.9549,
      "step": 137
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9982375522630155e-05,
      "loss": 0.9696,
      "step": 138
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9981546502742244e-05,
      "loss": 0.9791,
      "step": 139
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9980698448650805e-05,
      "loss": 0.9669,
      "step": 140
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9979831361973024e-05,
      "loss": 0.9601,
      "step": 141
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9978945244362387e-05,
      "loss": 0.9455,
      "step": 142
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9978040097508665e-05,
      "loss": 0.9534,
      "step": 143
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9977115923137912e-05,
      "loss": 0.933,
      "step": 144
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.997617272301248e-05,
      "loss": 0.9696,
      "step": 145
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9975210498930987e-05,
      "loss": 0.9491,
      "step": 146
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9974229252728345e-05,
      "loss": 0.9741,
      "step": 147
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.997322898627572e-05,
      "loss": 0.96,
      "step": 148
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.997220970148057e-05,
      "loss": 0.9458,
      "step": 149
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9971171400286602e-05,
      "loss": 0.9723,
      "step": 150
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9970114084673796e-05,
      "loss": 0.9586,
      "step": 151
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9969037756658394e-05,
      "loss": 0.9401,
      "step": 152
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.996794241829288e-05,
      "loss": 0.9474,
      "step": 153
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9966828071666e-05,
      "loss": 0.9362,
      "step": 154
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9965694718902745e-05,
      "loss": 0.9623,
      "step": 155
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9964542362164353e-05,
      "loss": 0.9405,
      "step": 156
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.99633710036483e-05,
      "loss": 0.9484,
      "step": 157
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.996218064558829e-05,
      "loss": 0.9477,
      "step": 158
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9960971290254262e-05,
      "loss": 0.9658,
      "step": 159
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9959742939952393e-05,
      "loss": 0.9245,
      "step": 160
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9958495597025062e-05,
      "loss": 0.9737,
      "step": 161
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9957229263850884e-05,
      "loss": 0.9494,
      "step": 162
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9955943942844677e-05,
      "loss": 0.9701,
      "step": 163
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9954639636457464e-05,
      "loss": 0.9594,
      "step": 164
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.995331634717649e-05,
      "loss": 0.9818,
      "step": 165
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9951974077525175e-05,
      "loss": 0.9504,
      "step": 166
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9950612830063155e-05,
      "loss": 0.9353,
      "step": 167
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9949232607386245e-05,
      "loss": 0.9599,
      "step": 168
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.994783341212644e-05,
      "loss": 0.9689,
      "step": 169
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9946415246951928e-05,
      "loss": 0.945,
      "step": 170
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.994497811456706e-05,
      "loss": 0.9566,
      "step": 171
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.994352201771236e-05,
      "loss": 0.9239,
      "step": 172
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9942046959164516e-05,
      "loss": 0.9525,
      "step": 173
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9940552941736376e-05,
      "loss": 0.9369,
      "step": 174
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9939039968276942e-05,
      "loss": 0.9158,
      "step": 175
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9937508041671363e-05,
      "loss": 0.9533,
      "step": 176
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9935957164840925e-05,
      "loss": 0.9295,
      "step": 177
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9934387340743063e-05,
      "loss": 0.9615,
      "step": 178
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.993279857237133e-05,
      "loss": 0.9446,
      "step": 179
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9931190862755416e-05,
      "loss": 0.9404,
      "step": 180
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9929564214961124e-05,
      "loss": 0.9434,
      "step": 181
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.992791863209037e-05,
      "loss": 0.9198,
      "step": 182
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.992625411728118e-05,
      "loss": 0.9669,
      "step": 183
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.992457067370769e-05,
      "loss": 0.9371,
      "step": 184
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.992286830458012e-05,
      "loss": 0.9513,
      "step": 185
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9921147013144782e-05,
      "loss": 0.9284,
      "step": 186
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9919406802684075e-05,
      "loss": 0.9693,
      "step": 187
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.991764767651648e-05,
      "loss": 0.9406,
      "step": 188
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9915869637996544e-05,
      "loss": 0.9496,
      "step": 189
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.991407269051487e-05,
      "loss": 0.9033,
      "step": 190
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9912256837498133e-05,
      "loss": 0.9046,
      "step": 191
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9910422082409057e-05,
      "loss": 0.9146,
      "step": 192
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9908568428746408e-05,
      "loss": 0.9361,
      "step": 193
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9906695880044986e-05,
      "loss": 0.9453,
      "step": 194
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9904804439875635e-05,
      "loss": 0.9564,
      "step": 195
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.990289411184521e-05,
      "loss": 0.9273,
      "step": 196
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.990096489959659e-05,
      "loss": 0.9098,
      "step": 197
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9899016806808675e-05,
      "loss": 0.9249,
      "step": 198
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.989704983719635e-05,
      "loss": 0.9249,
      "step": 199
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9895063994510512e-05,
      "loss": 0.9057,
      "step": 200
    },
    {
      "epoch": 0.25,
      "eval_loss": 0.9254624843597412,
      "eval_runtime": 182.5274,
      "eval_samples_per_second": 11.308,
      "eval_steps_per_second": 0.093,
      "step": 200
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.989305928253804e-05,
      "loss": 0.9289,
      "step": 201
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.98910357051018e-05,
      "loss": 0.9313,
      "step": 202
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.988899326606063e-05,
      "loss": 0.9461,
      "step": 203
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.988693196930934e-05,
      "loss": 0.9358,
      "step": 204
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9884851818778695e-05,
      "loss": 0.9441,
      "step": 205
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9882752818435415e-05,
      "loss": 0.9287,
      "step": 206
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9880634972282168e-05,
      "loss": 0.9182,
      "step": 207
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9878498284357557e-05,
      "loss": 0.9038,
      "step": 208
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9876342758736118e-05,
      "loss": 0.9262,
      "step": 209
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9874168399528307e-05,
      "loss": 0.9464,
      "step": 210
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9871975210880494e-05,
      "loss": 0.9069,
      "step": 211
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9869763196974957e-05,
      "loss": 0.9091,
      "step": 212
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9867532362029874e-05,
      "loss": 0.9083,
      "step": 213
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.986528271029931e-05,
      "loss": 0.9085,
      "step": 214
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9863014246073216e-05,
      "loss": 0.9179,
      "step": 215
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9860726973677416e-05,
      "loss": 0.9563,
      "step": 216
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9858420897473594e-05,
      "loss": 0.9021,
      "step": 217
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9856096021859305e-05,
      "loss": 0.9153,
      "step": 218
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.985375235126794e-05,
      "loss": 0.9098,
      "step": 219
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9851389890168738e-05,
      "loss": 0.9108,
      "step": 220
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9849008643066774e-05,
      "loss": 0.9227,
      "step": 221
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9846608614502934e-05,
      "loss": 0.9182,
      "step": 222
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9844189809053933e-05,
      "loss": 0.9255,
      "step": 223
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9841752231332286e-05,
      "loss": 0.9157,
      "step": 224
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.98392958859863e-05,
      "loss": 0.9412,
      "step": 225
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.983682077770008e-05,
      "loss": 0.9014,
      "step": 226
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9834326911193506e-05,
      "loss": 0.8872,
      "step": 227
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9831814291222233e-05,
      "loss": 0.9404,
      "step": 228
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9829282922577675e-05,
      "loss": 0.8956,
      "step": 229
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9826732810087e-05,
      "loss": 0.9145,
      "step": 230
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9824163958613107e-05,
      "loss": 0.9151,
      "step": 231
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.982157637305465e-05,
      "loss": 0.9047,
      "step": 232
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9818970058345997e-05,
      "loss": 0.9222,
      "step": 233
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.981634501945723e-05,
      "loss": 0.9056,
      "step": 234
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9813701261394136e-05,
      "loss": 0.9069,
      "step": 235
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9811038789198206e-05,
      "loss": 0.9425,
      "step": 236
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9808357607946613e-05,
      "loss": 0.9421,
      "step": 237
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9805657722752202e-05,
      "loss": 0.9084,
      "step": 238
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9802939138763495e-05,
      "loss": 0.9068,
      "step": 239
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9800201861164665e-05,
      "loss": 0.8996,
      "step": 240
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9797445895175533e-05,
      "loss": 0.929,
      "step": 241
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.979467124605156e-05,
      "loss": 0.9214,
      "step": 242
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9791877919083836e-05,
      "loss": 0.9014,
      "step": 243
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9789065919599064e-05,
      "loss": 0.9481,
      "step": 244
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9786235252959555e-05,
      "loss": 0.9084,
      "step": 245
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9783385924563218e-05,
      "loss": 0.9247,
      "step": 246
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9780517939843555e-05,
      "loss": 0.8979,
      "step": 247
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.977763130426963e-05,
      "loss": 0.8965,
      "step": 248
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.977472602334609e-05,
      "loss": 0.9232,
      "step": 249
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9771802102613127e-05,
      "loss": 0.8871,
      "step": 250
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9768859547646476e-05,
      "loss": 0.911,
      "step": 251
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9765898364057418e-05,
      "loss": 0.8846,
      "step": 252
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9762918557492748e-05,
      "loss": 0.9223,
      "step": 253
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.975992013363477e-05,
      "loss": 0.9301,
      "step": 254
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.975690309820131e-05,
      "loss": 0.917,
      "step": 255
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9753867456945653e-05,
      "loss": 0.9186,
      "step": 256
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.97508132156566e-05,
      "loss": 0.8978,
      "step": 257
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9747740380158392e-05,
      "loss": 0.8943,
      "step": 258
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9744648956310747e-05,
      "loss": 0.9117,
      "step": 259
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9741538950008817e-05,
      "loss": 0.8884,
      "step": 260
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9738410367183204e-05,
      "loss": 0.9011,
      "step": 261
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9735263213799914e-05,
      "loss": 0.9204,
      "step": 262
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9732097495860388e-05,
      "loss": 0.8826,
      "step": 263
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.972891321940145e-05,
      "loss": 0.8948,
      "step": 264
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.972571039049533e-05,
      "loss": 0.9024,
      "step": 265
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9722489015249618e-05,
      "loss": 0.9141,
      "step": 266
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.971924909980729e-05,
      "loss": 0.8836,
      "step": 267
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9715990650346667e-05,
      "loss": 0.9196,
      "step": 268
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.971271367308141e-05,
      "loss": 0.8981,
      "step": 269
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9709418174260523e-05,
      "loss": 0.8926,
      "step": 270
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9706104160168315e-05,
      "loss": 0.9048,
      "step": 271
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9702771637124408e-05,
      "loss": 0.8849,
      "step": 272
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.969942061148373e-05,
      "loss": 0.9038,
      "step": 273
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.969605108963648e-05,
      "loss": 0.8571,
      "step": 274
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.969266307800813e-05,
      "loss": 0.8953,
      "step": 275
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9689256583059416e-05,
      "loss": 0.9145,
      "step": 276
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9685831611286312e-05,
      "loss": 0.8775,
      "step": 277
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.9682388169220036e-05,
      "loss": 0.9167,
      "step": 278
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.967892626342702e-05,
      "loss": 0.8917,
      "step": 279
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.967544590050891e-05,
      "loss": 0.8945,
      "step": 280
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.967194708710255e-05,
      "loss": 0.8877,
      "step": 281
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9668429829879953e-05,
      "loss": 0.8763,
      "step": 282
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.966489413554833e-05,
      "loss": 0.8895,
      "step": 283
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9661340010850025e-05,
      "loss": 0.8838,
      "step": 284
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9657767462562544e-05,
      "loss": 0.8901,
      "step": 285
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.965417649749852e-05,
      "loss": 0.9071,
      "step": 286
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.96505671225057e-05,
      "loss": 0.9026,
      "step": 287
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9646939344466948e-05,
      "loss": 0.8879,
      "step": 288
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9643293170300216e-05,
      "loss": 0.9105,
      "step": 289
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9639628606958535e-05,
      "loss": 0.8884,
      "step": 290
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9635945661430006e-05,
      "loss": 0.8981,
      "step": 291
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9632244340737782e-05,
      "loss": 0.8907,
      "step": 292
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.962852465194006e-05,
      "loss": 0.8881,
      "step": 293
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9624786602130058e-05,
      "loss": 0.8909,
      "step": 294
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9621030198436007e-05,
      "loss": 0.8896,
      "step": 295
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9617255448021147e-05,
      "loss": 0.9147,
      "step": 296
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9613462358083697e-05,
      "loss": 0.893,
      "step": 297
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9609650935856847e-05,
      "loss": 0.8995,
      "step": 298
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9605821188608747e-05,
      "loss": 0.9098,
      "step": 299
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9601973123642493e-05,
      "loss": 0.8801,
      "step": 300
    },
    {
      "epoch": 0.38,
      "eval_loss": 0.8874989748001099,
      "eval_runtime": 182.6966,
      "eval_samples_per_second": 11.297,
      "eval_steps_per_second": 0.093,
      "step": 300
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9598106748296117e-05,
      "loss": 0.9046,
      "step": 301
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9594222069942556e-05,
      "loss": 0.869,
      "step": 302
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.959031909598966e-05,
      "loss": 0.8862,
      "step": 303
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9586397833880156e-05,
      "loss": 0.9149,
      "step": 304
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9582458291091664e-05,
      "loss": 0.9036,
      "step": 305
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9578500475136642e-05,
      "loss": 0.8947,
      "step": 306
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9574524393562417e-05,
      "loss": 0.8886,
      "step": 307
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9570530053951124e-05,
      "loss": 0.8643,
      "step": 308
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.956651746391973e-05,
      "loss": 0.8804,
      "step": 309
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9562486631120007e-05,
      "loss": 0.8715,
      "step": 310
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.95584375632385e-05,
      "loss": 0.8835,
      "step": 311
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9554370267996537e-05,
      "loss": 0.9177,
      "step": 312
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.955028475315021e-05,
      "loss": 0.8728,
      "step": 313
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.954618102649034e-05,
      "loss": 0.898,
      "step": 314
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9542059095842484e-05,
      "loss": 0.8816,
      "step": 315
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9537918969066923e-05,
      "loss": 0.853,
      "step": 316
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.953376065405862e-05,
      "loss": 0.9033,
      "step": 317
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9529584158747232e-05,
      "loss": 0.8923,
      "step": 318
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.952538949109708e-05,
      "loss": 0.8732,
      "step": 319
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.952117665910714e-05,
      "loss": 0.8665,
      "step": 320
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9516945670811035e-05,
      "loss": 0.8773,
      "step": 321
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9512696534276995e-05,
      "loss": 0.8861,
      "step": 322
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9508429257607866e-05,
      "loss": 0.8531,
      "step": 323
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9504143848941092e-05,
      "loss": 0.8855,
      "step": 324
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9499840316448675e-05,
      "loss": 0.8832,
      "step": 325
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9495518668337204e-05,
      "loss": 0.9156,
      "step": 326
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9491178912847794e-05,
      "loss": 0.8972,
      "step": 327
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9486821058256092e-05,
      "loss": 0.8774,
      "step": 328
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9482445112872265e-05,
      "loss": 0.8577,
      "step": 329
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9478051085040978e-05,
      "loss": 0.8896,
      "step": 330
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.947363898314137e-05,
      "loss": 0.8906,
      "step": 331
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9469208815587058e-05,
      "loss": 0.8787,
      "step": 332
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.94647605908261e-05,
      "loss": 0.9051,
      "step": 333
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.946029431734099e-05,
      "loss": 0.8842,
      "step": 334
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.945581000364864e-05,
      "loss": 0.8896,
      "step": 335
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9451307658300363e-05,
      "loss": 0.8987,
      "step": 336
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.944678728988186e-05,
      "loss": 0.89,
      "step": 337
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9442248907013203e-05,
      "loss": 0.86,
      "step": 338
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.943769251834881e-05,
      "loss": 0.8767,
      "step": 339
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9433118132577432e-05,
      "loss": 0.8826,
      "step": 340
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.942852575842215e-05,
      "loss": 0.8854,
      "step": 341
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.942391540464035e-05,
      "loss": 0.8767,
      "step": 342
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9419287080023685e-05,
      "loss": 0.8722,
      "step": 343
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9414640793398094e-05,
      "loss": 0.8764,
      "step": 344
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9409976553623767e-05,
      "loss": 0.8941,
      "step": 345
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.940529436959512e-05,
      "loss": 0.8949,
      "step": 346
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.94005942502408e-05,
      "loss": 0.8662,
      "step": 347
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9395876204523643e-05,
      "loss": 0.8981,
      "step": 348
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9391140241440678e-05,
      "loss": 0.8805,
      "step": 349
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9386386370023104e-05,
      "loss": 0.8632,
      "step": 350
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.938161459933626e-05,
      "loss": 0.8743,
      "step": 351
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9376824938479623e-05,
      "loss": 0.8851,
      "step": 352
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9372017396586792e-05,
      "loss": 0.8783,
      "step": 353
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.936719198282545e-05,
      "loss": 0.8715,
      "step": 354
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9362348706397374e-05,
      "loss": 0.8925,
      "step": 355
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9357487576538393e-05,
      "loss": 0.8518,
      "step": 356
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.935260860251839e-05,
      "loss": 0.8781,
      "step": 357
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9347711793641274e-05,
      "loss": 0.8857,
      "step": 358
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9342797159244958e-05,
      "loss": 0.8853,
      "step": 359
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.933786470870136e-05,
      "loss": 0.8686,
      "step": 360
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.933291445141635e-05,
      "loss": 0.9086,
      "step": 361
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9327946396829777e-05,
      "loss": 0.8768,
      "step": 362
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9322960554415422e-05,
      "loss": 0.8762,
      "step": 363
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.931795693368098e-05,
      "loss": 0.881,
      "step": 364
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.931293554416805e-05,
      "loss": 0.8802,
      "step": 365
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.930789639545212e-05,
      "loss": 0.8719,
      "step": 366
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9302839497142543e-05,
      "loss": 0.8667,
      "step": 367
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9297764858882516e-05,
      "loss": 0.8604,
      "step": 368
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9292672490349066e-05,
      "loss": 0.8499,
      "step": 369
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9287562401253023e-05,
      "loss": 0.8578,
      "step": 370
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.928243460133903e-05,
      "loss": 0.8884,
      "step": 371
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9277289100385482e-05,
      "loss": 0.8838,
      "step": 372
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9272125908204538e-05,
      "loss": 0.866,
      "step": 373
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.92669450346421e-05,
      "loss": 0.9028,
      "step": 374
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9261746489577767e-05,
      "loss": 0.8636,
      "step": 375
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9256530282924854e-05,
      "loss": 0.8787,
      "step": 376
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9251296424630355e-05,
      "loss": 0.8878,
      "step": 377
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9246044924674917e-05,
      "loss": 0.8466,
      "step": 378
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.924077579307283e-05,
      "loss": 0.8454,
      "step": 379
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.923548903987201e-05,
      "loss": 0.8519,
      "step": 380
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9230184675153974e-05,
      "loss": 0.8841,
      "step": 381
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9224862709033823e-05,
      "loss": 0.8598,
      "step": 382
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.921952315166023e-05,
      "loss": 0.8864,
      "step": 383
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9214166013215395e-05,
      "loss": 0.8596,
      "step": 384
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9208791303915063e-05,
      "loss": 0.8436,
      "step": 385
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.920339903400848e-05,
      "loss": 0.8799,
      "step": 386
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9197989213778375e-05,
      "loss": 0.8703,
      "step": 387
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.919256185354095e-05,
      "loss": 0.8699,
      "step": 388
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9187116963645845e-05,
      "loss": 0.8433,
      "step": 389
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.918165455447614e-05,
      "loss": 0.8771,
      "step": 390
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9176174636448315e-05,
      "loss": 0.883,
      "step": 391
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9170677220012244e-05,
      "loss": 0.8665,
      "step": 392
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9165162315651168e-05,
      "loss": 0.8722,
      "step": 393
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9159629933881666e-05,
      "loss": 0.8745,
      "step": 394
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9154080085253665e-05,
      "loss": 0.8506,
      "step": 395
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9148512780350384e-05,
      "loss": 0.8619,
      "step": 396
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9142928029788335e-05,
      "loss": 0.8706,
      "step": 397
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9137325844217304e-05,
      "loss": 0.8874,
      "step": 398
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9131706234320313e-05,
      "loss": 0.8949,
      "step": 399
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.912606921081362e-05,
      "loss": 0.8715,
      "step": 400
    },
    {
      "epoch": 0.51,
      "eval_loss": 0.8623976707458496,
      "eval_runtime": 182.5985,
      "eval_samples_per_second": 11.303,
      "eval_steps_per_second": 0.093,
      "step": 400
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9120414784446685e-05,
      "loss": 0.8348,
      "step": 401
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9114742966002157e-05,
      "loss": 0.8565,
      "step": 402
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.910905376629585e-05,
      "loss": 0.8473,
      "step": 403
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9103347196176726e-05,
      "loss": 0.881,
      "step": 404
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.909762326652686e-05,
      "loss": 0.8793,
      "step": 405
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9091881988261445e-05,
      "loss": 0.853,
      "step": 406
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.9086123372328748e-05,
      "loss": 0.8506,
      "step": 407
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.9080347429710097e-05,
      "loss": 0.8603,
      "step": 408
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.907455417141987e-05,
      "loss": 0.8687,
      "step": 409
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.9068743608505454e-05,
      "loss": 0.889,
      "step": 410
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.9062915752047246e-05,
      "loss": 0.868,
      "step": 411
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.905707061315861e-05,
      "loss": 0.8741,
      "step": 412
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.9051208202985872e-05,
      "loss": 0.8573,
      "step": 413
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.9045328532708293e-05,
      "loss": 0.8929,
      "step": 414
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.9039431613538047e-05,
      "loss": 0.8728,
      "step": 415
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.90335174567202e-05,
      "loss": 0.865,
      "step": 416
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.902758607353269e-05,
      "loss": 0.8799,
      "step": 417
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.902163747528631e-05,
      "loss": 0.8378,
      "step": 418
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.9015671673324666e-05,
      "loss": 0.8497,
      "step": 419
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.900968867902419e-05,
      "loss": 0.8528,
      "step": 420
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.900368850379409e-05,
      "loss": 0.8703,
      "step": 421
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.8997671159076328e-05,
      "loss": 0.8765,
      "step": 422
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.899163665634562e-05,
      "loss": 0.8552,
      "step": 423
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.898558500710939e-05,
      "loss": 0.8557,
      "step": 424
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8979516222907776e-05,
      "loss": 0.8481,
      "step": 425
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.897343031531358e-05,
      "loss": 0.8875,
      "step": 426
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.896732729593225e-05,
      "loss": 0.8447,
      "step": 427
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8961207176401875e-05,
      "loss": 0.8725,
      "step": 428
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8955069968393164e-05,
      "loss": 0.8577,
      "step": 429
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.8948915683609387e-05,
      "loss": 0.8704,
      "step": 430
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.89427443337864e-05,
      "loss": 0.8554,
      "step": 431
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8936555930692583e-05,
      "loss": 0.8713,
      "step": 432
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8930350486128855e-05,
      "loss": 0.8635,
      "step": 433
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8924128011928616e-05,
      "loss": 0.8419,
      "step": 434
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8917888519957756e-05,
      "loss": 0.8492,
      "step": 435
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.891163202211459e-05,
      "loss": 0.8608,
      "step": 436
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.8905358530329902e-05,
      "loss": 0.8541,
      "step": 437
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.889906805656684e-05,
      "loss": 0.8566,
      "step": 438
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8892760612820965e-05,
      "loss": 0.8513,
      "step": 439
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8886436211120195e-05,
      "loss": 0.8558,
      "step": 440
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.888009486352477e-05,
      "loss": 0.8485,
      "step": 441
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.887373658212726e-05,
      "loss": 0.8375,
      "step": 442
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8867361379052517e-05,
      "loss": 0.8686,
      "step": 443
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8860969266457676e-05,
      "loss": 0.8515,
      "step": 444
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8854560256532098e-05,
      "loss": 0.8577,
      "step": 445
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.8848134361497385e-05,
      "loss": 0.8751,
      "step": 446
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.884169159360732e-05,
      "loss": 0.8656,
      "step": 447
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.883523196514788e-05,
      "loss": 0.8763,
      "step": 448
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8828755488437177e-05,
      "loss": 0.8402,
      "step": 449
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8822262175825463e-05,
      "loss": 0.8483,
      "step": 450
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8815752039695087e-05,
      "loss": 0.8648,
      "step": 451
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8809225092460488e-05,
      "loss": 0.8462,
      "step": 452
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8802681346568155e-05,
      "loss": 0.8446,
      "step": 453
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8796120814496615e-05,
      "loss": 0.8569,
      "step": 454
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.878954350875641e-05,
      "loss": 0.8393,
      "step": 455
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8782949441890058e-05,
      "loss": 0.8696,
      "step": 456
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.877633862647204e-05,
      "loss": 0.8311,
      "step": 457
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8769711075108788e-05,
      "loss": 0.8558,
      "step": 458
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8763066800438638e-05,
      "loss": 0.8652,
      "step": 459
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8756405815131815e-05,
      "loss": 0.8852,
      "step": 460
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.874972813189042e-05,
      "loss": 0.8613,
      "step": 461
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8743033763448383e-05,
      "loss": 0.8535,
      "step": 462
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.873632272257146e-05,
      "loss": 0.8394,
      "step": 463
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.8729595022057206e-05,
      "loss": 0.8404,
      "step": 464
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.872285067473493e-05,
      "loss": 0.8274,
      "step": 465
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.8716089693465696e-05,
      "loss": 0.845,
      "step": 466
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.870931209114228e-05,
      "loss": 0.8554,
      "step": 467
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.8702517880689167e-05,
      "loss": 0.8548,
      "step": 468
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.8695707075062498e-05,
      "loss": 0.8599,
      "step": 469
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.8688879687250067e-05,
      "loss": 0.8621,
      "step": 470
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.868203573027129e-05,
      "loss": 0.8538,
      "step": 471
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8675175217177176e-05,
      "loss": 0.8615,
      "step": 472
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8668298161050308e-05,
      "loss": 0.849,
      "step": 473
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8661404575004818e-05,
      "loss": 0.8615,
      "step": 474
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8654494472186352e-05,
      "loss": 0.8255,
      "step": 475
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8647567865772054e-05,
      "loss": 0.8433,
      "step": 476
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8640624768970548e-05,
      "loss": 0.8506,
      "step": 477
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8633665195021898e-05,
      "loss": 0.8295,
      "step": 478
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8626689157197588e-05,
      "loss": 0.8515,
      "step": 479
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8619696668800494e-05,
      "loss": 0.8597,
      "step": 480
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8612687743164875e-05,
      "loss": 0.8537,
      "step": 481
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.860566239365632e-05,
      "loss": 0.8941,
      "step": 482
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.859862063367175e-05,
      "loss": 0.8616,
      "step": 483
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8591562476639374e-05,
      "loss": 0.8452,
      "step": 484
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8584487936018663e-05,
      "loss": 0.86,
      "step": 485
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.857739702530034e-05,
      "loss": 0.8596,
      "step": 486
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8570289758006346e-05,
      "loss": 0.8572,
      "step": 487
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8563166147689802e-05,
      "loss": 0.836,
      "step": 488
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8556026207935002e-05,
      "loss": 0.867,
      "step": 489
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.854886995235738e-05,
      "loss": 0.8326,
      "step": 490
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8541697394603483e-05,
      "loss": 0.859,
      "step": 491
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8534508548350934e-05,
      "loss": 0.8407,
      "step": 492
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.852730342730844e-05,
      "loss": 0.8518,
      "step": 493
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.852008204521572e-05,
      "loss": 0.8636,
      "step": 494
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8512844415843514e-05,
      "loss": 0.8556,
      "step": 495
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.850559055299354e-05,
      "loss": 0.8647,
      "step": 496
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.849832047049848e-05,
      "loss": 0.8453,
      "step": 497
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.849103418222194e-05,
      "loss": 0.8114,
      "step": 498
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8483731702058422e-05,
      "loss": 0.853,
      "step": 499
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8476413043933316e-05,
      "loss": 0.8399,
      "step": 500
    },
    {
      "epoch": 0.63,
      "eval_loss": 0.8413968682289124,
      "eval_runtime": 182.5951,
      "eval_samples_per_second": 11.304,
      "eval_steps_per_second": 0.093,
      "step": 500
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.846907822180286e-05,
      "loss": 0.8635,
      "step": 501
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.846172724965412e-05,
      "loss": 0.8283,
      "step": 502
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8454360141504943e-05,
      "loss": 0.8558,
      "step": 503
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8446976911403964e-05,
      "loss": 0.8526,
      "step": 504
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8439577573430557e-05,
      "loss": 0.8625,
      "step": 505
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8432162141694804e-05,
      "loss": 0.8521,
      "step": 506
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8424730630337486e-05,
      "loss": 0.8546,
      "step": 507
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8417283053530047e-05,
      "loss": 0.8346,
      "step": 508
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.840981942547456e-05,
      "loss": 0.8294,
      "step": 509
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8402339760403715e-05,
      "loss": 0.8266,
      "step": 510
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8394844072580772e-05,
      "loss": 0.8481,
      "step": 511
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8387332376299564e-05,
      "loss": 0.8671,
      "step": 512
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8379804685884432e-05,
      "loss": 0.8508,
      "step": 513
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8372261015690225e-05,
      "loss": 0.8386,
      "step": 514
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8364701380102267e-05,
      "loss": 0.86,
      "step": 515
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8357125793536328e-05,
      "loss": 0.8192,
      "step": 516
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.8349534270438583e-05,
      "loss": 0.8411,
      "step": 517
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8341926825285613e-05,
      "loss": 0.8548,
      "step": 518
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.833430347258435e-05,
      "loss": 0.8394,
      "step": 519
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8326664226872063e-05,
      "loss": 0.8352,
      "step": 520
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8319009102716338e-05,
      "loss": 0.8527,
      "step": 521
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.831133811471503e-05,
      "loss": 0.8211,
      "step": 522
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8303651277496245e-05,
      "loss": 0.8522,
      "step": 523
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8295948605718316e-05,
      "loss": 0.8449,
      "step": 524
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.828823011406977e-05,
      "loss": 0.8211,
      "step": 525
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.82804958172693e-05,
      "loss": 0.8524,
      "step": 526
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8272745730065743e-05,
      "loss": 0.8352,
      "step": 527
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.826497986723805e-05,
      "loss": 0.8246,
      "step": 528
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.825719824359524e-05,
      "loss": 0.8352,
      "step": 529
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.824940087397641e-05,
      "loss": 0.8079,
      "step": 530
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.824158777325066e-05,
      "loss": 0.8219,
      "step": 531
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8233758956317104e-05,
      "loss": 0.8572,
      "step": 532
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.822591443810482e-05,
      "loss": 0.8294,
      "step": 533
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8218054233572827e-05,
      "loss": 0.8245,
      "step": 534
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8210178357710057e-05,
      "loss": 0.8449,
      "step": 535
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.820228682553533e-05,
      "loss": 0.8453,
      "step": 536
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.819437965209732e-05,
      "loss": 0.824,
      "step": 537
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8186456852474523e-05,
      "loss": 0.8433,
      "step": 538
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8178518441775243e-05,
      "loss": 0.8527,
      "step": 539
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8170564435137542e-05,
      "loss": 0.8239,
      "step": 540
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.816259484772923e-05,
      "loss": 0.8612,
      "step": 541
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8154609694747826e-05,
      "loss": 0.8328,
      "step": 542
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8146608991420533e-05,
      "loss": 0.8466,
      "step": 543
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8138592753004207e-05,
      "loss": 0.8629,
      "step": 544
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8130560994785325e-05,
      "loss": 0.8488,
      "step": 545
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.812251373207997e-05,
      "loss": 0.833,
      "step": 546
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8114450980233775e-05,
      "loss": 0.8425,
      "step": 547
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8106372754621927e-05,
      "loss": 0.8356,
      "step": 548
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8098279070649106e-05,
      "loss": 0.8332,
      "step": 549
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 0.824,
      "step": 550
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.8082045389386654e-05,
      "loss": 0.8295,
      "step": 551
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.807390542305367e-05,
      "loss": 0.8311,
      "step": 552
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.8065750060272946e-05,
      "loss": 0.8418,
      "step": 553
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.805757931659626e-05,
      "loss": 0.8233,
      "step": 554
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.8049393207604734e-05,
      "loss": 0.8319,
      "step": 555
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.804119174890877e-05,
      "loss": 0.8316,
      "step": 556
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.8032974956148064e-05,
      "loss": 0.8576,
      "step": 557
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.8024742844991532e-05,
      "loss": 0.8503,
      "step": 558
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.8016495431137313e-05,
      "loss": 0.8494,
      "step": 559
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.8008232730312724e-05,
      "loss": 0.8239,
      "step": 560
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.799995475827424e-05,
      "loss": 0.8455,
      "step": 561
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.7991661530807443e-05,
      "loss": 0.8234,
      "step": 562
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.7983353063727014e-05,
      "loss": 0.8302,
      "step": 563
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.7975029372876706e-05,
      "loss": 0.8354,
      "step": 564
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.7966690474129285e-05,
      "loss": 0.8354,
      "step": 565
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7958336383386525e-05,
      "loss": 0.8236,
      "step": 566
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7949967116579173e-05,
      "loss": 0.8539,
      "step": 567
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.794158268966691e-05,
      "loss": 0.8251,
      "step": 568
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7933183118638328e-05,
      "loss": 0.8495,
      "step": 569
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7924768419510906e-05,
      "loss": 0.8453,
      "step": 570
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.791633860833096e-05,
      "loss": 0.827,
      "step": 571
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.7907893701173626e-05,
      "loss": 0.8377,
      "step": 572
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7899433714142836e-05,
      "loss": 0.8195,
      "step": 573
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7890958663371268e-05,
      "loss": 0.8476,
      "step": 574
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7882468565020327e-05,
      "loss": 0.8383,
      "step": 575
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7873963435280122e-05,
      "loss": 0.8167,
      "step": 576
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7865443290369414e-05,
      "loss": 0.855,
      "step": 577
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7856908146535602e-05,
      "loss": 0.8311,
      "step": 578
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7848358020054693e-05,
      "loss": 0.8176,
      "step": 579
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.7839792927231253e-05,
      "loss": 0.8352,
      "step": 580
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.7831212884398402e-05,
      "loss": 0.813,
      "step": 581
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.782261790791775e-05,
      "loss": 0.8426,
      "step": 582
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.7814008014179407e-05,
      "loss": 0.836,
      "step": 583
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.7805383219601912e-05,
      "loss": 0.836,
      "step": 584
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.7796743540632226e-05,
      "loss": 0.8284,
      "step": 585
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.778808899374569e-05,
      "loss": 0.8461,
      "step": 586
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.7779419595446005e-05,
      "loss": 0.8556,
      "step": 587
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.7770735362265185e-05,
      "loss": 0.8244,
      "step": 588
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.7762036310763533e-05,
      "loss": 0.8301,
      "step": 589
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.7753322457529615e-05,
      "loss": 0.826,
      "step": 590
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.774459381918022e-05,
      "loss": 0.835,
      "step": 591
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.7735850412360332e-05,
      "loss": 0.8279,
      "step": 592
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.77270922537431e-05,
      "loss": 0.8173,
      "step": 593
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.7718319360029794e-05,
      "loss": 0.8159,
      "step": 594
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.7709531747949796e-05,
      "loss": 0.8174,
      "step": 595
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.770072943426055e-05,
      "loss": 0.8203,
      "step": 596
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7691912435747526e-05,
      "loss": 0.7989,
      "step": 597
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7683080769224218e-05,
      "loss": 0.8416,
      "step": 598
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7674234451532065e-05,
      "loss": 0.8278,
      "step": 599
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7665373499540464e-05,
      "loss": 0.8335,
      "step": 600
    },
    {
      "epoch": 0.76,
      "eval_loss": 0.8235897421836853,
      "eval_runtime": 182.4348,
      "eval_samples_per_second": 11.314,
      "eval_steps_per_second": 0.093,
      "step": 600
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7656497930146715e-05,
      "loss": 0.8214,
      "step": 601
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7647607760275987e-05,
      "loss": 0.8547,
      "step": 602
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7638703006881295e-05,
      "loss": 0.7989,
      "step": 603
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.762978368694346e-05,
      "loss": 0.8015,
      "step": 604
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7620849817471094e-05,
      "loss": 0.8215,
      "step": 605
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7611901415500536e-05,
      "loss": 0.8182,
      "step": 606
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7602938498095845e-05,
      "loss": 0.8063,
      "step": 607
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7593961082348765e-05,
      "loss": 0.8063,
      "step": 608
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.758496918537868e-05,
      "loss": 0.843,
      "step": 609
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7575962824332595e-05,
      "loss": 0.8005,
      "step": 610
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7566942016385088e-05,
      "loss": 0.8265,
      "step": 611
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.75579067787383e-05,
      "loss": 0.8137,
      "step": 612
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7548857128621878e-05,
      "loss": 0.8252,
      "step": 613
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7539793083292953e-05,
      "loss": 0.8399,
      "step": 614
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7530714660036112e-05,
      "loss": 0.8101,
      "step": 615
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7521621876163355e-05,
      "loss": 0.8362,
      "step": 616
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.751251474901407e-05,
      "loss": 0.8331,
      "step": 617
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7503393295955002e-05,
      "loss": 0.8145,
      "step": 618
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.74942575343802e-05,
      "loss": 0.8136,
      "step": 619
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7485107481711014e-05,
      "loss": 0.8293,
      "step": 620
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.7475943155396032e-05,
      "loss": 0.8642,
      "step": 621
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.746676457291107e-05,
      "loss": 0.8303,
      "step": 622
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.7457571751759133e-05,
      "loss": 0.8262,
      "step": 623
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.744836470947037e-05,
      "loss": 0.8174,
      "step": 624
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.7439143463602052e-05,
      "loss": 0.8105,
      "step": 625
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.742990803173853e-05,
      "loss": 0.8016,
      "step": 626
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.7420658431491224e-05,
      "loss": 0.8123,
      "step": 627
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.741139468049855e-05,
      "loss": 0.8042,
      "step": 628
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.740211679642592e-05,
      "loss": 0.8196,
      "step": 629
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7392824796965703e-05,
      "loss": 0.8462,
      "step": 630
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7383518699837174e-05,
      "loss": 0.8122,
      "step": 631
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7374198522786497e-05,
      "loss": 0.831,
      "step": 632
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.736486428358668e-05,
      "loss": 0.8097,
      "step": 633
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7355516000037555e-05,
      "loss": 0.8084,
      "step": 634
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.734615368996573e-05,
      "loss": 0.8428,
      "step": 635
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.733677737122456e-05,
      "loss": 0.8297,
      "step": 636
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7327387061694113e-05,
      "loss": 0.7953,
      "step": 637
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7317982779281153e-05,
      "loss": 0.8221,
      "step": 638
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7308564541919066e-05,
      "loss": 0.8181,
      "step": 639
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7299132367567856e-05,
      "loss": 0.8119,
      "step": 640
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7289686274214116e-05,
      "loss": 0.8148,
      "step": 641
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7280226279870967e-05,
      "loss": 0.8062,
      "step": 642
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.727075240257805e-05,
      "loss": 0.8343,
      "step": 643
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7261264660401463e-05,
      "loss": 0.821,
      "step": 644
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7251763071433767e-05,
      "loss": 0.8238,
      "step": 645
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7242247653793908e-05,
      "loss": 0.816,
      "step": 646
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7232718425627213e-05,
      "loss": 0.8181,
      "step": 647
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.722317540510534e-05,
      "loss": 0.7972,
      "step": 648
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7213618610426253e-05,
      "loss": 0.8433,
      "step": 649
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7204048059814175e-05,
      "loss": 0.8202,
      "step": 650
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7194463771519572e-05,
      "loss": 0.8497,
      "step": 651
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.7184865763819095e-05,
      "loss": 0.7878,
      "step": 652
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.717525405501556e-05,
      "loss": 0.8362,
      "step": 653
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.7165628663437923e-05,
      "loss": 0.8205,
      "step": 654
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.715598960744121e-05,
      "loss": 0.814,
      "step": 655
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.7146336905406525e-05,
      "loss": 0.8172,
      "step": 656
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.7136670575740975e-05,
      "loss": 0.7825,
      "step": 657
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.712699063687768e-05,
      "loss": 0.8123,
      "step": 658
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.7117297107275677e-05,
      "loss": 0.8342,
      "step": 659
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.710759000541995e-05,
      "loss": 0.8102,
      "step": 660
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.709786934982135e-05,
      "loss": 0.8186,
      "step": 661
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.7088135159016584e-05,
      "loss": 0.8263,
      "step": 662
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.7078387451568152e-05,
      "loss": 0.7968,
      "step": 663
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.706862624606435e-05,
      "loss": 0.8249,
      "step": 664
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.7058851561119198e-05,
      "loss": 0.8192,
      "step": 665
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.7049063415372433e-05,
      "loss": 0.7872,
      "step": 666
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.7039261827489452e-05,
      "loss": 0.8199,
      "step": 667
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.7029446816161286e-05,
      "loss": 0.8267,
      "step": 668
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.7019618400104572e-05,
      "loss": 0.7915,
      "step": 669
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.7009776598061496e-05,
      "loss": 0.8399,
      "step": 670
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.699992142879978e-05,
      "loss": 0.8138,
      "step": 671
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.6990052911112637e-05,
      "loss": 0.8145,
      "step": 672
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.698017106381873e-05,
      "loss": 0.8273,
      "step": 673
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.6970275905762143e-05,
      "loss": 0.8217,
      "step": 674
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.6960367455812336e-05,
      "loss": 0.7787,
      "step": 675
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.695044573286413e-05,
      "loss": 0.8073,
      "step": 676
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.6940510755837642e-05,
      "loss": 0.8111,
      "step": 677
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.6930562543678275e-05,
      "loss": 0.8345,
      "step": 678
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.6920601115356663e-05,
      "loss": 0.8027,
      "step": 679
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.691062648986865e-05,
      "loss": 0.7931,
      "step": 680
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.6900638686235234e-05,
      "loss": 0.8244,
      "step": 681
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.6890637723502555e-05,
      "loss": 0.8064,
      "step": 682
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.6880623620741843e-05,
      "loss": 0.8489,
      "step": 683
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.687059639704938e-05,
      "loss": 0.8086,
      "step": 684
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.686055607154648e-05,
      "loss": 0.8118,
      "step": 685
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.6850502663379433e-05,
      "loss": 0.8009,
      "step": 686
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.684043619171947e-05,
      "loss": 0.8319,
      "step": 687
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.683035667576275e-05,
      "loss": 0.8319,
      "step": 688
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.6820264134730298e-05,
      "loss": 0.8077,
      "step": 689
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.6810158587867973e-05,
      "loss": 0.8124,
      "step": 690
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.6800040054446442e-05,
      "loss": 0.8113,
      "step": 691
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.6789908553761137e-05,
      "loss": 0.7904,
      "step": 692
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.677976410513221e-05,
      "loss": 0.812,
      "step": 693
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.676960672790451e-05,
      "loss": 0.8399,
      "step": 694
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.6759436441447544e-05,
      "loss": 0.8065,
      "step": 695
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.6749253265155428e-05,
      "loss": 0.8095,
      "step": 696
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.673905721844686e-05,
      "loss": 0.8121,
      "step": 697
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.672884832076508e-05,
      "loss": 0.8173,
      "step": 698
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.6718626591577843e-05,
      "loss": 0.7973,
      "step": 699
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6708392050377365e-05,
      "loss": 0.8268,
      "step": 700
    },
    {
      "epoch": 0.89,
      "eval_loss": 0.8047187328338623,
      "eval_runtime": 183.1517,
      "eval_samples_per_second": 11.269,
      "eval_steps_per_second": 0.093,
      "step": 700
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6698144716680288e-05,
      "loss": 0.7737,
      "step": 701
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6687884610027666e-05,
      "loss": 0.8033,
      "step": 702
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6677611749984894e-05,
      "loss": 0.8156,
      "step": 703
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.666732615614169e-05,
      "loss": 0.7893,
      "step": 704
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6657027848112064e-05,
      "loss": 0.8083,
      "step": 705
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.664671684553426e-05,
      "loss": 0.8479,
      "step": 706
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6636393168070733e-05,
      "loss": 0.8,
      "step": 707
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.6626056835408114e-05,
      "loss": 0.8156,
      "step": 708
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.6615707867257163e-05,
      "loss": 0.7859,
      "step": 709
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.660534628335273e-05,
      "loss": 0.8284,
      "step": 710
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.6594972103453727e-05,
      "loss": 0.796,
      "step": 711
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.6584585347343085e-05,
      "loss": 0.7997,
      "step": 712
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.657418603482772e-05,
      "loss": 0.8228,
      "step": 713
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.6563774185738494e-05,
      "loss": 0.8067,
      "step": 714
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.6553349819930167e-05,
      "loss": 0.814,
      "step": 715
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.654291295728137e-05,
      "loss": 0.8114,
      "step": 716
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.653246361769457e-05,
      "loss": 0.8099,
      "step": 717
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.652200182109602e-05,
      "loss": 0.7965,
      "step": 718
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6511527587435736e-05,
      "loss": 0.7943,
      "step": 719
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6501040936687444e-05,
      "loss": 0.8196,
      "step": 720
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.649054188884855e-05,
      "loss": 0.8067,
      "step": 721
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.64800304639401e-05,
      "loss": 0.7846,
      "step": 722
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.646950668200675e-05,
      "loss": 0.8062,
      "step": 723
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6458970563116703e-05,
      "loss": 0.812,
      "step": 724
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6448422127361707e-05,
      "loss": 0.803,
      "step": 725
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6437861394856986e-05,
      "loss": 0.8106,
      "step": 726
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6427288385741218e-05,
      "loss": 0.7794,
      "step": 727
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6416703120176485e-05,
      "loss": 0.7924,
      "step": 728
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.640610561834825e-05,
      "loss": 0.814,
      "step": 729
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6395495900465306e-05,
      "loss": 0.8098,
      "step": 730
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.6384873986759742e-05,
      "loss": 0.8238,
      "step": 731
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.63742398974869e-05,
      "loss": 0.8037,
      "step": 732
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.6363593652925343e-05,
      "loss": 0.8148,
      "step": 733
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.6352935273376814e-05,
      "loss": 0.7857,
      "step": 734
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.63422647791662e-05,
      "loss": 0.817,
      "step": 735
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.633158219064148e-05,
      "loss": 0.8083,
      "step": 736
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.632088752817371e-05,
      "loss": 0.8124,
      "step": 737
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.6310180812156952e-05,
      "loss": 0.8102,
      "step": 738
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6299462063008272e-05,
      "loss": 0.8065,
      "step": 739
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6288731301167667e-05,
      "loss": 0.8165,
      "step": 740
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.627798854709806e-05,
      "loss": 0.8221,
      "step": 741
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.626723382128522e-05,
      "loss": 0.8062,
      "step": 742
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.625646714423776e-05,
      "loss": 0.8112,
      "step": 743
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6245688536487084e-05,
      "loss": 0.824,
      "step": 744
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6234898018587336e-05,
      "loss": 0.8151,
      "step": 745
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6224095611115385e-05,
      "loss": 0.8081,
      "step": 746
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.621328133467076e-05,
      "loss": 0.8039,
      "step": 747
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.620245520987564e-05,
      "loss": 0.8088,
      "step": 748
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6191617257374782e-05,
      "loss": 0.8003,
      "step": 749
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6180767497835503e-05,
      "loss": 0.765,
      "step": 750
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6169905951947637e-05,
      "loss": 0.7972,
      "step": 751
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.615903264042349e-05,
      "loss": 0.7952,
      "step": 752
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6148147583997813e-05,
      "loss": 0.7766,
      "step": 753
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6137250803427744e-05,
      "loss": 0.8248,
      "step": 754
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.6126342319492784e-05,
      "loss": 0.7834,
      "step": 755
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.6115422152994752e-05,
      "loss": 0.7866,
      "step": 756
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.6104490324757742e-05,
      "loss": 0.8078,
      "step": 757
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.6093546855628085e-05,
      "loss": 0.7777,
      "step": 758
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.608259176647431e-05,
      "loss": 0.7921,
      "step": 759
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.6071625078187113e-05,
      "loss": 0.7942,
      "step": 760
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.6060646811679302e-05,
      "loss": 0.7936,
      "step": 761
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.6049656987885764e-05,
      "loss": 0.8234,
      "step": 762
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.6038655627763425e-05,
      "loss": 0.7969,
      "step": 763
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.602764275229121e-05,
      "loss": 0.7871,
      "step": 764
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.6016618382470014e-05,
      "loss": 0.7975,
      "step": 765
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.6005582539322627e-05,
      "loss": 0.8316,
      "step": 766
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.5994535243893742e-05,
      "loss": 0.825,
      "step": 767
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.598347651724988e-05,
      "loss": 0.7838,
      "step": 768
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.5972406380479356e-05,
      "loss": 0.7842,
      "step": 769
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.5961324854692254e-05,
      "loss": 0.7604,
      "step": 770
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.5950231961020373e-05,
      "loss": 0.7944,
      "step": 771
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.593912772061718e-05,
      "loss": 0.7891,
      "step": 772
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.59280121546578e-05,
      "loss": 0.7909,
      "step": 773
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.5916885284338937e-05,
      "loss": 0.7857,
      "step": 774
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.5905747130878853e-05,
      "loss": 0.8019,
      "step": 775
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.5894597715517334e-05,
      "loss": 0.7941,
      "step": 776
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.588343705951564e-05,
      "loss": 0.8191,
      "step": 777
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.5872265184156463e-05,
      "loss": 0.7926,
      "step": 778
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.5861082110743897e-05,
      "loss": 0.7693,
      "step": 779
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.5849887860603374e-05,
      "loss": 0.769,
      "step": 780
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.5838682455081657e-05,
      "loss": 0.8112,
      "step": 781
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.582746591554677e-05,
      "loss": 0.7894,
      "step": 782
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.581623826338798e-05,
      "loss": 0.8109,
      "step": 783
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.5804999520015735e-05,
      "loss": 0.8017,
      "step": 784
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.5793749706861637e-05,
      "loss": 0.7951,
      "step": 785
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.5782488845378394e-05,
      "loss": 0.8209,
      "step": 786
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.577121695703979e-05,
      "loss": 0.7979,
      "step": 787
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5759934063340627e-05,
      "loss": 0.8093,
      "step": 788
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.57486401857967e-05,
      "loss": 0.7818,
      "step": 789
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5737335345944758e-05,
      "loss": 0.774,
      "step": 790
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.572601956534243e-05,
      "loss": 0.8038,
      "step": 791
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5714692865568232e-05,
      "loss": 0.7597,
      "step": 792
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5703355268221493e-05,
      "loss": 0.7535,
      "step": 793
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5692006794922316e-05,
      "loss": 0.7738,
      "step": 794
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.568064746731156e-05,
      "loss": 0.7732,
      "step": 795
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5669277307050768e-05,
      "loss": 0.7706,
      "step": 796
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5657896335822147e-05,
      "loss": 0.7666,
      "step": 797
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.564650457532852e-05,
      "loss": 0.7863,
      "step": 798
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.563510204729328e-05,
      "loss": 0.7849,
      "step": 799
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5623688773460358e-05,
      "loss": 0.7739,
      "step": 800
    },
    {
      "epoch": 1.01,
      "eval_loss": 0.7900829315185547,
      "eval_runtime": 182.5273,
      "eval_samples_per_second": 11.308,
      "eval_steps_per_second": 0.093,
      "step": 800
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5612264775594176e-05,
      "loss": 0.7849,
      "step": 801
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5600830075479604e-05,
      "loss": 0.7509,
      "step": 802
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5589384694921917e-05,
      "loss": 0.7665,
      "step": 803
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5577928655746767e-05,
      "loss": 0.7695,
      "step": 804
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.556646197980012e-05,
      "loss": 0.7771,
      "step": 805
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.555498468894824e-05,
      "loss": 0.78,
      "step": 806
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5543496805077617e-05,
      "loss": 0.7818,
      "step": 807
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5531998350094954e-05,
      "loss": 0.7714,
      "step": 808
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5520489345927095e-05,
      "loss": 0.7885,
      "step": 809
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5508969814521026e-05,
      "loss": 0.7889,
      "step": 810
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.549743977784379e-05,
      "loss": 0.7711,
      "step": 811
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5485899257882465e-05,
      "loss": 0.7903,
      "step": 812
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.547434827664413e-05,
      "loss": 0.7659,
      "step": 813
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5462786856155792e-05,
      "loss": 0.765,
      "step": 814
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5451215018464386e-05,
      "loss": 0.7952,
      "step": 815
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5439632785636707e-05,
      "loss": 0.7554,
      "step": 816
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5428040179759366e-05,
      "loss": 0.7888,
      "step": 817
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.541643722293876e-05,
      "loss": 0.7713,
      "step": 818
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.5404823937301026e-05,
      "loss": 0.7768,
      "step": 819
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.5393200344991993e-05,
      "loss": 0.7875,
      "step": 820
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.5381566468177145e-05,
      "loss": 0.7744,
      "step": 821
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.5369922329041583e-05,
      "loss": 0.7614,
      "step": 822
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.5358267949789968e-05,
      "loss": 0.7756,
      "step": 823
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.53466033526465e-05,
      "loss": 0.763,
      "step": 824
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.533492855985485e-05,
      "loss": 0.7659,
      "step": 825
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5323243593678153e-05,
      "loss": 0.7845,
      "step": 826
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5311548476398918e-05,
      "loss": 0.7629,
      "step": 827
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5299843230319026e-05,
      "loss": 0.7819,
      "step": 828
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5288127877759675e-05,
      "loss": 0.773,
      "step": 829
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.527640244106133e-05,
      "loss": 0.7518,
      "step": 830
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5264666942583685e-05,
      "loss": 0.7505,
      "step": 831
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5252921404705627e-05,
      "loss": 0.748,
      "step": 832
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5241165849825186e-05,
      "loss": 0.8051,
      "step": 833
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5229400300359483e-05,
      "loss": 0.7506,
      "step": 834
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5217624778744718e-05,
      "loss": 0.7694,
      "step": 835
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5205839307436088e-05,
      "loss": 0.7799,
      "step": 836
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5194043908907774e-05,
      "loss": 0.771,
      "step": 837
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5182238605652883e-05,
      "loss": 0.7488,
      "step": 838
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5170423420183422e-05,
      "loss": 0.7618,
      "step": 839
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5158598375030218e-05,
      "loss": 0.7891,
      "step": 840
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5146763492742926e-05,
      "loss": 0.7313,
      "step": 841
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.5134918795889944e-05,
      "loss": 0.7416,
      "step": 842
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.512306430705839e-05,
      "loss": 0.7794,
      "step": 843
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.5111200048854055e-05,
      "loss": 0.7859,
      "step": 844
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.5099326043901361e-05,
      "loss": 0.7879,
      "step": 845
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.5087442314843312e-05,
      "loss": 0.7692,
      "step": 846
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.5075548884341462e-05,
      "loss": 0.7767,
      "step": 847
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.506364577507586e-05,
      "loss": 0.769,
      "step": 848
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.5051733009745013e-05,
      "loss": 0.779,
      "step": 849
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.503981061106584e-05,
      "loss": 0.7766,
      "step": 850
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.5027878601773633e-05,
      "loss": 0.7577,
      "step": 851
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.501593700462201e-05,
      "loss": 0.766,
      "step": 852
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.5003985842382873e-05,
      "loss": 0.7562,
      "step": 853
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.499202513784636e-05,
      "loss": 0.778,
      "step": 854
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.4980054913820814e-05,
      "loss": 0.7684,
      "step": 855
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.496807519313272e-05,
      "loss": 0.7889,
      "step": 856
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.4956085998626684e-05,
      "loss": 0.7802,
      "step": 857
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.494408735316537e-05,
      "loss": 0.7812,
      "step": 858
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.4932079279629464e-05,
      "loss": 0.7617,
      "step": 859
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.4920061800917637e-05,
      "loss": 0.755,
      "step": 860
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.4908034939946493e-05,
      "loss": 0.7835,
      "step": 861
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.4895998719650526e-05,
      "loss": 0.7425,
      "step": 862
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.4883953162982071e-05,
      "loss": 0.75,
      "step": 863
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.4871898292911285e-05,
      "loss": 0.7697,
      "step": 864
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.485983413242606e-05,
      "loss": 0.7738,
      "step": 865
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.4847760704532033e-05,
      "loss": 0.7619,
      "step": 866
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.4835678032252484e-05,
      "loss": 0.7453,
      "step": 867
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.482358613862835e-05,
      "loss": 0.7358,
      "step": 868
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.4811485046718124e-05,
      "loss": 0.7656,
      "step": 869
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.4799374779597866e-05,
      "loss": 0.7715,
      "step": 870
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.4787255360361113e-05,
      "loss": 0.7417,
      "step": 871
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.4775126812118865e-05,
      "loss": 0.7596,
      "step": 872
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.476298915799953e-05,
      "loss": 0.7569,
      "step": 873
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4750842421148872e-05,
      "loss": 0.7708,
      "step": 874
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4738686624729987e-05,
      "loss": 0.7964,
      "step": 875
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4726521791923238e-05,
      "loss": 0.7718,
      "step": 876
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4714347945926223e-05,
      "loss": 0.7734,
      "step": 877
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4702165109953732e-05,
      "loss": 0.7766,
      "step": 878
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4689973307237687e-05,
      "loss": 0.7632,
      "step": 879
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4677772561027121e-05,
      "loss": 0.7507,
      "step": 880
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4665562894588115e-05,
      "loss": 0.76,
      "step": 881
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.465334433120376e-05,
      "loss": 0.7605,
      "step": 882
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.464111689417412e-05,
      "loss": 0.7621,
      "step": 883
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4628880606816172e-05,
      "loss": 0.7672,
      "step": 884
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4616635492463775e-05,
      "loss": 0.7584,
      "step": 885
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4604381574467616e-05,
      "loss": 0.7493,
      "step": 886
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4592118876195178e-05,
      "loss": 0.746,
      "step": 887
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4579847421030677e-05,
      "loss": 0.7882,
      "step": 888
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4567567232375038e-05,
      "loss": 0.7499,
      "step": 889
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4555278333645833e-05,
      "loss": 0.7719,
      "step": 890
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4542980748277247e-05,
      "loss": 0.7706,
      "step": 891
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.453067449972003e-05,
      "loss": 0.7844,
      "step": 892
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4518359611441452e-05,
      "loss": 0.7518,
      "step": 893
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4506036106925255e-05,
      "loss": 0.7936,
      "step": 894
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4493704009671614e-05,
      "loss": 0.7745,
      "step": 895
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4481363343197095e-05,
      "loss": 0.7712,
      "step": 896
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4469014131034594e-05,
      "loss": 0.7607,
      "step": 897
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4456656396733313e-05,
      "loss": 0.7539,
      "step": 898
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.44442901638587e-05,
      "loss": 0.7662,
      "step": 899
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4431915455992416e-05,
      "loss": 0.7613,
      "step": 900
    },
    {
      "epoch": 1.14,
      "eval_loss": 0.7758861184120178,
      "eval_runtime": 182.8985,
      "eval_samples_per_second": 11.285,
      "eval_steps_per_second": 0.093,
      "step": 900
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4419532296732271e-05,
      "loss": 0.7746,
      "step": 901
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.44071407096922e-05,
      "loss": 0.742,
      "step": 902
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4394740718502207e-05,
      "loss": 0.746,
      "step": 903
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4382332346808325e-05,
      "loss": 0.7745,
      "step": 904
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4369915618272568e-05,
      "loss": 0.7632,
      "step": 905
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4357490556572877e-05,
      "loss": 0.7564,
      "step": 906
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.43450571854031e-05,
      "loss": 0.7651,
      "step": 907
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4332615528472912e-05,
      "loss": 0.7467,
      "step": 908
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4320165609507804e-05,
      "loss": 0.7652,
      "step": 909
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4307707452249013e-05,
      "loss": 0.7398,
      "step": 910
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.429524108045349e-05,
      "loss": 0.7462,
      "step": 911
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4282766517893852e-05,
      "loss": 0.7848,
      "step": 912
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.4270283788358328e-05,
      "loss": 0.7419,
      "step": 913
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.4257792915650728e-05,
      "loss": 0.7577,
      "step": 914
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.424529392359039e-05,
      "loss": 0.7628,
      "step": 915
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.423278683601213e-05,
      "loss": 0.7584,
      "step": 916
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.422027167676621e-05,
      "loss": 0.7278,
      "step": 917
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.4207748469718276e-05,
      "loss": 0.7668,
      "step": 918
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.4195217238749325e-05,
      "loss": 0.7403,
      "step": 919
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.4182678007755653e-05,
      "loss": 0.7717,
      "step": 920
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.4170130800648814e-05,
      "loss": 0.7553,
      "step": 921
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.4157575641355568e-05,
      "loss": 0.7401,
      "step": 922
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.4145012553817848e-05,
      "loss": 0.7658,
      "step": 923
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.4132441561992689e-05,
      "loss": 0.7383,
      "step": 924
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.4119862689852224e-05,
      "loss": 0.7723,
      "step": 925
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.4107275961383584e-05,
      "loss": 0.7615,
      "step": 926
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.4094681400588908e-05,
      "loss": 0.7772,
      "step": 927
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.4082079031485253e-05,
      "loss": 0.7254,
      "step": 928
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.4069468878104573e-05,
      "loss": 0.7674,
      "step": 929
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.4056850964493668e-05,
      "loss": 0.7505,
      "step": 930
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.4044225314714131e-05,
      "loss": 0.7557,
      "step": 931
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.4031591952842313e-05,
      "loss": 0.7549,
      "step": 932
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.4018950902969262e-05,
      "loss": 0.7286,
      "step": 933
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.4006302189200703e-05,
      "loss": 0.7297,
      "step": 934
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.3993645835656955e-05,
      "loss": 0.7435,
      "step": 935
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.398098186647293e-05,
      "loss": 0.7407,
      "step": 936
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.3968310305798032e-05,
      "loss": 0.7621,
      "step": 937
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.395563117779617e-05,
      "loss": 0.7527,
      "step": 938
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.3942944506645666e-05,
      "loss": 0.7835,
      "step": 939
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.3930250316539237e-05,
      "loss": 0.7252,
      "step": 940
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.3917548631683931e-05,
      "loss": 0.76,
      "step": 941
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.3904839476301091e-05,
      "loss": 0.7564,
      "step": 942
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.3892122874626301e-05,
      "loss": 0.7648,
      "step": 943
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.3879398850909355e-05,
      "loss": 0.762,
      "step": 944
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.3866667429414188e-05,
      "loss": 0.7161,
      "step": 945
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.3853928634418856e-05,
      "loss": 0.7736,
      "step": 946
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.3841182490215461e-05,
      "loss": 0.7609,
      "step": 947
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.382842902111013e-05,
      "loss": 0.7499,
      "step": 948
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.3815668251422953e-05,
      "loss": 0.7514,
      "step": 949
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.3802900205487948e-05,
      "loss": 0.7496,
      "step": 950
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.3790124907652996e-05,
      "loss": 0.7712,
      "step": 951
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.377734238227982e-05,
      "loss": 0.7489,
      "step": 952
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.376455265374392e-05,
      "loss": 0.7449,
      "step": 953
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3751755746434533e-05,
      "loss": 0.745,
      "step": 954
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3738951684754585e-05,
      "loss": 0.7592,
      "step": 955
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3726140493120639e-05,
      "loss": 0.7748,
      "step": 956
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3713322195962866e-05,
      "loss": 0.7517,
      "step": 957
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3700496817724978e-05,
      "loss": 0.7599,
      "step": 958
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3687664382864197e-05,
      "loss": 0.7511,
      "step": 959
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3674824915851193e-05,
      "loss": 0.7273,
      "step": 960
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3661978441170054e-05,
      "loss": 0.7682,
      "step": 961
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3649124983318224e-05,
      "loss": 0.7426,
      "step": 962
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3636264566806473e-05,
      "loss": 0.7512,
      "step": 963
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3623397216158826e-05,
      "loss": 0.7606,
      "step": 964
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3610522955912551e-05,
      "loss": 0.753,
      "step": 965
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3597641810618071e-05,
      "loss": 0.7263,
      "step": 966
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3584753804838955e-05,
      "loss": 0.7487,
      "step": 967
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3571858963151845e-05,
      "loss": 0.742,
      "step": 968
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3558957310146424e-05,
      "loss": 0.733,
      "step": 969
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3546048870425356e-05,
      "loss": 0.7557,
      "step": 970
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3533133668604262e-05,
      "loss": 0.7535,
      "step": 971
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3520211729311639e-05,
      "loss": 0.7436,
      "step": 972
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3507283077188844e-05,
      "loss": 0.7598,
      "step": 973
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3494347736890037e-05,
      "loss": 0.7613,
      "step": 974
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3481405733082118e-05,
      "loss": 0.7326,
      "step": 975
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3468457090444712e-05,
      "loss": 0.7201,
      "step": 976
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3455501833670089e-05,
      "loss": 0.7451,
      "step": 977
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.344253998746314e-05,
      "loss": 0.7565,
      "step": 978
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3429571576541315e-05,
      "loss": 0.721,
      "step": 979
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3416596625634595e-05,
      "loss": 0.7432,
      "step": 980
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3403615159485415e-05,
      "loss": 0.7602,
      "step": 981
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3390627202848652e-05,
      "loss": 0.7363,
      "step": 982
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3377632780491545e-05,
      "loss": 0.759,
      "step": 983
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3364631917193671e-05,
      "loss": 0.7407,
      "step": 984
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3351624637746885e-05,
      "loss": 0.7591,
      "step": 985
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3338610966955286e-05,
      "loss": 0.7678,
      "step": 986
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3325590929635146e-05,
      "loss": 0.7424,
      "step": 987
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3312564550614894e-05,
      "loss": 0.708,
      "step": 988
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3299531854735042e-05,
      "loss": 0.7576,
      "step": 989
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3286492866848143e-05,
      "loss": 0.7543,
      "step": 990
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3273447611818768e-05,
      "loss": 0.7621,
      "step": 991
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.326039611452342e-05,
      "loss": 0.7499,
      "step": 992
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.3247338399850512e-05,
      "loss": 0.7477,
      "step": 993
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.3234274492700316e-05,
      "loss": 0.7408,
      "step": 994
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.3221204417984907e-05,
      "loss": 0.7921,
      "step": 995
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.3208128200628127e-05,
      "loss": 0.7526,
      "step": 996
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.319504586556553e-05,
      "loss": 0.7399,
      "step": 997
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.3181957437744333e-05,
      "loss": 0.741,
      "step": 998
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.3168862942123374e-05,
      "loss": 0.7143,
      "step": 999
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.3155762403673065e-05,
      "loss": 0.7469,
      "step": 1000
    },
    {
      "epoch": 1.27,
      "eval_loss": 0.7623376250267029,
      "eval_runtime": 182.3699,
      "eval_samples_per_second": 11.318,
      "eval_steps_per_second": 0.093,
      "step": 1000
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.3142655847375332e-05,
      "loss": 0.7479,
      "step": 1001
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.3129543298223583e-05,
      "loss": 0.7754,
      "step": 1002
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.311642478122266e-05,
      "loss": 0.738,
      "step": 1003
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.3103300321388775e-05,
      "loss": 0.7399,
      "step": 1004
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 0.7468,
      "step": 1005
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.3077033673343597e-05,
      "loss": 0.7598,
      "step": 1006
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.3063891535221208e-05,
      "loss": 0.744,
      "step": 1007
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.305074355444357e-05,
      "loss": 0.7312,
      "step": 1008
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.3037589756083085e-05,
      "loss": 0.743,
      "step": 1009
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.3024430165223245e-05,
      "loss": 0.7448,
      "step": 1010
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.3011264806958597e-05,
      "loss": 0.7469,
      "step": 1011
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.2998093706394676e-05,
      "loss": 0.7534,
      "step": 1012
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.2984916888647971e-05,
      "loss": 0.7368,
      "step": 1013
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.2971734378845877e-05,
      "loss": 0.7392,
      "step": 1014
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.2958546202126638e-05,
      "loss": 0.75,
      "step": 1015
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2945352383639311e-05,
      "loss": 0.7514,
      "step": 1016
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.29321529485437e-05,
      "loss": 0.7584,
      "step": 1017
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2918947922010336e-05,
      "loss": 0.7365,
      "step": 1018
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2905737329220394e-05,
      "loss": 0.7591,
      "step": 1019
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2892521195365679e-05,
      "loss": 0.736,
      "step": 1020
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2879299545648551e-05,
      "loss": 0.7416,
      "step": 1021
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2866072405281901e-05,
      "loss": 0.738,
      "step": 1022
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.2852839799489076e-05,
      "loss": 0.7218,
      "step": 1023
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2839601753503863e-05,
      "loss": 0.7419,
      "step": 1024
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2826358292570398e-05,
      "loss": 0.7259,
      "step": 1025
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2813109441943166e-05,
      "loss": 0.7539,
      "step": 1026
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2799855226886919e-05,
      "loss": 0.7378,
      "step": 1027
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2786595672676643e-05,
      "loss": 0.7425,
      "step": 1028
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2773330804597503e-05,
      "loss": 0.7539,
      "step": 1029
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2760060647944794e-05,
      "loss": 0.7322,
      "step": 1030
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.2746785228023904e-05,
      "loss": 0.7337,
      "step": 1031
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.273350457015025e-05,
      "loss": 0.7537,
      "step": 1032
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.2720218699649243e-05,
      "loss": 0.7882,
      "step": 1033
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.2706927641856228e-05,
      "loss": 0.7549,
      "step": 1034
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.2693631422116455e-05,
      "loss": 0.762,
      "step": 1035
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.2680330065784998e-05,
      "loss": 0.7404,
      "step": 1036
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.266702359822674e-05,
      "loss": 0.7179,
      "step": 1037
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.2653712044816308e-05,
      "loss": 0.72,
      "step": 1038
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.2640395430938029e-05,
      "loss": 0.7385,
      "step": 1039
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.262707378198587e-05,
      "loss": 0.7489,
      "step": 1040
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2613747123363418e-05,
      "loss": 0.7345,
      "step": 1041
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.260041548048379e-05,
      "loss": 0.7536,
      "step": 1042
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2587078878769632e-05,
      "loss": 0.7422,
      "step": 1043
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2573737343653026e-05,
      "loss": 0.7554,
      "step": 1044
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2560390900575472e-05,
      "loss": 0.7161,
      "step": 1045
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2547039574987829e-05,
      "loss": 0.7212,
      "step": 1046
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2533683392350264e-05,
      "loss": 0.7421,
      "step": 1047
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2520322378132207e-05,
      "loss": 0.7385,
      "step": 1048
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.25069565578123e-05,
      "loss": 0.7638,
      "step": 1049
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2493585956878354e-05,
      "loss": 0.7371,
      "step": 1050
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.248021060082729e-05,
      "loss": 0.7538,
      "step": 1051
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2466830515165102e-05,
      "loss": 0.7346,
      "step": 1052
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2453445725406805e-05,
      "loss": 0.7487,
      "step": 1053
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2440056257076376e-05,
      "loss": 0.743,
      "step": 1054
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.242666213570672e-05,
      "loss": 0.7537,
      "step": 1055
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2413263386839616e-05,
      "loss": 0.7243,
      "step": 1056
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.239986003602566e-05,
      "loss": 0.7259,
      "step": 1057
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2386452108824236e-05,
      "loss": 0.7448,
      "step": 1058
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.237303963080344e-05,
      "loss": 0.7491,
      "step": 1059
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2359622627540059e-05,
      "loss": 0.7287,
      "step": 1060
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2346201124619502e-05,
      "loss": 0.7559,
      "step": 1061
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2332775147635762e-05,
      "loss": 0.7351,
      "step": 1062
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.231934472219136e-05,
      "loss": 0.7267,
      "step": 1063
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.2305909873897305e-05,
      "loss": 0.7537,
      "step": 1064
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.229247062837304e-05,
      "loss": 0.7407,
      "step": 1065
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.2279027011246384e-05,
      "loss": 0.743,
      "step": 1066
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.2265579048153502e-05,
      "loss": 0.7356,
      "step": 1067
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.2252126764738845e-05,
      "loss": 0.7299,
      "step": 1068
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.2238670186655099e-05,
      "loss": 0.7192,
      "step": 1069
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.2225209339563144e-05,
      "loss": 0.7391,
      "step": 1070
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2211744249132001e-05,
      "loss": 0.7183,
      "step": 1071
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2198274941038773e-05,
      "loss": 0.7474,
      "step": 1072
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.218480144096862e-05,
      "loss": 0.7569,
      "step": 1073
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2171323774614686e-05,
      "loss": 0.7268,
      "step": 1074
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2157841967678064e-05,
      "loss": 0.7551,
      "step": 1075
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.214435604586774e-05,
      "loss": 0.7374,
      "step": 1076
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2130866034900552e-05,
      "loss": 0.7107,
      "step": 1077
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2117371960501128e-05,
      "loss": 0.7397,
      "step": 1078
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.2103873848401853e-05,
      "loss": 0.7434,
      "step": 1079
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.2090371724342804e-05,
      "loss": 0.748,
      "step": 1080
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.2076865614071718e-05,
      "loss": 0.7331,
      "step": 1081
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.2063355543343925e-05,
      "loss": 0.709,
      "step": 1082
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.2049841537922307e-05,
      "loss": 0.7099,
      "step": 1083
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.2036323623577253e-05,
      "loss": 0.7357,
      "step": 1084
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.2022801826086609e-05,
      "loss": 0.7236,
      "step": 1085
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.200927617123562e-05,
      "loss": 0.728,
      "step": 1086
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1995746684816892e-05,
      "loss": 0.7278,
      "step": 1087
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1982213392630331e-05,
      "loss": 0.7239,
      "step": 1088
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1968676320483103e-05,
      "loss": 0.7524,
      "step": 1089
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.195513549418959e-05,
      "loss": 0.7346,
      "step": 1090
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1941590939571317e-05,
      "loss": 0.7343,
      "step": 1091
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1928042682456936e-05,
      "loss": 0.7232,
      "step": 1092
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1914490748682146e-05,
      "loss": 0.7287,
      "step": 1093
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.1900935164089664e-05,
      "loss": 0.7348,
      "step": 1094
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1887375954529167e-05,
      "loss": 0.7428,
      "step": 1095
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.187381314585725e-05,
      "loss": 0.7294,
      "step": 1096
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1860246763937354e-05,
      "loss": 0.7299,
      "step": 1097
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1846676834639758e-05,
      "loss": 0.7259,
      "step": 1098
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1833103383841491e-05,
      "loss": 0.7199,
      "step": 1099
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1819526437426298e-05,
      "loss": 0.744,
      "step": 1100
    },
    {
      "epoch": 1.39,
      "eval_loss": 0.748179018497467,
      "eval_runtime": 182.3312,
      "eval_samples_per_second": 11.32,
      "eval_steps_per_second": 0.093,
      "step": 1100
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.1805946021284589e-05,
      "loss": 0.7125,
      "step": 1101
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.17923621613134e-05,
      "loss": 0.7321,
      "step": 1102
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1778774883416325e-05,
      "loss": 0.7338,
      "step": 1103
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1765184213503477e-05,
      "loss": 0.7461,
      "step": 1104
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1751590177491441e-05,
      "loss": 0.7223,
      "step": 1105
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1737992801303217e-05,
      "loss": 0.7341,
      "step": 1106
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1724392110868179e-05,
      "loss": 0.7187,
      "step": 1107
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1710788132122015e-05,
      "loss": 0.7387,
      "step": 1108
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.169718089100669e-05,
      "loss": 0.7401,
      "step": 1109
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1683570413470384e-05,
      "loss": 0.7197,
      "step": 1110
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.166995672546746e-05,
      "loss": 0.7233,
      "step": 1111
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1656339852958382e-05,
      "loss": 0.7153,
      "step": 1112
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1642719821909711e-05,
      "loss": 0.7363,
      "step": 1113
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1629096658294012e-05,
      "loss": 0.7107,
      "step": 1114
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1615470388089836e-05,
      "loss": 0.7134,
      "step": 1115
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1601841037281652e-05,
      "loss": 0.7403,
      "step": 1116
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1588208631859808e-05,
      "loss": 0.6933,
      "step": 1117
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1574573197820464e-05,
      "loss": 0.7202,
      "step": 1118
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1560934761165578e-05,
      "loss": 0.7269,
      "step": 1119
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1547293347902813e-05,
      "loss": 0.7402,
      "step": 1120
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.153364898404552e-05,
      "loss": 0.7192,
      "step": 1121
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1520001695612675e-05,
      "loss": 0.7512,
      "step": 1122
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1506351508628819e-05,
      "loss": 0.7237,
      "step": 1123
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1492698449124042e-05,
      "loss": 0.756,
      "step": 1124
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1479042543133895e-05,
      "loss": 0.7251,
      "step": 1125
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.146538381669936e-05,
      "loss": 0.7471,
      "step": 1126
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1451722295866805e-05,
      "loss": 0.7334,
      "step": 1127
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1438058006687918e-05,
      "loss": 0.7401,
      "step": 1128
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1424390975219666e-05,
      "loss": 0.7264,
      "step": 1129
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1410721227524256e-05,
      "loss": 0.7372,
      "step": 1130
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1397048789669061e-05,
      "loss": 0.7287,
      "step": 1131
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1383373687726595e-05,
      "loss": 0.735,
      "step": 1132
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1369695947774443e-05,
      "loss": 0.7328,
      "step": 1133
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1356015595895233e-05,
      "loss": 0.7102,
      "step": 1134
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1342332658176556e-05,
      "loss": 0.734,
      "step": 1135
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.132864716071095e-05,
      "loss": 0.7242,
      "step": 1136
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1314959129595828e-05,
      "loss": 0.7105,
      "step": 1137
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1301268590933434e-05,
      "loss": 0.7204,
      "step": 1138
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1287575570830794e-05,
      "loss": 0.7475,
      "step": 1139
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1273880095399667e-05,
      "loss": 0.7251,
      "step": 1140
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1260182190756492e-05,
      "loss": 0.7238,
      "step": 1141
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1246481883022347e-05,
      "loss": 0.7293,
      "step": 1142
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1232779198322881e-05,
      "loss": 0.7301,
      "step": 1143
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1219074162788285e-05,
      "loss": 0.7222,
      "step": 1144
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1205366802553231e-05,
      "loss": 0.706,
      "step": 1145
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1191657143756825e-05,
      "loss": 0.7074,
      "step": 1146
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1177945212542552e-05,
      "loss": 0.7214,
      "step": 1147
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1164231035058228e-05,
      "loss": 0.733,
      "step": 1148
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1150514637455969e-05,
      "loss": 0.7408,
      "step": 1149
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.1136796045892102e-05,
      "loss": 0.7819,
      "step": 1150
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.1123075286527153e-05,
      "loss": 0.7223,
      "step": 1151
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.1109352385525782e-05,
      "loss": 0.7052,
      "step": 1152
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.1095627369056722e-05,
      "loss": 0.7167,
      "step": 1153
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.1081900263292746e-05,
      "loss": 0.7308,
      "step": 1154
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.1068171094410618e-05,
      "loss": 0.7572,
      "step": 1155
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.1054439888591023e-05,
      "loss": 0.7121,
      "step": 1156
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.1040706672018548e-05,
      "loss": 0.736,
      "step": 1157
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.1026971470881589e-05,
      "loss": 0.7371,
      "step": 1158
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.1013234311372353e-05,
      "loss": 0.721,
      "step": 1159
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0999495219686762e-05,
      "loss": 0.717,
      "step": 1160
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0985754222024437e-05,
      "loss": 0.7218,
      "step": 1161
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.097201134458862e-05,
      "loss": 0.7256,
      "step": 1162
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.095826661358615e-05,
      "loss": 0.7505,
      "step": 1163
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0944520055227395e-05,
      "loss": 0.7482,
      "step": 1164
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.0930771695726201e-05,
      "loss": 0.7114,
      "step": 1165
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0917021561299864e-05,
      "loss": 0.718,
      "step": 1166
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0903269678169046e-05,
      "loss": 0.7132,
      "step": 1167
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0889516072557764e-05,
      "loss": 0.6975,
      "step": 1168
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0875760770693307e-05,
      "loss": 0.7084,
      "step": 1169
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0862003798806195e-05,
      "loss": 0.7279,
      "step": 1170
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0848245183130143e-05,
      "loss": 0.7269,
      "step": 1171
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0834484949901997e-05,
      "loss": 0.7084,
      "step": 1172
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.0820723125361685e-05,
      "loss": 0.7164,
      "step": 1173
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0806959735752174e-05,
      "loss": 0.7236,
      "step": 1174
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.079319480731941e-05,
      "loss": 0.7241,
      "step": 1175
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0779428366312277e-05,
      "loss": 0.72,
      "step": 1176
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0765660438982541e-05,
      "loss": 0.7156,
      "step": 1177
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0751891051584811e-05,
      "loss": 0.7303,
      "step": 1178
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0738120230376463e-05,
      "loss": 0.7302,
      "step": 1179
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.0724348001617626e-05,
      "loss": 0.7191,
      "step": 1180
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.07105743915711e-05,
      "loss": 0.7281,
      "step": 1181
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0696799426502325e-05,
      "loss": 0.7606,
      "step": 1182
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0683023132679323e-05,
      "loss": 0.7136,
      "step": 1183
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0669245536372652e-05,
      "loss": 0.6964,
      "step": 1184
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0655466663855349e-05,
      "loss": 0.7263,
      "step": 1185
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.064168654140289e-05,
      "loss": 0.733,
      "step": 1186
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0627905195293135e-05,
      "loss": 0.7209,
      "step": 1187
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0614122651806272e-05,
      "loss": 0.7426,
      "step": 1188
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0600338937224773e-05,
      "loss": 0.7079,
      "step": 1189
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0586554077833346e-05,
      "loss": 0.7459,
      "step": 1190
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0572768099918887e-05,
      "loss": 0.723,
      "step": 1191
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0558981029770414e-05,
      "loss": 0.7418,
      "step": 1192
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0545192893679036e-05,
      "loss": 0.7265,
      "step": 1193
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0531403717937888e-05,
      "loss": 0.7135,
      "step": 1194
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0517613528842096e-05,
      "loss": 0.6953,
      "step": 1195
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0503822352688708e-05,
      "loss": 0.7313,
      "step": 1196
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.049003021577667e-05,
      "loss": 0.7184,
      "step": 1197
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.047623714440674e-05,
      "loss": 0.7373,
      "step": 1198
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0462443164881474e-05,
      "loss": 0.7194,
      "step": 1199
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.044864830350515e-05,
      "loss": 0.7064,
      "step": 1200
    },
    {
      "epoch": 1.52,
      "eval_loss": 0.7351406812667847,
      "eval_runtime": 182.402,
      "eval_samples_per_second": 11.316,
      "eval_steps_per_second": 0.093,
      "step": 1200
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0434852586583737e-05,
      "loss": 0.7323,
      "step": 1201
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0421056040424824e-05,
      "loss": 0.7154,
      "step": 1202
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0407258691337593e-05,
      "loss": 0.7391,
      "step": 1203
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.039346056563275e-05,
      "loss": 0.7566,
      "step": 1204
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0379661689622477e-05,
      "loss": 0.7121,
      "step": 1205
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0365862089620402e-05,
      "loss": 0.7293,
      "step": 1206
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0352061791941522e-05,
      "loss": 0.7238,
      "step": 1207
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0338260822902166e-05,
      "loss": 0.6738,
      "step": 1208
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0324459208819946e-05,
      "loss": 0.735,
      "step": 1209
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0310656976013704e-05,
      "loss": 0.7386,
      "step": 1210
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0296854150803454e-05,
      "loss": 0.7118,
      "step": 1211
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.0283050759510356e-05,
      "loss": 0.6867,
      "step": 1212
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.026924682845663e-05,
      "loss": 0.7208,
      "step": 1213
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.0255442383965539e-05,
      "loss": 0.7137,
      "step": 1214
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.0241637452361323e-05,
      "loss": 0.706,
      "step": 1215
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.022783205996915e-05,
      "loss": 0.716,
      "step": 1216
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.021402623311506e-05,
      "loss": 0.7124,
      "step": 1217
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.0200219998125928e-05,
      "loss": 0.697,
      "step": 1218
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.018641338132941e-05,
      "loss": 0.7291,
      "step": 1219
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.0172606409053887e-05,
      "loss": 0.705,
      "step": 1220
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.0158799107628415e-05,
      "loss": 0.7263,
      "step": 1221
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.0144991503382676e-05,
      "loss": 0.7375,
      "step": 1222
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.013118362264694e-05,
      "loss": 0.7221,
      "step": 1223
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.0117375491751995e-05,
      "loss": 0.7088,
      "step": 1224
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.0103567137029111e-05,
      "loss": 0.7258,
      "step": 1225
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.008975858480998e-05,
      "loss": 0.7173,
      "step": 1226
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.0075949861426677e-05,
      "loss": 0.7117,
      "step": 1227
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.006214099321159e-05,
      "loss": 0.7228,
      "step": 1228
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.0048332006497406e-05,
      "loss": 0.7168,
      "step": 1229
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.0034522927617014e-05,
      "loss": 0.6837,
      "step": 1230
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.0020713782903494e-05,
      "loss": 0.6961,
      "step": 1231
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.0006904598690048e-05,
      "loss": 0.7152,
      "step": 1232
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.993095401309955e-06,
      "loss": 0.7093,
      "step": 1233
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.979286217096506e-06,
      "loss": 0.7227,
      "step": 1234
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.965477072382989e-06,
      "loss": 0.7142,
      "step": 1235
    },
    {
      "epoch": 1.56,
      "learning_rate": 9.951667993502599e-06,
      "loss": 0.7282,
      "step": 1236
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.937859006788411e-06,
      "loss": 0.719,
      "step": 1237
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.924050138573328e-06,
      "loss": 0.6805,
      "step": 1238
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.910241415190022e-06,
      "loss": 0.7148,
      "step": 1239
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.896432862970892e-06,
      "loss": 0.6954,
      "step": 1240
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.882624508248005e-06,
      "loss": 0.7165,
      "step": 1241
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.868816377353063e-06,
      "loss": 0.7193,
      "step": 1242
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.855008496617326e-06,
      "loss": 0.7205,
      "step": 1243
    },
    {
      "epoch": 1.57,
      "learning_rate": 9.84120089237159e-06,
      "loss": 0.7183,
      "step": 1244
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.827393590946116e-06,
      "loss": 0.7305,
      "step": 1245
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.81358661867059e-06,
      "loss": 0.7155,
      "step": 1246
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.799780001874073e-06,
      "loss": 0.7175,
      "step": 1247
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.785973766884947e-06,
      "loss": 0.7184,
      "step": 1248
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.772167940030854e-06,
      "loss": 0.7188,
      "step": 1249
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.75836254763868e-06,
      "loss": 0.7132,
      "step": 1250
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.744557616034461e-06,
      "loss": 0.7258,
      "step": 1251
    },
    {
      "epoch": 1.58,
      "learning_rate": 9.730753171543374e-06,
      "loss": 0.6931,
      "step": 1252
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.716949240489651e-06,
      "loss": 0.7204,
      "step": 1253
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.703145849196547e-06,
      "loss": 0.7304,
      "step": 1254
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.689343023986303e-06,
      "loss": 0.7231,
      "step": 1255
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.675540791180056e-06,
      "loss": 0.7346,
      "step": 1256
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.661739177097836e-06,
      "loss": 0.6861,
      "step": 1257
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.64793820805848e-06,
      "loss": 0.6853,
      "step": 1258
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.6341379103796e-06,
      "loss": 0.7198,
      "step": 1259
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.620338310377526e-06,
      "loss": 0.7146,
      "step": 1260
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.606539434367254e-06,
      "loss": 0.6897,
      "step": 1261
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.59274130866241e-06,
      "loss": 0.7119,
      "step": 1262
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.578943959575176e-06,
      "loss": 0.7237,
      "step": 1263
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.565147413416266e-06,
      "loss": 0.7171,
      "step": 1264
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.551351696494854e-06,
      "loss": 0.7199,
      "step": 1265
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.537556835118529e-06,
      "loss": 0.7483,
      "step": 1266
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.523762855593265e-06,
      "loss": 0.7007,
      "step": 1267
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.509969784223332e-06,
      "loss": 0.7108,
      "step": 1268
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.496177647311294e-06,
      "loss": 0.7067,
      "step": 1269
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.482386471157905e-06,
      "loss": 0.7255,
      "step": 1270
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.468596282062114e-06,
      "loss": 0.678,
      "step": 1271
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.454807106320969e-06,
      "loss": 0.7041,
      "step": 1272
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.441018970229587e-06,
      "loss": 0.7099,
      "step": 1273
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.427231900081115e-06,
      "loss": 0.7166,
      "step": 1274
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.413445922166654e-06,
      "loss": 0.7104,
      "step": 1275
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.399661062775229e-06,
      "loss": 0.7171,
      "step": 1276
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.385877348193733e-06,
      "loss": 0.7121,
      "step": 1277
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.372094804706867e-06,
      "loss": 0.7311,
      "step": 1278
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.358313458597111e-06,
      "loss": 0.7207,
      "step": 1279
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.344533336144653e-06,
      "loss": 0.7078,
      "step": 1280
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.330754463627351e-06,
      "loss": 0.7122,
      "step": 1281
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.316976867320682e-06,
      "loss": 0.7162,
      "step": 1282
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.303200573497676e-06,
      "loss": 0.7171,
      "step": 1283
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.289425608428905e-06,
      "loss": 0.7124,
      "step": 1284
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.275651998382377e-06,
      "loss": 0.7106,
      "step": 1285
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.26187976962354e-06,
      "loss": 0.715,
      "step": 1286
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.248108948415194e-06,
      "loss": 0.6732,
      "step": 1287
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.23433956101746e-06,
      "loss": 0.7149,
      "step": 1288
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.220571633687727e-06,
      "loss": 0.7269,
      "step": 1289
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.206805192680592e-06,
      "loss": 0.7036,
      "step": 1290
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.19304026424783e-06,
      "loss": 0.6922,
      "step": 1291
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.179276874638315e-06,
      "loss": 0.7178,
      "step": 1292
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.165515050098004e-06,
      "loss": 0.6752,
      "step": 1293
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.151754816869859e-06,
      "loss": 0.7144,
      "step": 1294
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.137996201193807e-06,
      "loss": 0.7049,
      "step": 1295
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.124239229306696e-06,
      "loss": 0.7273,
      "step": 1296
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.110483927442236e-06,
      "loss": 0.7307,
      "step": 1297
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.096730321830955e-06,
      "loss": 0.7256,
      "step": 1298
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.082978438700138e-06,
      "loss": 0.7057,
      "step": 1299
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.069228304273802e-06,
      "loss": 0.7274,
      "step": 1300
    },
    {
      "epoch": 1.65,
      "eval_loss": 0.7208890914916992,
      "eval_runtime": 182.6999,
      "eval_samples_per_second": 11.297,
      "eval_steps_per_second": 0.093,
      "step": 1300
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.05547994477261e-06,
      "loss": 0.7169,
      "step": 1301
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.041733386413851e-06,
      "loss": 0.7107,
      "step": 1302
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.027988655411383e-06,
      "loss": 0.7043,
      "step": 1303
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.014245777975565e-06,
      "loss": 0.6988,
      "step": 1304
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.00050478031324e-06,
      "loss": 0.6851,
      "step": 1305
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.986765688627652e-06,
      "loss": 0.6975,
      "step": 1306
    },
    {
      "epoch": 1.65,
      "learning_rate": 8.973028529118413e-06,
      "loss": 0.7067,
      "step": 1307
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.959293327981457e-06,
      "loss": 0.7015,
      "step": 1308
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.945560111408977e-06,
      "loss": 0.7114,
      "step": 1309
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.931828905589385e-06,
      "loss": 0.6976,
      "step": 1310
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.918099736707254e-06,
      "loss": 0.699,
      "step": 1311
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.904372630943281e-06,
      "loss": 0.6995,
      "step": 1312
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.890647614474223e-06,
      "loss": 0.7012,
      "step": 1313
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.876924713472845e-06,
      "loss": 0.7011,
      "step": 1314
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.863203954107902e-06,
      "loss": 0.693,
      "step": 1315
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.849485362544033e-06,
      "loss": 0.6902,
      "step": 1316
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.835768964941773e-06,
      "loss": 0.6958,
      "step": 1317
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.822054787457455e-06,
      "loss": 0.7189,
      "step": 1318
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.808342856243179e-06,
      "loss": 0.7105,
      "step": 1319
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.79463319744677e-06,
      "loss": 0.6984,
      "step": 1320
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.780925837211716e-06,
      "loss": 0.7007,
      "step": 1321
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.767220801677122e-06,
      "loss": 0.6691,
      "step": 1322
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.753518116977658e-06,
      "loss": 0.7009,
      "step": 1323
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.739817809243511e-06,
      "loss": 0.7172,
      "step": 1324
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.726119904600337e-06,
      "loss": 0.7069,
      "step": 1325
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.712424429169208e-06,
      "loss": 0.7064,
      "step": 1326
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.69873140906657e-06,
      "loss": 0.6897,
      "step": 1327
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.685040870404174e-06,
      "loss": 0.6957,
      "step": 1328
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.671352839289052e-06,
      "loss": 0.691,
      "step": 1329
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.657667341823449e-06,
      "loss": 0.7143,
      "step": 1330
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.64398440410477e-06,
      "loss": 0.7048,
      "step": 1331
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.63030405222556e-06,
      "loss": 0.7338,
      "step": 1332
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.616626312273407e-06,
      "loss": 0.7031,
      "step": 1333
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.602951210330942e-06,
      "loss": 0.6848,
      "step": 1334
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.58927877247575e-06,
      "loss": 0.685,
      "step": 1335
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.575609024780336e-06,
      "loss": 0.6786,
      "step": 1336
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.561941993312087e-06,
      "loss": 0.6867,
      "step": 1337
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.548277704133198e-06,
      "loss": 0.7041,
      "step": 1338
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.534616183300642e-06,
      "loss": 0.7043,
      "step": 1339
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.520957456866107e-06,
      "loss": 0.7039,
      "step": 1340
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.50730155087596e-06,
      "loss": 0.721,
      "step": 1341
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.493648491371183e-06,
      "loss": 0.7168,
      "step": 1342
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.479998304387329e-06,
      "loss": 0.692,
      "step": 1343
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.466351015954482e-06,
      "loss": 0.6991,
      "step": 1344
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.452706652097187e-06,
      "loss": 0.7128,
      "step": 1345
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.439065238834424e-06,
      "loss": 0.6832,
      "step": 1346
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.425426802179539e-06,
      "loss": 0.6758,
      "step": 1347
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.411791368140197e-06,
      "loss": 0.6684,
      "step": 1348
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.398158962718353e-06,
      "loss": 0.6828,
      "step": 1349
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.384529611910164e-06,
      "loss": 0.6769,
      "step": 1350
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.370903341705991e-06,
      "loss": 0.7197,
      "step": 1351
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.357280178090294e-06,
      "loss": 0.6815,
      "step": 1352
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.34366014704162e-06,
      "loss": 0.6664,
      "step": 1353
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.330043274532546e-06,
      "loss": 0.6683,
      "step": 1354
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.316429586529616e-06,
      "loss": 0.7058,
      "step": 1355
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.302819108993311e-06,
      "loss": 0.7026,
      "step": 1356
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.289211867877986e-06,
      "loss": 0.6987,
      "step": 1357
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.275607889131823e-06,
      "loss": 0.6935,
      "step": 1358
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.262007198696786e-06,
      "loss": 0.7131,
      "step": 1359
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.24840982250856e-06,
      "loss": 0.6888,
      "step": 1360
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.234815786496526e-06,
      "loss": 0.6913,
      "step": 1361
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.221225116583677e-06,
      "loss": 0.6808,
      "step": 1362
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.207637838686602e-06,
      "loss": 0.7214,
      "step": 1363
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.194053978715415e-06,
      "loss": 0.6873,
      "step": 1364
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.180473562573705e-06,
      "loss": 0.6878,
      "step": 1365
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.166896616158514e-06,
      "loss": 0.6886,
      "step": 1366
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.15332316536024e-06,
      "loss": 0.6785,
      "step": 1367
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.139753236062649e-06,
      "loss": 0.7007,
      "step": 1368
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.126186854142752e-06,
      "loss": 0.7139,
      "step": 1369
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.112624045470834e-06,
      "loss": 0.6711,
      "step": 1370
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.099064835910339e-06,
      "loss": 0.6994,
      "step": 1371
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.085509251317857e-06,
      "loss": 0.6945,
      "step": 1372
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.07195731754307e-06,
      "loss": 0.694,
      "step": 1373
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.058409060428684e-06,
      "loss": 0.6977,
      "step": 1374
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.044864505810415e-06,
      "loss": 0.6878,
      "step": 1375
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.0313236795169e-06,
      "loss": 0.689,
      "step": 1376
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.017786607369674e-06,
      "loss": 0.7174,
      "step": 1377
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.004253315183113e-06,
      "loss": 0.7047,
      "step": 1378
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.99072382876438e-06,
      "loss": 0.6847,
      "step": 1379
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.977198173913394e-06,
      "loss": 0.7027,
      "step": 1380
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.963676376422754e-06,
      "loss": 0.6888,
      "step": 1381
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.950158462077697e-06,
      "loss": 0.67,
      "step": 1382
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.936644456656082e-06,
      "loss": 0.6727,
      "step": 1383
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.923134385928282e-06,
      "loss": 0.7346,
      "step": 1384
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.909628275657199e-06,
      "loss": 0.7007,
      "step": 1385
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.896126151598149e-06,
      "loss": 0.7117,
      "step": 1386
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.882628039498876e-06,
      "loss": 0.675,
      "step": 1387
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.869133965099454e-06,
      "loss": 0.6717,
      "step": 1388
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.855643954132263e-06,
      "loss": 0.6893,
      "step": 1389
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.84215803232194e-06,
      "loss": 0.6912,
      "step": 1390
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.828676225385316e-06,
      "loss": 0.657,
      "step": 1391
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.815198559031383e-06,
      "loss": 0.6848,
      "step": 1392
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.80172505896123e-06,
      "loss": 0.6979,
      "step": 1393
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.788255750868002e-06,
      "loss": 0.6903,
      "step": 1394
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.774790660436857e-06,
      "loss": 0.7002,
      "step": 1395
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.761329813344901e-06,
      "loss": 0.6767,
      "step": 1396
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.747873235261157e-06,
      "loss": 0.7207,
      "step": 1397
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.734420951846498e-06,
      "loss": 0.6975,
      "step": 1398
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.720972988753619e-06,
      "loss": 0.6794,
      "step": 1399
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.707529371626966e-06,
      "loss": 0.6991,
      "step": 1400
    },
    {
      "epoch": 1.77,
      "eval_loss": 0.7086923122406006,
      "eval_runtime": 182.7999,
      "eval_samples_per_second": 11.291,
      "eval_steps_per_second": 0.093,
      "step": 1400
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.694090126102697e-06,
      "loss": 0.7017,
      "step": 1401
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.680655277808642e-06,
      "loss": 0.6942,
      "step": 1402
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.66722485236424e-06,
      "loss": 0.6707,
      "step": 1403
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.6537988753805e-06,
      "loss": 0.6898,
      "step": 1404
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.640377372459944e-06,
      "loss": 0.7162,
      "step": 1405
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.626960369196562e-06,
      "loss": 0.6852,
      "step": 1406
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.613547891175769e-06,
      "loss": 0.7025,
      "step": 1407
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.600139963974341e-06,
      "loss": 0.6902,
      "step": 1408
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.5867366131603885e-06,
      "loss": 0.6663,
      "step": 1409
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.573337864293283e-06,
      "loss": 0.6599,
      "step": 1410
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.559943742923626e-06,
      "loss": 0.7152,
      "step": 1411
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.546554274593197e-06,
      "loss": 0.6736,
      "step": 1412
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.533169484834897e-06,
      "loss": 0.7245,
      "step": 1413
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.519789399172713e-06,
      "loss": 0.6808,
      "step": 1414
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.506414043121647e-06,
      "loss": 0.6998,
      "step": 1415
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.4930434421877016e-06,
      "loss": 0.6875,
      "step": 1416
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.479677621867797e-06,
      "loss": 0.6631,
      "step": 1417
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.4663166076497376e-06,
      "loss": 0.7119,
      "step": 1418
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.452960425012173e-06,
      "loss": 0.6694,
      "step": 1419
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.4396090994245295e-06,
      "loss": 0.6966,
      "step": 1420
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.426262656346979e-06,
      "loss": 0.6763,
      "step": 1421
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.412921121230373e-06,
      "loss": 0.6815,
      "step": 1422
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.399584519516212e-06,
      "loss": 0.7266,
      "step": 1423
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.386252876636588e-06,
      "loss": 0.6943,
      "step": 1424
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.372926218014131e-06,
      "loss": 0.7012,
      "step": 1425
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.3596045690619755e-06,
      "loss": 0.7072,
      "step": 1426
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.346287955183693e-06,
      "loss": 0.7327,
      "step": 1427
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.332976401773262e-06,
      "loss": 0.6965,
      "step": 1428
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.319669934215008e-06,
      "loss": 0.6928,
      "step": 1429
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.306368577883547e-06,
      "loss": 0.7051,
      "step": 1430
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.293072358143773e-06,
      "loss": 0.7144,
      "step": 1431
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.279781300350758e-06,
      "loss": 0.7195,
      "step": 1432
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.266495429849754e-06,
      "loss": 0.7071,
      "step": 1433
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.2532147719761e-06,
      "loss": 0.6939,
      "step": 1434
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.239939352055208e-06,
      "loss": 0.6909,
      "step": 1435
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.2266691954025015e-06,
      "loss": 0.6868,
      "step": 1436
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.213404327323359e-06,
      "loss": 0.6871,
      "step": 1437
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.2001447731130836e-06,
      "loss": 0.6964,
      "step": 1438
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.186890558056836e-06,
      "loss": 0.6945,
      "step": 1439
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.173641707429606e-06,
      "loss": 0.7004,
      "step": 1440
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.160398246496143e-06,
      "loss": 0.6743,
      "step": 1441
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.147160200510923e-06,
      "loss": 0.6798,
      "step": 1442
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.133927594718102e-06,
      "loss": 0.6788,
      "step": 1443
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.120700454351449e-06,
      "loss": 0.6839,
      "step": 1444
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.107478804634324e-06,
      "loss": 0.6838,
      "step": 1445
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.094262670779611e-06,
      "loss": 0.6798,
      "step": 1446
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.081052077989668e-06,
      "loss": 0.6968,
      "step": 1447
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.067847051456303e-06,
      "loss": 0.702,
      "step": 1448
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.0546476163606905e-06,
      "loss": 0.6998,
      "step": 1449
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.041453797873363e-06,
      "loss": 0.6736,
      "step": 1450
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.028265621154127e-06,
      "loss": 0.6745,
      "step": 1451
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.015083111352031e-06,
      "loss": 0.6614,
      "step": 1452
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.001906293605329e-06,
      "loss": 0.6709,
      "step": 1453
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.988735193041406e-06,
      "loss": 0.6826,
      "step": 1454
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.975569834776757e-06,
      "loss": 0.6605,
      "step": 1455
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.962410243916917e-06,
      "loss": 0.6905,
      "step": 1456
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.949256445556433e-06,
      "loss": 0.6845,
      "step": 1457
    },
    {
      "epoch": 1.84,
      "learning_rate": 6.936108464778795e-06,
      "loss": 0.6936,
      "step": 1458
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.922966326656406e-06,
      "loss": 0.6546,
      "step": 1459
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.909830056250527e-06,
      "loss": 0.6941,
      "step": 1460
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.896699678611226e-06,
      "loss": 0.6808,
      "step": 1461
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.883575218777341e-06,
      "loss": 0.6952,
      "step": 1462
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.87045670177642e-06,
      "loss": 0.6608,
      "step": 1463
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.8573441526246695e-06,
      "loss": 0.7096,
      "step": 1464
    },
    {
      "epoch": 1.85,
      "learning_rate": 6.844237596326941e-06,
      "loss": 0.7026,
      "step": 1465
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.8311370578766255e-06,
      "loss": 0.6886,
      "step": 1466
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.81804256225567e-06,
      "loss": 0.6619,
      "step": 1467
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.804954134434469e-06,
      "loss": 0.7029,
      "step": 1468
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.791871799371874e-06,
      "loss": 0.7018,
      "step": 1469
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.778795582015096e-06,
      "loss": 0.6873,
      "step": 1470
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.765725507299688e-06,
      "loss": 0.6709,
      "step": 1471
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.7526616001494905e-06,
      "loss": 0.6889,
      "step": 1472
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.7396038854765825e-06,
      "loss": 0.6679,
      "step": 1473
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.7265523881812335e-06,
      "loss": 0.6898,
      "step": 1474
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.7135071331518575e-06,
      "loss": 0.6918,
      "step": 1475
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.7004681452649625e-06,
      "loss": 0.69,
      "step": 1476
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.687435449385109e-06,
      "loss": 0.6911,
      "step": 1477
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.674409070364854e-06,
      "loss": 0.6776,
      "step": 1478
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.661389033044717e-06,
      "loss": 0.685,
      "step": 1479
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.648375362253119e-06,
      "loss": 0.6571,
      "step": 1480
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.6353680828063306e-06,
      "loss": 0.7027,
      "step": 1481
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.62236721950846e-06,
      "loss": 0.6832,
      "step": 1482
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.609372797151349e-06,
      "loss": 0.7081,
      "step": 1483
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.596384840514587e-06,
      "loss": 0.6833,
      "step": 1484
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.583403374365406e-06,
      "loss": 0.6771,
      "step": 1485
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.570428423458687e-06,
      "loss": 0.6748,
      "step": 1486
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.557460012536864e-06,
      "loss": 0.7006,
      "step": 1487
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.5444981663299135e-06,
      "loss": 0.6726,
      "step": 1488
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.531542909555292e-06,
      "loss": 0.6576,
      "step": 1489
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.518594266917883e-06,
      "loss": 0.6659,
      "step": 1490
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.505652263109967e-06,
      "loss": 0.6965,
      "step": 1491
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.492716922811158e-06,
      "loss": 0.665,
      "step": 1492
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.479788270688363e-06,
      "loss": 0.6813,
      "step": 1493
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.466866331395742e-06,
      "loss": 0.6673,
      "step": 1494
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.453951129574644e-06,
      "loss": 0.6871,
      "step": 1495
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.44104268985358e-06,
      "loss": 0.6794,
      "step": 1496
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.428141036848157e-06,
      "loss": 0.6657,
      "step": 1497
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.415246195161046e-06,
      "loss": 0.6868,
      "step": 1498
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.4023581893819345e-06,
      "loss": 0.6861,
      "step": 1499
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.3894770440874525e-06,
      "loss": 0.6803,
      "step": 1500
    },
    {
      "epoch": 1.9,
      "eval_loss": 0.6962260007858276,
      "eval_runtime": 182.5589,
      "eval_samples_per_second": 11.306,
      "eval_steps_per_second": 0.093,
      "step": 1500
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.376602783841177e-06,
      "loss": 0.7019,
      "step": 1501
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.36373543319353e-06,
      "loss": 0.7141,
      "step": 1502
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.350875016681779e-06,
      "loss": 0.6687,
      "step": 1503
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.338021558829951e-06,
      "loss": 0.6711,
      "step": 1504
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.325175084148809e-06,
      "loss": 0.6706,
      "step": 1505
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.312335617135806e-06,
      "loss": 0.6672,
      "step": 1506
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.299503182275023e-06,
      "loss": 0.6962,
      "step": 1507
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.286677804037137e-06,
      "loss": 0.6682,
      "step": 1508
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.273859506879365e-06,
      "loss": 0.665,
      "step": 1509
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.261048315245419e-06,
      "loss": 0.6911,
      "step": 1510
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.24824425356547e-06,
      "loss": 0.6568,
      "step": 1511
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.23544734625608e-06,
      "loss": 0.6708,
      "step": 1512
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.222657617720182e-06,
      "loss": 0.6839,
      "step": 1513
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.209875092347004e-06,
      "loss": 0.6936,
      "step": 1514
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.197099794512056e-06,
      "loss": 0.6858,
      "step": 1515
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.18433174857705e-06,
      "loss": 0.6939,
      "step": 1516
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.171570978889872e-06,
      "loss": 0.6896,
      "step": 1517
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.158817509784542e-06,
      "loss": 0.6719,
      "step": 1518
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.1460713655811465e-06,
      "loss": 0.669,
      "step": 1519
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.133332570585813e-06,
      "loss": 0.6966,
      "step": 1520
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.120601149090649e-06,
      "loss": 0.6416,
      "step": 1521
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.1078771253737e-06,
      "loss": 0.6356,
      "step": 1522
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.095160523698913e-06,
      "loss": 0.7008,
      "step": 1523
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.082451368316072e-06,
      "loss": 0.6825,
      "step": 1524
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.069749683460765e-06,
      "loss": 0.658,
      "step": 1525
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.057055493354334e-06,
      "loss": 0.6736,
      "step": 1526
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.044368822203833e-06,
      "loss": 0.6777,
      "step": 1527
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.031689694201972e-06,
      "loss": 0.6656,
      "step": 1528
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.019018133527075e-06,
      "loss": 0.6732,
      "step": 1529
    },
    {
      "epoch": 1.94,
      "learning_rate": 6.006354164343047e-06,
      "loss": 0.6971,
      "step": 1530
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.9936978107992995e-06,
      "loss": 0.6577,
      "step": 1531
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.98104909703074e-06,
      "loss": 0.6842,
      "step": 1532
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.9684080471576925e-06,
      "loss": 0.6822,
      "step": 1533
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.955774685285872e-06,
      "loss": 0.6754,
      "step": 1534
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.943149035506337e-06,
      "loss": 0.6681,
      "step": 1535
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.930531121895429e-06,
      "loss": 0.6604,
      "step": 1536
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.9179209685147525e-06,
      "loss": 0.6837,
      "step": 1537
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.9053185994110975e-06,
      "loss": 0.6636,
      "step": 1538
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.892724038616419e-06,
      "loss": 0.6793,
      "step": 1539
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.880137310147782e-06,
      "loss": 0.6787,
      "step": 1540
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.867558438007311e-06,
      "loss": 0.6605,
      "step": 1541
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.854987446182156e-06,
      "loss": 0.6892,
      "step": 1542
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.842424358644432e-06,
      "loss": 0.6736,
      "step": 1543
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.829869199351188e-06,
      "loss": 0.6778,
      "step": 1544
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.8173219922443516e-06,
      "loss": 0.6617,
      "step": 1545
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.804782761250678e-06,
      "loss": 0.645,
      "step": 1546
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.7922515302817254e-06,
      "loss": 0.6802,
      "step": 1547
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.77972832323379e-06,
      "loss": 0.6673,
      "step": 1548
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.7672131639878725e-06,
      "loss": 0.6961,
      "step": 1549
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.754706076409613e-06,
      "loss": 0.6687,
      "step": 1550
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.742207084349274e-06,
      "loss": 0.696,
      "step": 1551
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.7297162116416774e-06,
      "loss": 0.6478,
      "step": 1552
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.717233482106153e-06,
      "loss": 0.6679,
      "step": 1553
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.704758919546511e-06,
      "loss": 0.6656,
      "step": 1554
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.692292547750989e-06,
      "loss": 0.7189,
      "step": 1555
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.6798343904922005e-06,
      "loss": 0.6776,
      "step": 1556
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.667384471527093e-06,
      "loss": 0.6937,
      "step": 1557
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.654942814596902e-06,
      "loss": 0.7149,
      "step": 1558
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.642509443427124e-06,
      "loss": 0.6722,
      "step": 1559
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.630084381727434e-06,
      "loss": 0.6517,
      "step": 1560
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.617667653191678e-06,
      "loss": 0.6771,
      "step": 1561
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.605259281497796e-06,
      "loss": 0.6697,
      "step": 1562
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.592859290307804e-06,
      "loss": 0.685,
      "step": 1563
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.580467703267736e-06,
      "loss": 0.6537,
      "step": 1564
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.5680845440075885e-06,
      "loss": 0.6704,
      "step": 1565
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.5557098361413e-06,
      "loss": 0.6388,
      "step": 1566
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.5433436032666864e-06,
      "loss": 0.675,
      "step": 1567
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.530985868965408e-06,
      "loss": 0.6817,
      "step": 1568
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.5186366568029084e-06,
      "loss": 0.6812,
      "step": 1569
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.5062959903283855e-06,
      "loss": 0.6848,
      "step": 1570
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.493963893074749e-06,
      "loss": 0.6546,
      "step": 1571
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.481640388558551e-06,
      "loss": 0.6657,
      "step": 1572
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.46932550027997e-06,
      "loss": 0.6877,
      "step": 1573
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.457019251722755e-06,
      "loss": 0.6978,
      "step": 1574
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.444721666354169e-06,
      "loss": 0.6743,
      "step": 1575
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.432432767624967e-06,
      "loss": 0.6741,
      "step": 1576
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.420152578969327e-06,
      "loss": 0.6552,
      "step": 1577
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.407881123804826e-06,
      "loss": 0.6558,
      "step": 1578
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.39561842553239e-06,
      "loss": 0.6464,
      "step": 1579
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.3833645075362295e-06,
      "loss": 0.6647,
      "step": 1580
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.37111939318383e-06,
      "loss": 0.6425,
      "step": 1581
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.3588831058258805e-06,
      "loss": 0.6327,
      "step": 1582
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.346655668796241e-06,
      "loss": 0.6117,
      "step": 1583
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.334437105411887e-06,
      "loss": 0.6423,
      "step": 1584
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.32222743897288e-06,
      "loss": 0.618,
      "step": 1585
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.310026692762316e-06,
      "loss": 0.6583,
      "step": 1586
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.2978348900462705e-06,
      "loss": 0.6528,
      "step": 1587
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.28565205407378e-06,
      "loss": 0.6597,
      "step": 1588
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.273478208076761e-06,
      "loss": 0.661,
      "step": 1589
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.2613133752700145e-06,
      "loss": 0.6532,
      "step": 1590
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.2491575788511315e-06,
      "loss": 0.6514,
      "step": 1591
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.237010842000474e-06,
      "loss": 0.628,
      "step": 1592
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.224873187881136e-06,
      "loss": 0.6468,
      "step": 1593
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.212744639638889e-06,
      "loss": 0.6394,
      "step": 1594
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.200625220402139e-06,
      "loss": 0.673,
      "step": 1595
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.188514953281878e-06,
      "loss": 0.6248,
      "step": 1596
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.176413861371654e-06,
      "loss": 0.6159,
      "step": 1597
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.164321967747519e-06,
      "loss": 0.6215,
      "step": 1598
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.152239295467971e-06,
      "loss": 0.6508,
      "step": 1599
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.14016586757394e-06,
      "loss": 0.6272,
      "step": 1600
    },
    {
      "epoch": 2.02,
      "eval_loss": 0.6868788599967957,
      "eval_runtime": 182.5254,
      "eval_samples_per_second": 11.308,
      "eval_steps_per_second": 0.093,
      "step": 1600
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.128101707088717e-06,
      "loss": 0.6343,
      "step": 1601
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.116046837017929e-06,
      "loss": 0.6597,
      "step": 1602
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.10400128034948e-06,
      "loss": 0.626,
      "step": 1603
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.091965060053505e-06,
      "loss": 0.6417,
      "step": 1604
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.079938199082363e-06,
      "loss": 0.6522,
      "step": 1605
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.067920720370536e-06,
      "loss": 0.6464,
      "step": 1606
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.0559126468346354e-06,
      "loss": 0.6529,
      "step": 1607
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.043914001373319e-06,
      "loss": 0.6322,
      "step": 1608
    },
    {
      "epoch": 2.04,
      "learning_rate": 5.031924806867281e-06,
      "loss": 0.6172,
      "step": 1609
    },
    {
      "epoch": 2.04,
      "learning_rate": 5.019945086179192e-06,
      "loss": 0.6579,
      "step": 1610
    },
    {
      "epoch": 2.04,
      "learning_rate": 5.007974862153643e-06,
      "loss": 0.6487,
      "step": 1611
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.996014157617131e-06,
      "loss": 0.6112,
      "step": 1612
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.984062995377992e-06,
      "loss": 0.6468,
      "step": 1613
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.972121398226371e-06,
      "loss": 0.6102,
      "step": 1614
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.960189388934163e-06,
      "loss": 0.6334,
      "step": 1615
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.9482669902549896e-06,
      "loss": 0.6263,
      "step": 1616
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.936354224924143e-06,
      "loss": 0.6534,
      "step": 1617
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.924451115658539e-06,
      "loss": 0.6257,
      "step": 1618
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.912557685156688e-06,
      "loss": 0.6456,
      "step": 1619
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.900673956098644e-06,
      "loss": 0.6399,
      "step": 1620
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.888799951145948e-06,
      "loss": 0.6307,
      "step": 1621
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.876935692941615e-06,
      "loss": 0.6171,
      "step": 1622
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.865081204110057e-06,
      "loss": 0.6631,
      "step": 1623
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.853236507257076e-06,
      "loss": 0.6362,
      "step": 1624
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.841401624969782e-06,
      "loss": 0.6325,
      "step": 1625
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.829576579816584e-06,
      "loss": 0.6788,
      "step": 1626
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.817761394347117e-06,
      "loss": 0.6109,
      "step": 1627
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.805956091092228e-06,
      "loss": 0.6418,
      "step": 1628
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.794160692563917e-06,
      "loss": 0.6456,
      "step": 1629
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.7823752212552855e-06,
      "loss": 0.6595,
      "step": 1630
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.7705996996405165e-06,
      "loss": 0.6388,
      "step": 1631
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.758834150174819e-06,
      "loss": 0.6353,
      "step": 1632
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.747078595294375e-06,
      "loss": 0.6646,
      "step": 1633
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.735333057416316e-06,
      "loss": 0.6128,
      "step": 1634
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.7235975589386715e-06,
      "loss": 0.6329,
      "step": 1635
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.7118721222403295e-06,
      "loss": 0.6225,
      "step": 1636
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.700156769680981e-06,
      "loss": 0.6237,
      "step": 1637
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.6884515236010864e-06,
      "loss": 0.6142,
      "step": 1638
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.6767564063218525e-06,
      "loss": 0.6353,
      "step": 1639
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.66507144014515e-06,
      "loss": 0.613,
      "step": 1640
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.653396647353506e-06,
      "loss": 0.6136,
      "step": 1641
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.641732050210032e-06,
      "loss": 0.6507,
      "step": 1642
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.630077670958421e-06,
      "loss": 0.6506,
      "step": 1643
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.618433531822859e-06,
      "loss": 0.6431,
      "step": 1644
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.606799655008009e-06,
      "loss": 0.6313,
      "step": 1645
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.595176062698975e-06,
      "loss": 0.6375,
      "step": 1646
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.583562777061239e-06,
      "loss": 0.6352,
      "step": 1647
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.571959820240637e-06,
      "loss": 0.6367,
      "step": 1648
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.560367214363295e-06,
      "loss": 0.663,
      "step": 1649
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.5487849815356145e-06,
      "loss": 0.6571,
      "step": 1650
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.5372131438442135e-06,
      "loss": 0.6227,
      "step": 1651
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.525651723355876e-06,
      "loss": 0.6215,
      "step": 1652
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.514100742117535e-06,
      "loss": 0.6052,
      "step": 1653
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.5025602221562105e-06,
      "loss": 0.6367,
      "step": 1654
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.491030185478976e-06,
      "loss": 0.6229,
      "step": 1655
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.479510654072909e-06,
      "loss": 0.6456,
      "step": 1656
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.46800164990505e-06,
      "loss": 0.6241,
      "step": 1657
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.456503194922386e-06,
      "loss": 0.5923,
      "step": 1658
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.445015311051761e-06,
      "loss": 0.6436,
      "step": 1659
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.433538020199882e-06,
      "loss": 0.6431,
      "step": 1660
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.422071344253237e-06,
      "loss": 0.648,
      "step": 1661
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.410615305078087e-06,
      "loss": 0.6338,
      "step": 1662
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.399169924520403e-06,
      "loss": 0.6292,
      "step": 1663
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.387735224405828e-06,
      "loss": 0.6206,
      "step": 1664
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.3763112265396445e-06,
      "loss": 0.6282,
      "step": 1665
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.364897952706726e-06,
      "loss": 0.646,
      "step": 1666
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.353495424671485e-06,
      "loss": 0.6514,
      "step": 1667
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.342103664177856e-06,
      "loss": 0.6214,
      "step": 1668
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.330722692949233e-06,
      "loss": 0.6617,
      "step": 1669
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.319352532688444e-06,
      "loss": 0.6162,
      "step": 1670
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.3079932050776855e-06,
      "loss": 0.6341,
      "step": 1671
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.296644731778509e-06,
      "loss": 0.6405,
      "step": 1672
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.28530713443177e-06,
      "loss": 0.6323,
      "step": 1673
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.273980434657571e-06,
      "loss": 0.6285,
      "step": 1674
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.262664654055247e-06,
      "loss": 0.6242,
      "step": 1675
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.2513598142032995e-06,
      "loss": 0.6312,
      "step": 1676
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.240065936659374e-06,
      "loss": 0.6228,
      "step": 1677
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.228783042960216e-06,
      "loss": 0.6029,
      "step": 1678
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.21751115462161e-06,
      "loss": 0.6434,
      "step": 1679
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.206250293138366e-06,
      "loss": 0.6479,
      "step": 1680
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.195000479984264e-06,
      "loss": 0.6195,
      "step": 1681
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.183761736612022e-06,
      "loss": 0.6329,
      "step": 1682
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.17253408445323e-06,
      "loss": 0.6348,
      "step": 1683
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.161317544918345e-06,
      "loss": 0.6092,
      "step": 1684
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.15011213939663e-06,
      "loss": 0.6627,
      "step": 1685
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.138917889256109e-06,
      "loss": 0.6361,
      "step": 1686
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.127734815843537e-06,
      "loss": 0.6167,
      "step": 1687
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.11656294048436e-06,
      "loss": 0.6148,
      "step": 1688
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.105402284482668e-06,
      "loss": 0.6237,
      "step": 1689
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.094252869121153e-06,
      "loss": 0.6167,
      "step": 1690
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.083114715661069e-06,
      "loss": 0.6464,
      "step": 1691
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.0719878453422005e-06,
      "loss": 0.6493,
      "step": 1692
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.060872279382817e-06,
      "loss": 0.6468,
      "step": 1693
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.049768038979631e-06,
      "loss": 0.6676,
      "step": 1694
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.038675145307747e-06,
      "loss": 0.6249,
      "step": 1695
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.027593619520646e-06,
      "loss": 0.6124,
      "step": 1696
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.016523482750126e-06,
      "loss": 0.6201,
      "step": 1697
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.0054647561062625e-06,
      "loss": 0.6283,
      "step": 1698
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.994417460677375e-06,
      "loss": 0.6278,
      "step": 1699
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.98338161752999e-06,
      "loss": 0.6231,
      "step": 1700
    },
    {
      "epoch": 2.15,
      "eval_loss": 0.6741244196891785,
      "eval_runtime": 182.574,
      "eval_samples_per_second": 11.305,
      "eval_steps_per_second": 0.093,
      "step": 1700
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.972357247708791e-06,
      "loss": 0.6329,
      "step": 1701
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.9613443722365815e-06,
      "loss": 0.6604,
      "step": 1702
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.950343012114238e-06,
      "loss": 0.6372,
      "step": 1703
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.939353188320702e-06,
      "loss": 0.6455,
      "step": 1704
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.9283749218128885e-06,
      "loss": 0.6224,
      "step": 1705
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.917408233525694e-06,
      "loss": 0.6419,
      "step": 1706
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.90645314437192e-06,
      "loss": 0.6119,
      "step": 1707
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.895509675242261e-06,
      "loss": 0.6279,
      "step": 1708
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.884577847005251e-06,
      "loss": 0.6338,
      "step": 1709
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.8736576805072165e-06,
      "loss": 0.6275,
      "step": 1710
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.862749196572256e-06,
      "loss": 0.6111,
      "step": 1711
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.8518524160021876e-06,
      "loss": 0.6164,
      "step": 1712
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.840967359576513e-06,
      "loss": 0.6384,
      "step": 1713
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.830094048052367e-06,
      "loss": 0.6354,
      "step": 1714
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.819232502164499e-06,
      "loss": 0.6049,
      "step": 1715
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.808382742625222e-06,
      "loss": 0.5921,
      "step": 1716
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.797544790124361e-06,
      "loss": 0.6363,
      "step": 1717
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.7867186653292374e-06,
      "loss": 0.6391,
      "step": 1718
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.775904388884618e-06,
      "loss": 0.6246,
      "step": 1719
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.7651019814126656e-06,
      "loss": 0.6529,
      "step": 1720
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.754311463512922e-06,
      "loss": 0.6102,
      "step": 1721
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.7435328557622395e-06,
      "loss": 0.6466,
      "step": 1722
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.732766178714784e-06,
      "loss": 0.6127,
      "step": 1723
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.7220114529019437e-06,
      "loss": 0.6618,
      "step": 1724
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.7112686988323353e-06,
      "loss": 0.6369,
      "step": 1725
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.700537936991733e-06,
      "loss": 0.6207,
      "step": 1726
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.68981918784305e-06,
      "loss": 0.6248,
      "step": 1727
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.679112471826296e-06,
      "loss": 0.6343,
      "step": 1728
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.668417809358521e-06,
      "loss": 0.624,
      "step": 1729
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.6577352208338015e-06,
      "loss": 0.6254,
      "step": 1730
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.6470647266231883e-06,
      "loss": 0.6228,
      "step": 1731
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.6364063470746603e-06,
      "loss": 0.622,
      "step": 1732
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.625760102513103e-06,
      "loss": 0.614,
      "step": 1733
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.6151260132402598e-06,
      "loss": 0.6096,
      "step": 1734
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.604504099534696e-06,
      "loss": 0.6305,
      "step": 1735
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.5938943816517544e-06,
      "loss": 0.6373,
      "step": 1736
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.5832968798235158e-06,
      "loss": 0.6163,
      "step": 1737
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.572711614258786e-06,
      "loss": 0.6014,
      "step": 1738
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.562138605143015e-06,
      "loss": 0.6184,
      "step": 1739
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.5515778726382967e-06,
      "loss": 0.6261,
      "step": 1740
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.5410294368832967e-06,
      "loss": 0.6141,
      "step": 1741
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.5304933179932543e-06,
      "loss": 0.6693,
      "step": 1742
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.519969536059903e-06,
      "loss": 0.6384,
      "step": 1743
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.5094581111514535e-06,
      "loss": 0.6199,
      "step": 1744
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.4989590633125583e-06,
      "loss": 0.6311,
      "step": 1745
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.4884724125642646e-06,
      "loss": 0.6144,
      "step": 1746
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.477998178903982e-06,
      "loss": 0.662,
      "step": 1747
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.467536382305433e-06,
      "loss": 0.6402,
      "step": 1748
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.457087042718631e-06,
      "loss": 0.6157,
      "step": 1749
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.446650180069837e-06,
      "loss": 0.5892,
      "step": 1750
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.4362258142615067e-06,
      "loss": 0.6244,
      "step": 1751
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.4258139651722776e-06,
      "loss": 0.6488,
      "step": 1752
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.4154146526569144e-06,
      "loss": 0.6383,
      "step": 1753
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.405027896546277e-06,
      "loss": 0.6153,
      "step": 1754
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.394653716647277e-06,
      "loss": 0.6257,
      "step": 1755
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.384292132742839e-06,
      "loss": 0.613,
      "step": 1756
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.3739431645918874e-06,
      "loss": 0.644,
      "step": 1757
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.363606831929268e-06,
      "loss": 0.5996,
      "step": 1758
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.3532831544657464e-06,
      "loss": 0.6404,
      "step": 1759
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.342972151887941e-06,
      "loss": 0.6105,
      "step": 1760
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.3326738438583116e-06,
      "loss": 0.6245,
      "step": 1761
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.3223882500151116e-06,
      "loss": 0.6244,
      "step": 1762
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.312115389972337e-06,
      "loss": 0.6231,
      "step": 1763
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.3018552833197115e-06,
      "loss": 0.6301,
      "step": 1764
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.2916079496226407e-06,
      "loss": 0.604,
      "step": 1765
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.2813734084221584e-06,
      "loss": 0.5995,
      "step": 1766
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.2711516792349206e-06,
      "loss": 0.65,
      "step": 1767
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.2609427815531426e-06,
      "loss": 0.6132,
      "step": 1768
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.2507467348445753e-06,
      "loss": 0.6103,
      "step": 1769
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.2405635585524566e-06,
      "loss": 0.6503,
      "step": 1770
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.2303932720954887e-06,
      "loss": 0.5852,
      "step": 1771
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.220235894867794e-06,
      "loss": 0.6325,
      "step": 1772
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.2100914462388655e-06,
      "loss": 0.6061,
      "step": 1773
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.199959945553561e-06,
      "loss": 0.6104,
      "step": 1774
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.1898414121320277e-06,
      "loss": 0.6187,
      "step": 1775
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.1797358652697052e-06,
      "loss": 0.6301,
      "step": 1776
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.1696433242372537e-06,
      "loss": 0.6114,
      "step": 1777
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.1595638082805324e-06,
      "loss": 0.6001,
      "step": 1778
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.1494973366205717e-06,
      "loss": 0.6275,
      "step": 1779
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.1394439284535206e-06,
      "loss": 0.646,
      "step": 1780
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.1294036029506215e-06,
      "loss": 0.6344,
      "step": 1781
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.11937637925816e-06,
      "loss": 0.6117,
      "step": 1782
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.109362276497446e-06,
      "loss": 0.5945,
      "step": 1783
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0993613137647705e-06,
      "loss": 0.6288,
      "step": 1784
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.089373510131354e-06,
      "loss": 0.6431,
      "step": 1785
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0793988846433375e-06,
      "loss": 0.6065,
      "step": 1786
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0694374563217244e-06,
      "loss": 0.6329,
      "step": 1787
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0594892441623593e-06,
      "loss": 0.6286,
      "step": 1788
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.0495542671358745e-06,
      "loss": 0.6421,
      "step": 1789
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.0396325441876627e-06,
      "loss": 0.6289,
      "step": 1790
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.0297240942378603e-06,
      "loss": 0.6261,
      "step": 1791
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.0198289361812695e-06,
      "loss": 0.5952,
      "step": 1792
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.009947088887364e-06,
      "loss": 0.6269,
      "step": 1793
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.00007857120022e-06,
      "loss": 0.6048,
      "step": 1794
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9902234019385056e-06,
      "loss": 0.5928,
      "step": 1795
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.9803815998954334e-06,
      "loss": 0.6335,
      "step": 1796
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.970553183838717e-06,
      "loss": 0.6343,
      "step": 1797
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.960738172510551e-06,
      "loss": 0.6019,
      "step": 1798
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.9509365846275673e-06,
      "loss": 0.6199,
      "step": 1799
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.941148438880803e-06,
      "loss": 0.6185,
      "step": 1800
    },
    {
      "epoch": 2.28,
      "eval_loss": 0.6662217378616333,
      "eval_runtime": 182.4504,
      "eval_samples_per_second": 11.313,
      "eval_steps_per_second": 0.093,
      "step": 1800
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.9313737539356525e-06,
      "loss": 0.6266,
      "step": 1801
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.921612548431848e-06,
      "loss": 0.6034,
      "step": 1802
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.9118648409834205e-06,
      "loss": 0.6271,
      "step": 1803
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.9021306501786496e-06,
      "loss": 0.6263,
      "step": 1804
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.8924099945800533e-06,
      "loss": 0.6176,
      "step": 1805
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.882702892724326e-06,
      "loss": 0.6016,
      "step": 1806
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.8730093631223243e-06,
      "loss": 0.6338,
      "step": 1807
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.863329424259026e-06,
      "loss": 0.6062,
      "step": 1808
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.8536630945934783e-06,
      "loss": 0.6114,
      "step": 1809
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.8440103925587904e-06,
      "loss": 0.6142,
      "step": 1810
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.834371336562077e-06,
      "loss": 0.6149,
      "step": 1811
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.8247459449844395e-06,
      "loss": 0.6188,
      "step": 1812
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.8151342361809077e-06,
      "loss": 0.5934,
      "step": 1813
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.805536228480429e-06,
      "loss": 0.5872,
      "step": 1814
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.795951940185827e-06,
      "loss": 0.6232,
      "step": 1815
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.7863813895737503e-06,
      "loss": 0.6409,
      "step": 1816
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.7768245948946615e-06,
      "loss": 0.5895,
      "step": 1817
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.767281574372791e-06,
      "loss": 0.6271,
      "step": 1818
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.757752346206095e-06,
      "loss": 0.6061,
      "step": 1819
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.748236928566238e-06,
      "loss": 0.6371,
      "step": 1820
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.738735339598536e-06,
      "loss": 0.6043,
      "step": 1821
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.7292475974219544e-06,
      "loss": 0.6096,
      "step": 1822
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.719773720129033e-06,
      "loss": 0.6259,
      "step": 1823
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.7103137257858867e-06,
      "loss": 0.6076,
      "step": 1824
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.700867632432145e-06,
      "loss": 0.6217,
      "step": 1825
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6914354580809364e-06,
      "loss": 0.6464,
      "step": 1826
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.682017220718851e-06,
      "loss": 0.631,
      "step": 1827
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6726129383058862e-06,
      "loss": 0.5923,
      "step": 1828
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.663222628775445e-06,
      "loss": 0.6456,
      "step": 1829
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.6538463100342773e-06,
      "loss": 0.6329,
      "step": 1830
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.6444839999624496e-06,
      "loss": 0.6146,
      "step": 1831
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.635135716413322e-06,
      "loss": 0.6346,
      "step": 1832
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.6258014772135045e-06,
      "loss": 0.6022,
      "step": 1833
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.6164813001628275e-06,
      "loss": 0.6143,
      "step": 1834
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.607175203034299e-06,
      "loss": 0.6248,
      "step": 1835
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.5978832035740777e-06,
      "loss": 0.6397,
      "step": 1836
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.5886053195014537e-06,
      "loss": 0.6431,
      "step": 1837
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5793415685087797e-06,
      "loss": 0.6278,
      "step": 1838
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.570091968261472e-06,
      "loss": 0.61,
      "step": 1839
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.56085653639795e-06,
      "loss": 0.6555,
      "step": 1840
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5516352905296325e-06,
      "loss": 0.6018,
      "step": 1841
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5424282482408703e-06,
      "loss": 0.6276,
      "step": 1842
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5332354270889313e-06,
      "loss": 0.6286,
      "step": 1843
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.5240568446039716e-06,
      "loss": 0.6034,
      "step": 1844
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.514892518288988e-06,
      "loss": 0.6149,
      "step": 1845
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.5057424656198013e-06,
      "loss": 0.6337,
      "step": 1846
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.496606704044999e-06,
      "loss": 0.6173,
      "step": 1847
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.487485250985928e-06,
      "loss": 0.6204,
      "step": 1848
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4783781238366477e-06,
      "loss": 0.6149,
      "step": 1849
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.469285339963892e-06,
      "loss": 0.611,
      "step": 1850
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.4602069167070496e-06,
      "loss": 0.6347,
      "step": 1851
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.451142871378124e-06,
      "loss": 0.6131,
      "step": 1852
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.442093221261702e-06,
      "loss": 0.6024,
      "step": 1853
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.4330579836149148e-06,
      "loss": 0.6025,
      "step": 1854
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.4240371756674063e-06,
      "loss": 0.6665,
      "step": 1855
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.415030814621322e-06,
      "loss": 0.6443,
      "step": 1856
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.4060389176512366e-06,
      "loss": 0.5979,
      "step": 1857
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.397061501904159e-06,
      "loss": 0.6317,
      "step": 1858
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.3880985844994674e-06,
      "loss": 0.5806,
      "step": 1859
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.379150182528909e-06,
      "loss": 0.6082,
      "step": 1860
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.3702163130565415e-06,
      "loss": 0.623,
      "step": 1861
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3612969931187093e-06,
      "loss": 0.6087,
      "step": 1862
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3523922397240163e-06,
      "loss": 0.5897,
      "step": 1863
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3435020698532894e-06,
      "loss": 0.6098,
      "step": 1864
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.334626500459539e-06,
      "loss": 0.6406,
      "step": 1865
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3257655484679376e-06,
      "loss": 0.6351,
      "step": 1866
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.3169192307757846e-06,
      "loss": 0.6098,
      "step": 1867
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.308087564252475e-06,
      "loss": 0.6033,
      "step": 1868
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.299270565739453e-06,
      "loss": 0.6093,
      "step": 1869
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.290468252050204e-06,
      "loss": 0.6084,
      "step": 1870
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2816806399702083e-06,
      "loss": 0.6199,
      "step": 1871
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2729077462569028e-06,
      "loss": 0.5901,
      "step": 1872
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.2641495876396713e-06,
      "loss": 0.6468,
      "step": 1873
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.25540618081978e-06,
      "loss": 0.5909,
      "step": 1874
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.246677542470388e-06,
      "loss": 0.6076,
      "step": 1875
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.237963689236472e-06,
      "loss": 0.6467,
      "step": 1876
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.2292646377348205e-06,
      "loss": 0.5948,
      "step": 1877
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.220580404553998e-06,
      "loss": 0.596,
      "step": 1878
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.2119110062543105e-06,
      "loss": 0.631,
      "step": 1879
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.2032564593677773e-06,
      "loss": 0.6314,
      "step": 1880
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.19461678039809e-06,
      "loss": 0.6219,
      "step": 1881
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.1859919858205936e-06,
      "loss": 0.6096,
      "step": 1882
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.1773820920822507e-06,
      "loss": 0.6186,
      "step": 1883
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.1687871156016016e-06,
      "loss": 0.6079,
      "step": 1884
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1602070727687463e-06,
      "loss": 0.6368,
      "step": 1885
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.151641979945307e-06,
      "loss": 0.6167,
      "step": 1886
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1430918534643996e-06,
      "loss": 0.6164,
      "step": 1887
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.134556709630591e-06,
      "loss": 0.6354,
      "step": 1888
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.12603656471988e-06,
      "loss": 0.604,
      "step": 1889
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.117531434979675e-06,
      "loss": 0.601,
      "step": 1890
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1090413366287353e-06,
      "loss": 0.6048,
      "step": 1891
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1005662858571684e-06,
      "loss": 0.6263,
      "step": 1892
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.092106298826375e-06,
      "loss": 0.5904,
      "step": 1893
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.083661391669043e-06,
      "loss": 0.6064,
      "step": 1894
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.075231580489098e-06,
      "loss": 0.6225,
      "step": 1895
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.066816881361673e-06,
      "loss": 0.6443,
      "step": 1896
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0584173103330927e-06,
      "loss": 0.6124,
      "step": 1897
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.0500328834208295e-06,
      "loss": 0.6148,
      "step": 1898
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.041663616613477e-06,
      "loss": 0.6312,
      "step": 1899
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.033309525870717e-06,
      "loss": 0.6017,
      "step": 1900
    },
    {
      "epoch": 2.4,
      "eval_loss": 0.6589770317077637,
      "eval_runtime": 182.4932,
      "eval_samples_per_second": 11.31,
      "eval_steps_per_second": 0.093,
      "step": 1900
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.024970627123295e-06,
      "loss": 0.5802,
      "step": 1901
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.0166469362729868e-06,
      "loss": 0.6035,
      "step": 1902
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.008338469192561e-06,
      "loss": 0.6037,
      "step": 1903
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.0000452417257633e-06,
      "loss": 0.6283,
      "step": 1904
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.991767269687278e-06,
      "loss": 0.6118,
      "step": 1905
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.9835045688626897e-06,
      "loss": 0.5971,
      "step": 1906
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.975257155008472e-06,
      "loss": 0.6139,
      "step": 1907
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.967025043851939e-06,
      "loss": 0.6256,
      "step": 1908
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.95880825109123e-06,
      "loss": 0.615,
      "step": 1909
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9506067923952676e-06,
      "loss": 0.6231,
      "step": 1910
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9424206834037416e-06,
      "loss": 0.5874,
      "step": 1911
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.934249939727059e-06,
      "loss": 0.612,
      "step": 1912
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.926094576946332e-06,
      "loss": 0.5928,
      "step": 1913
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9179546106133497e-06,
      "loss": 0.6271,
      "step": 1914
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 0.5851,
      "step": 1915
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.9017209293508975e-06,
      "loss": 0.5964,
      "step": 1916
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8936272453780768e-06,
      "loss": 0.6446,
      "step": 1917
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8855490197662263e-06,
      "loss": 0.5981,
      "step": 1918
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8774862679200322e-06,
      "loss": 0.6069,
      "step": 1919
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8694390052146737e-06,
      "loss": 0.605,
      "step": 1920
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8614072469957966e-06,
      "loss": 0.6192,
      "step": 1921
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8533910085794714e-06,
      "loss": 0.6189,
      "step": 1922
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8453903052521783e-06,
      "loss": 0.5942,
      "step": 1923
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.8374051522707736e-06,
      "loss": 0.6314,
      "step": 1924
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.8294355648624607e-06,
      "loss": 0.6101,
      "step": 1925
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.8214815582247613e-06,
      "loss": 0.5876,
      "step": 1926
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.8135431475254783e-06,
      "loss": 0.6102,
      "step": 1927
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.8056203479026812e-06,
      "loss": 0.6115,
      "step": 1928
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.7977131744646724e-06,
      "loss": 0.6048,
      "step": 1929
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.789821642289945e-06,
      "loss": 0.6086,
      "step": 1930
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.7819457664271755e-06,
      "loss": 0.6007,
      "step": 1931
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.7740855618951814e-06,
      "loss": 0.6056,
      "step": 1932
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7662410436828981e-06,
      "loss": 0.6214,
      "step": 1933
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.758412226749343e-06,
      "loss": 0.6333,
      "step": 1934
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.750599126023591e-06,
      "loss": 0.6158,
      "step": 1935
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7428017564047594e-06,
      "loss": 0.6188,
      "step": 1936
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7350201327619532e-06,
      "loss": 0.6097,
      "step": 1937
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7272542699342598e-06,
      "loss": 0.6257,
      "step": 1938
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7195041827307024e-06,
      "loss": 0.6149,
      "step": 1939
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7117698859302357e-06,
      "loss": 0.5937,
      "step": 1940
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.7040513942816905e-06,
      "loss": 0.6363,
      "step": 1941
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.696348722503759e-06,
      "loss": 0.5823,
      "step": 1942
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.6886618852849723e-06,
      "loss": 0.6132,
      "step": 1943
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.6809908972836607e-06,
      "loss": 0.6217,
      "step": 1944
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.6733357731279375e-06,
      "loss": 0.6237,
      "step": 1945
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.6656965274156545e-06,
      "loss": 0.6394,
      "step": 1946
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.6580731747143907e-06,
      "loss": 0.6008,
      "step": 1947
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6504657295614202e-06,
      "loss": 0.6144,
      "step": 1948
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.642874206463676e-06,
      "loss": 0.6264,
      "step": 1949
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6352986198977327e-06,
      "loss": 0.6286,
      "step": 1950
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6277389843097746e-06,
      "loss": 0.6223,
      "step": 1951
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6201953141155712e-06,
      "loss": 0.6145,
      "step": 1952
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6126676237004401e-06,
      "loss": 0.6161,
      "step": 1953
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.6051559274192275e-06,
      "loss": 0.607,
      "step": 1954
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.5976602395962892e-06,
      "loss": 0.6093,
      "step": 1955
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.5901805745254417e-06,
      "loss": 0.617,
      "step": 1956
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.5827169464699576e-06,
      "loss": 0.6025,
      "step": 1957
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.5752693696625165e-06,
      "loss": 0.5815,
      "step": 1958
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.567837858305199e-06,
      "loss": 0.6061,
      "step": 1959
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.560422426569449e-06,
      "loss": 0.5933,
      "step": 1960
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.5530230885960386e-06,
      "loss": 0.612,
      "step": 1961
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.5456398584950594e-06,
      "loss": 0.6212,
      "step": 1962
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.5382727503458861e-06,
      "loss": 0.63,
      "step": 1963
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5309217781971419e-06,
      "loss": 0.6188,
      "step": 1964
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.523586956066686e-06,
      "loss": 0.5999,
      "step": 1965
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.516268297941581e-06,
      "loss": 0.6053,
      "step": 1966
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5089658177780653e-06,
      "loss": 0.617,
      "step": 1967
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5016795295015207e-06,
      "loss": 0.6347,
      "step": 1968
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.4944094470064585e-06,
      "loss": 0.594,
      "step": 1969
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.4871555841564889e-06,
      "loss": 0.6,
      "step": 1970
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.4799179547842823e-06,
      "loss": 0.6388,
      "step": 1971
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4726965726915642e-06,
      "loss": 0.6008,
      "step": 1972
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4654914516490638e-06,
      "loss": 0.5676,
      "step": 1973
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.458302605396521e-06,
      "loss": 0.6249,
      "step": 1974
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4511300476426227e-06,
      "loss": 0.6158,
      "step": 1975
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.443973792065e-06,
      "loss": 0.6079,
      "step": 1976
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.436833852310201e-06,
      "loss": 0.6224,
      "step": 1977
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4297102419936559e-06,
      "loss": 0.5837,
      "step": 1978
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4226029746996616e-06,
      "loss": 0.6091,
      "step": 1979
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.4155120639813392e-06,
      "loss": 0.5884,
      "step": 1980
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.4084375233606285e-06,
      "loss": 0.6041,
      "step": 1981
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.4013793663282515e-06,
      "loss": 0.5901,
      "step": 1982
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.3943376063436808e-06,
      "loss": 0.6314,
      "step": 1983
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.387312256835127e-06,
      "loss": 0.5958,
      "step": 1984
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.3803033311995072e-06,
      "loss": 0.5939,
      "step": 1985
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.3733108428024167e-06,
      "loss": 0.61,
      "step": 1986
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.3663348049781066e-06,
      "loss": 0.5917,
      "step": 1987
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3593752310294527e-06,
      "loss": 0.6346,
      "step": 1988
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3524321342279478e-06,
      "loss": 0.6366,
      "step": 1989
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.345505527813652e-06,
      "loss": 0.5911,
      "step": 1990
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3385954249951861e-06,
      "loss": 0.5967,
      "step": 1991
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3317018389496927e-06,
      "loss": 0.5916,
      "step": 1992
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3248247828228244e-06,
      "loss": 0.6376,
      "step": 1993
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3179642697287132e-06,
      "loss": 0.6028,
      "step": 1994
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.311120312749935e-06,
      "loss": 0.5987,
      "step": 1995
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.3042929249375046e-06,
      "loss": 0.5972,
      "step": 1996
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2974821193108356e-06,
      "loss": 0.5715,
      "step": 1997
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2906879088577218e-06,
      "loss": 0.6023,
      "step": 1998
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2839103065343084e-06,
      "loss": 0.6276,
      "step": 1999
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2771493252650723e-06,
      "loss": 0.6096,
      "step": 2000
    },
    {
      "epoch": 2.53,
      "eval_loss": 0.6545730829238892,
      "eval_runtime": 182.4816,
      "eval_samples_per_second": 11.311,
      "eval_steps_per_second": 0.093,
      "step": 2000
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.2704049779427973e-06,
      "loss": 0.6033,
      "step": 2001
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.26367727742854e-06,
      "loss": 0.5787,
      "step": 2002
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.256966236551619e-06,
      "loss": 0.6056,
      "step": 2003
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2502718681095838e-06,
      "loss": 0.615,
      "step": 2004
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2435941848681864e-06,
      "loss": 0.6033,
      "step": 2005
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2369331995613664e-06,
      "loss": 0.6082,
      "step": 2006
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.230288924891212e-06,
      "loss": 0.6079,
      "step": 2007
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2236613735279611e-06,
      "loss": 0.6155,
      "step": 2008
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2170505581099446e-06,
      "loss": 0.5905,
      "step": 2009
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2104564912435924e-06,
      "loss": 0.6019,
      "step": 2010
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2038791855033839e-06,
      "loss": 0.5959,
      "step": 2011
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1973186534318449e-06,
      "loss": 0.6048,
      "step": 2012
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1907749075395147e-06,
      "loss": 0.6265,
      "step": 2013
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.184247960304914e-06,
      "loss": 0.5928,
      "step": 2014
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1777378241745385e-06,
      "loss": 0.6094,
      "step": 2015
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1712445115628245e-06,
      "loss": 0.6211,
      "step": 2016
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1647680348521206e-06,
      "loss": 0.6547,
      "step": 2017
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.1583084063926775e-06,
      "loss": 0.6189,
      "step": 2018
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.151865638502615e-06,
      "loss": 0.6134,
      "step": 2019
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1454397434679022e-06,
      "loss": 0.6153,
      "step": 2020
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1390307335423279e-06,
      "loss": 0.6193,
      "step": 2021
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1326386209474827e-06,
      "loss": 0.6097,
      "step": 2022
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.126263417872745e-06,
      "loss": 0.6449,
      "step": 2023
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1199051364752334e-06,
      "loss": 0.6107,
      "step": 2024
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1135637888798101e-06,
      "loss": 0.5905,
      "step": 2025
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1072393871790365e-06,
      "loss": 0.6156,
      "step": 2026
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.1009319434331623e-06,
      "loss": 0.6036,
      "step": 2027
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0946414696701048e-06,
      "loss": 0.6389,
      "step": 2028
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0883679778854095e-06,
      "loss": 0.6145,
      "step": 2029
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0821114800422482e-06,
      "loss": 0.6041,
      "step": 2030
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.075871988071383e-06,
      "loss": 0.6159,
      "step": 2031
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0696495138711472e-06,
      "loss": 0.6034,
      "step": 2032
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0634440693074176e-06,
      "loss": 0.592,
      "step": 2033
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.0572556662136036e-06,
      "loss": 0.6382,
      "step": 2034
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.0510843163906148e-06,
      "loss": 0.612,
      "step": 2035
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.0449300316068378e-06,
      "loss": 0.6204,
      "step": 2036
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.038792823598125e-06,
      "loss": 0.6215,
      "step": 2037
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.032672704067752e-06,
      "loss": 0.6191,
      "step": 2038
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.0265696846864249e-06,
      "loss": 0.6062,
      "step": 2039
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.020483777092226e-06,
      "loss": 0.6,
      "step": 2040
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.014414992890611e-06,
      "loss": 0.6035,
      "step": 2041
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.008363343654385e-06,
      "loss": 0.5858,
      "step": 2042
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.0023288409236743e-06,
      "loss": 0.6036,
      "step": 2043
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.96311496205914e-07,
      "loss": 0.6188,
      "step": 2044
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.903113209758098e-07,
      "loss": 0.5998,
      "step": 2045
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.843283266753333e-07,
      "loss": 0.6034,
      "step": 2046
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.783625247136952e-07,
      "loss": 0.6168,
      "step": 2047
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.724139264673116e-07,
      "loss": 0.5904,
      "step": 2048
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.66482543279803e-07,
      "loss": 0.6295,
      "step": 2049
    },
    {
      "epoch": 2.59,
      "learning_rate": 9.605683864619574e-07,
      "loss": 0.6291,
      "step": 2050
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.546714672917113e-07,
      "loss": 0.5899,
      "step": 2051
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.487917970141324e-07,
      "loss": 0.5957,
      "step": 2052
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.429293868413914e-07,
      "loss": 0.6231,
      "step": 2053
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.370842479527553e-07,
      "loss": 0.5893,
      "step": 2054
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.312563914945461e-07,
      "loss": 0.6263,
      "step": 2055
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.254458285801326e-07,
      "loss": 0.603,
      "step": 2056
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.19652570289905e-07,
      "loss": 0.5796,
      "step": 2057
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.138766276712552e-07,
      "loss": 0.6228,
      "step": 2058
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.081180117385591e-07,
      "loss": 0.5863,
      "step": 2059
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.023767334731426e-07,
      "loss": 0.6124,
      "step": 2060
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.966528038232769e-07,
      "loss": 0.6158,
      "step": 2061
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.909462337041508e-07,
      "loss": 0.6114,
      "step": 2062
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.852570339978439e-07,
      "loss": 0.5954,
      "step": 2063
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.795852155533158e-07,
      "loss": 0.596,
      "step": 2064
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.739307891863813e-07,
      "loss": 0.6083,
      "step": 2065
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.682937656796886e-07,
      "loss": 0.6015,
      "step": 2066
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.626741557826979e-07,
      "loss": 0.6108,
      "step": 2067
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.570719702116647e-07,
      "loss": 0.5797,
      "step": 2068
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.514872196496182e-07,
      "loss": 0.6457,
      "step": 2069
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.459199147463371e-07,
      "loss": 0.596,
      "step": 2070
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.403700661183356e-07,
      "loss": 0.6209,
      "step": 2071
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.348376843488348e-07,
      "loss": 0.6367,
      "step": 2072
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.293227799877579e-07,
      "loss": 0.5878,
      "step": 2073
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.23825363551688e-07,
      "loss": 0.6349,
      "step": 2074
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.183454455238638e-07,
      "loss": 0.602,
      "step": 2075
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.128830363541574e-07,
      "loss": 0.6342,
      "step": 2076
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.074381464590531e-07,
      "loss": 0.6301,
      "step": 2077
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.020107862216264e-07,
      "loss": 0.5976,
      "step": 2078
    },
    {
      "epoch": 2.63,
      "learning_rate": 7.966009659915208e-07,
      "loss": 0.6122,
      "step": 2079
    },
    {
      "epoch": 2.63,
      "learning_rate": 7.912086960849374e-07,
      "loss": 0.6116,
      "step": 2080
    },
    {
      "epoch": 2.63,
      "learning_rate": 7.858339867846077e-07,
      "loss": 0.6267,
      "step": 2081
    },
    {
      "epoch": 2.63,
      "learning_rate": 7.804768483397752e-07,
      "loss": 0.6129,
      "step": 2082
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.75137290966177e-07,
      "loss": 0.6192,
      "step": 2083
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.698153248460271e-07,
      "loss": 0.6278,
      "step": 2084
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.645109601279921e-07,
      "loss": 0.5773,
      "step": 2085
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.592242069271738e-07,
      "loss": 0.5973,
      "step": 2086
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.539550753250846e-07,
      "loss": 0.6303,
      "step": 2087
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.487035753696459e-07,
      "loss": 0.6079,
      "step": 2088
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.434697170751459e-07,
      "loss": 0.5828,
      "step": 2089
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.382535104222366e-07,
      "loss": 0.6238,
      "step": 2090
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.330549653579045e-07,
      "loss": 0.587,
      "step": 2091
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.278740917954608e-07,
      "loss": 0.6215,
      "step": 2092
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.227108996145194e-07,
      "loss": 0.5941,
      "step": 2093
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.175653986609721e-07,
      "loss": 0.6018,
      "step": 2094
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.124375987469767e-07,
      "loss": 0.5878,
      "step": 2095
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.07327509650938e-07,
      "loss": 0.5906,
      "step": 2096
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.022351411174866e-07,
      "loss": 0.6161,
      "step": 2097
    },
    {
      "epoch": 2.65,
      "learning_rate": 6.97160502857458e-07,
      "loss": 0.5976,
      "step": 2098
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.921036045478791e-07,
      "loss": 0.6148,
      "step": 2099
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.870644558319528e-07,
      "loss": 0.6141,
      "step": 2100
    },
    {
      "epoch": 2.66,
      "eval_loss": 0.6509197354316711,
      "eval_runtime": 182.452,
      "eval_samples_per_second": 11.313,
      "eval_steps_per_second": 0.093,
      "step": 2100
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.820430663190236e-07,
      "loss": 0.6068,
      "step": 2101
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.770394455845797e-07,
      "loss": 0.5804,
      "step": 2102
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.72053603170224e-07,
      "loss": 0.5974,
      "step": 2103
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.670855485836525e-07,
      "loss": 0.6078,
      "step": 2104
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.621352912986468e-07,
      "loss": 0.5638,
      "step": 2105
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.572028407550412e-07,
      "loss": 0.5961,
      "step": 2106
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.522882063587276e-07,
      "loss": 0.6181,
      "step": 2107
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.473913974816092e-07,
      "loss": 0.6027,
      "step": 2108
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.425124234616087e-07,
      "loss": 0.624,
      "step": 2109
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.37651293602628e-07,
      "loss": 0.5986,
      "step": 2110
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.32808017174551e-07,
      "loss": 0.6019,
      "step": 2111
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.279826034132108e-07,
      "loss": 0.6259,
      "step": 2112
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.231750615203769e-07,
      "loss": 0.6136,
      "step": 2113
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.183854006637424e-07,
      "loss": 0.6065,
      "step": 2114
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.136136299768991e-07,
      "loss": 0.6322,
      "step": 2115
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.088597585593236e-07,
      "loss": 0.596,
      "step": 2116
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.041237954763601e-07,
      "loss": 0.6188,
      "step": 2117
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.994057497592032e-07,
      "loss": 0.6245,
      "step": 2118
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.947056304048826e-07,
      "loss": 0.5844,
      "step": 2119
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.900234463762367e-07,
      "loss": 0.6082,
      "step": 2120
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.853592066019054e-07,
      "loss": 0.5958,
      "step": 2121
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.807129199763172e-07,
      "loss": 0.5739,
      "step": 2122
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.760845953596527e-07,
      "loss": 0.6126,
      "step": 2123
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.714742415778496e-07,
      "loss": 0.586,
      "step": 2124
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.668818674225684e-07,
      "loss": 0.6182,
      "step": 2125
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.623074816511942e-07,
      "loss": 0.6472,
      "step": 2126
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.577510929868002e-07,
      "loss": 0.5928,
      "step": 2127
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.532127101181406e-07,
      "loss": 0.5991,
      "step": 2128
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.486923416996393e-07,
      "loss": 0.6047,
      "step": 2129
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.441899963513631e-07,
      "loss": 0.6057,
      "step": 2130
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.397056826590141e-07,
      "loss": 0.5934,
      "step": 2131
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.352394091739022e-07,
      "loss": 0.6111,
      "step": 2132
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.307911844129421e-07,
      "loss": 0.6227,
      "step": 2133
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.263610168586308e-07,
      "loss": 0.5858,
      "step": 2134
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.219489149590251e-07,
      "loss": 0.6042,
      "step": 2135
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.175548871277358e-07,
      "loss": 0.592,
      "step": 2136
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.131789417439104e-07,
      "loss": 0.5922,
      "step": 2137
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.088210871522104e-07,
      "loss": 0.6235,
      "step": 2138
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.044813316627994e-07,
      "loss": 0.5895,
      "step": 2139
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.001596835513256e-07,
      "loss": 0.6131,
      "step": 2140
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.958561510589122e-07,
      "loss": 0.6275,
      "step": 2141
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.915707423921346e-07,
      "loss": 0.5888,
      "step": 2142
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.87303465723008e-07,
      "loss": 0.6094,
      "step": 2143
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.830543291889667e-07,
      "loss": 0.6063,
      "step": 2144
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.788233408928588e-07,
      "loss": 0.6107,
      "step": 2145
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.746105089029229e-07,
      "loss": 0.5762,
      "step": 2146
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.704158412527715e-07,
      "loss": 0.5916,
      "step": 2147
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.6623934594138165e-07,
      "loss": 0.5858,
      "step": 2148
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.620810309330803e-07,
      "loss": 0.6041,
      "step": 2149
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.5794090415751666e-07,
      "loss": 0.613,
      "step": 2150
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.538189735096643e-07,
      "loss": 0.5944,
      "step": 2151
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.497152468497934e-07,
      "loss": 0.6204,
      "step": 2152
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.4562973200346413e-07,
      "loss": 0.6121,
      "step": 2153
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.4156243676150235e-07,
      "loss": 0.6107,
      "step": 2154
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.3751336887999597e-07,
      "loss": 0.5868,
      "step": 2155
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.334825360802686e-07,
      "loss": 0.5925,
      "step": 2156
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.2946994604887715e-07,
      "loss": 0.6021,
      "step": 2157
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.2547560643758534e-07,
      "loss": 0.6234,
      "step": 2158
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.214995248633569e-07,
      "loss": 0.5885,
      "step": 2159
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.1754170890833777e-07,
      "loss": 0.6204,
      "step": 2160
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.136021661198453e-07,
      "loss": 0.6155,
      "step": 2161
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.0968090401034444e-07,
      "loss": 0.6048,
      "step": 2162
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.05777930057446e-07,
      "loss": 0.6198,
      "step": 2163
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.018932517038854e-07,
      "loss": 0.5938,
      "step": 2164
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.980268763575079e-07,
      "loss": 0.58,
      "step": 2165
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.9417881139125593e-07,
      "loss": 0.639,
      "step": 2166
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.903490641431573e-07,
      "loss": 0.5888,
      "step": 2167
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.8653764191630694e-07,
      "loss": 0.6084,
      "step": 2168
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.827445519788553e-07,
      "loss": 0.5995,
      "step": 2169
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.7896980156399533e-07,
      "loss": 0.6106,
      "step": 2170
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.752133978699457e-07,
      "loss": 0.6372,
      "step": 2171
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.714753480599431e-07,
      "loss": 0.5853,
      "step": 2172
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.6775565926222e-07,
      "loss": 0.5938,
      "step": 2173
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.6405433856999684e-07,
      "loss": 0.5609,
      "step": 2174
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.603713930414676e-07,
      "loss": 0.5982,
      "step": 2175
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.5670682969978665e-07,
      "loss": 0.6106,
      "step": 2176
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.5306065553305404e-07,
      "loss": 0.6308,
      "step": 2177
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.494328774943023e-07,
      "loss": 0.58,
      "step": 2178
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.458235025014833e-07,
      "loss": 0.6274,
      "step": 2179
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.42232537437458e-07,
      "loss": 0.5998,
      "step": 2180
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.3865998914997645e-07,
      "loss": 0.5813,
      "step": 2181
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.351058644516736e-07,
      "loss": 0.5797,
      "step": 2182
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.31570170120048e-07,
      "loss": 0.6156,
      "step": 2183
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.2805291289745635e-07,
      "loss": 0.613,
      "step": 2184
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.245540994910934e-07,
      "loss": 0.6243,
      "step": 2185
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.21073736572981e-07,
      "loss": 0.6056,
      "step": 2186
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.1761183077996585e-07,
      "loss": 0.5923,
      "step": 2187
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.1416838871368925e-07,
      "loss": 0.6168,
      "step": 2188
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.107434169405865e-07,
      "loss": 0.5913,
      "step": 2189
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.073369219918698e-07,
      "loss": 0.6394,
      "step": 2190
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.039489103635196e-07,
      "loss": 0.606,
      "step": 2191
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.0057938851627e-07,
      "loss": 0.5723,
      "step": 2192
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.972283628755912e-07,
      "loss": 0.5858,
      "step": 2193
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.9389583983168826e-07,
      "loss": 0.6015,
      "step": 2194
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.905818257394799e-07,
      "loss": 0.5985,
      "step": 2195
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.872863269185899e-07,
      "loss": 0.6201,
      "step": 2196
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.840093496533336e-07,
      "loss": 0.6153,
      "step": 2197
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.8075090019271e-07,
      "loss": 0.5984,
      "step": 2198
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.775109847503832e-07,
      "loss": 0.591,
      "step": 2199
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.742896095046732e-07,
      "loss": 0.6081,
      "step": 2200
    },
    {
      "epoch": 2.78,
      "eval_loss": 0.6491526961326599,
      "eval_runtime": 182.5497,
      "eval_samples_per_second": 11.307,
      "eval_steps_per_second": 0.093,
      "step": 2200
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.7108678059855064e-07,
      "loss": 0.5816,
      "step": 2201
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.679025041396155e-07,
      "loss": 0.5951,
      "step": 2202
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.6473678620008714e-07,
      "loss": 0.603,
      "step": 2203
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.6158963281680106e-07,
      "loss": 0.6009,
      "step": 2204
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.584610499911833e-07,
      "loss": 0.6071,
      "step": 2205
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.553510436892559e-07,
      "loss": 0.5835,
      "step": 2206
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.522596198416094e-07,
      "loss": 0.6417,
      "step": 2207
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.4918678434340373e-07,
      "loss": 0.6033,
      "step": 2208
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.461325430543482e-07,
      "loss": 0.6185,
      "step": 2209
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.4309690179869503e-07,
      "loss": 0.616,
      "step": 2210
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.400798663652304e-07,
      "loss": 0.6278,
      "step": 2211
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.370814425072554e-07,
      "loss": 0.5982,
      "step": 2212
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.3410163594258296e-07,
      "loss": 0.5754,
      "step": 2213
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.3114045235352433e-07,
      "loss": 0.6113,
      "step": 2214
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.2819789738687482e-07,
      "loss": 0.5694,
      "step": 2215
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.2527397665391026e-07,
      "loss": 0.6064,
      "step": 2216
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.2236869573036946e-07,
      "loss": 0.6022,
      "step": 2217
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.1948206015644845e-07,
      "loss": 0.5869,
      "step": 2218
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.1661407543678404e-07,
      "loss": 0.6123,
      "step": 2219
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.1376474704044693e-07,
      "loss": 0.5885,
      "step": 2220
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.1093408040093855e-07,
      "loss": 0.6109,
      "step": 2221
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.081220809161655e-07,
      "loss": 0.5966,
      "step": 2222
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.0532875394844053e-07,
      "loss": 0.5928,
      "step": 2223
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.0255410482446837e-07,
      "loss": 0.605,
      "step": 2224
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.9979813883533762e-07,
      "loss": 0.6086,
      "step": 2225
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.9706086123650768e-07,
      "loss": 0.6119,
      "step": 2226
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.9434227724779987e-07,
      "loss": 0.5927,
      "step": 2227
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.9164239205339053e-07,
      "loss": 0.5947,
      "step": 2228
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.8896121080179574e-07,
      "loss": 0.5951,
      "step": 2229
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.8629873860586567e-07,
      "loss": 0.6179,
      "step": 2230
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.836549805427734e-07,
      "loss": 0.5795,
      "step": 2231
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.8102994165400512e-07,
      "loss": 0.5849,
      "step": 2232
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.784236269453521e-07,
      "loss": 0.6054,
      "step": 2233
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.7583604138689535e-07,
      "loss": 0.5903,
      "step": 2234
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.7326718991300563e-07,
      "loss": 0.5721,
      "step": 2235
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.7071707742232546e-07,
      "loss": 0.6144,
      "step": 2236
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.681857087777672e-07,
      "loss": 0.6233,
      "step": 2237
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.6567308880649501e-07,
      "loss": 0.5903,
      "step": 2238
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.631792222999229e-07,
      "loss": 0.5704,
      "step": 2239
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.6070411401370335e-07,
      "loss": 0.5862,
      "step": 2240
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.582477686677175e-07,
      "loss": 0.6076,
      "step": 2241
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.5581019094606853e-07,
      "loss": 0.6283,
      "step": 2242
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.53391385497067e-07,
      "loss": 0.6195,
      "step": 2243
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.5099135693322776e-07,
      "loss": 0.5973,
      "step": 2244
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.4861010983126202e-07,
      "loss": 0.5993,
      "step": 2245
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.4624764873206188e-07,
      "loss": 0.6093,
      "step": 2246
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.4390397814069812e-07,
      "loss": 0.5902,
      "step": 2247
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.4157910252640795e-07,
      "loss": 0.622,
      "step": 2248
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.392730263225883e-07,
      "loss": 0.6082,
      "step": 2249
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.3698575392678492e-07,
      "loss": 0.6191,
      "step": 2250
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.3471728970068986e-07,
      "loss": 0.5774,
      "step": 2251
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.324676379701273e-07,
      "loss": 0.5886,
      "step": 2252
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.3023680302504338e-07,
      "loss": 0.6247,
      "step": 2253
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.2802478911950744e-07,
      "loss": 0.5943,
      "step": 2254
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.258316004716953e-07,
      "loss": 0.6048,
      "step": 2255
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.2365724126388367e-07,
      "loss": 0.6133,
      "step": 2256
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.2150171564244474e-07,
      "loss": 0.5906,
      "step": 2257
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.1936502771783488e-07,
      "loss": 0.5783,
      "step": 2258
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.1724718156458814e-07,
      "loss": 0.5881,
      "step": 2259
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.1514818122130844e-07,
      "loss": 0.606,
      "step": 2260
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.1306803069066175e-07,
      "loss": 0.5945,
      "step": 2261
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.1100673393936945e-07,
      "loss": 0.6139,
      "step": 2262
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.089642948982006e-07,
      "loss": 0.5978,
      "step": 2263
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.0694071746196077e-07,
      "loss": 0.618,
      "step": 2264
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0493600548948879e-07,
      "loss": 0.5898,
      "step": 2265
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0295016280365111e-07,
      "loss": 0.5819,
      "step": 2266
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0098319319132743e-07,
      "loss": 0.6206,
      "step": 2267
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.903510040340958e-08,
      "loss": 0.5645,
      "step": 2268
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.710588815479371e-08,
      "loss": 0.6079,
      "step": 2269
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.519556012436815e-08,
      "loss": 0.6111,
      "step": 2270
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.330411995501442e-08,
      "loss": 0.5906,
      "step": 2271
    },
    {
      "epoch": 2.88,
      "learning_rate": 9.143157125359514e-08,
      "loss": 0.6095,
      "step": 2272
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.957791759094392e-08,
      "loss": 0.5739,
      "step": 2273
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.774316250186654e-08,
      "loss": 0.5961,
      "step": 2274
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.592730948513205e-08,
      "loss": 0.6023,
      "step": 2275
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.413036200346058e-08,
      "loss": 0.6144,
      "step": 2276
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.235232348352107e-08,
      "loss": 0.5969,
      "step": 2277
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.059319731592574e-08,
      "loss": 0.6069,
      "step": 2278
    },
    {
      "epoch": 2.88,
      "learning_rate": 7.885298685522235e-08,
      "loss": 0.6037,
      "step": 2279
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.71316954198853e-08,
      "loss": 0.6041,
      "step": 2280
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.542932629231337e-08,
      "loss": 0.5856,
      "step": 2281
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.374588271882088e-08,
      "loss": 0.5707,
      "step": 2282
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.208136790963326e-08,
      "loss": 0.6139,
      "step": 2283
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.043578503888037e-08,
      "loss": 0.5994,
      "step": 2284
    },
    {
      "epoch": 2.89,
      "learning_rate": 6.880913724458538e-08,
      "loss": 0.6194,
      "step": 2285
    },
    {
      "epoch": 2.89,
      "learning_rate": 6.720142762867032e-08,
      "loss": 0.5942,
      "step": 2286
    },
    {
      "epoch": 2.89,
      "learning_rate": 6.561265925693838e-08,
      "loss": 0.6036,
      "step": 2287
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.404283515907494e-08,
      "loss": 0.5799,
      "step": 2288
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.249195832863875e-08,
      "loss": 0.5938,
      "step": 2289
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.096003172305742e-08,
      "loss": 0.5944,
      "step": 2290
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.944705826362418e-08,
      "loss": 0.5926,
      "step": 2291
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.795304083548559e-08,
      "loss": 0.6317,
      "step": 2292
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.647798228764156e-08,
      "loss": 0.5976,
      "step": 2293
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.5021885432942065e-08,
      "loss": 0.607,
      "step": 2294
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.3584753048073756e-08,
      "loss": 0.6,
      "step": 2295
    },
    {
      "epoch": 2.91,
      "learning_rate": 5.216658787355999e-08,
      "loss": 0.6031,
      "step": 2296
    },
    {
      "epoch": 2.91,
      "learning_rate": 5.076739261375751e-08,
      "loss": 0.5849,
      "step": 2297
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.9387169936845335e-08,
      "loss": 0.5814,
      "step": 2298
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.802592247482585e-08,
      "loss": 0.584,
      "step": 2299
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.6683652823513725e-08,
      "loss": 0.6161,
      "step": 2300
    },
    {
      "epoch": 2.91,
      "eval_loss": 0.6485294103622437,
      "eval_runtime": 182.6801,
      "eval_samples_per_second": 11.298,
      "eval_steps_per_second": 0.093,
      "step": 2300
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.5360363542537034e-08,
      "loss": 0.5574,
      "step": 2301
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.405605715532724e-08,
      "loss": 0.6231,
      "step": 2302
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.27707361491192e-08,
      "loss": 0.6428,
      "step": 2303
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.150440297493896e-08,
      "loss": 0.5812,
      "step": 2304
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.025706004760932e-08,
      "loss": 0.6105,
      "step": 2305
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.9028709745738694e-08,
      "loss": 0.5831,
      "step": 2306
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.7819354411713364e-08,
      "loss": 0.6181,
      "step": 2307
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.662899635170192e-08,
      "loss": 0.598,
      "step": 2308
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.5457637835646375e-08,
      "loss": 0.6113,
      "step": 2309
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.430528109725439e-08,
      "loss": 0.5891,
      "step": 2310
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.3171928334001513e-08,
      "loss": 0.5958,
      "step": 2311
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.205758170712225e-08,
      "loss": 0.5887,
      "step": 2312
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.096224334160791e-08,
      "loss": 0.6268,
      "step": 2313
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.988591532620322e-08,
      "loss": 0.5879,
      "step": 2314
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.8828599713398575e-08,
      "loss": 0.605,
      "step": 2315
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.779029851943227e-08,
      "loss": 0.5883,
      "step": 2316
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.677101372428159e-08,
      "loss": 0.5889,
      "step": 2317
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.5770747271659512e-08,
      "loss": 0.5691,
      "step": 2318
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.4789501069014675e-08,
      "loss": 0.6048,
      "step": 2319
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.3827276987524738e-08,
      "loss": 0.6122,
      "step": 2320
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.2884076862089712e-08,
      "loss": 0.6182,
      "step": 2321
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.1959902491338615e-08,
      "loss": 0.6008,
      "step": 2322
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.1054755637612833e-08,
      "loss": 0.6092,
      "step": 2323
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.0168638026974996e-08,
      "loss": 0.5915,
      "step": 2324
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.9301551349195648e-08,
      "loss": 0.6203,
      "step": 2325
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.845349725775769e-08,
      "loss": 0.5633,
      "step": 2326
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.7624477369847516e-08,
      "loss": 0.597,
      "step": 2327
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.6814493266357202e-08,
      "loss": 0.6055,
      "step": 2328
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.6023546491877872e-08,
      "loss": 0.6239,
      "step": 2329
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.5251638554694137e-08,
      "loss": 0.6022,
      "step": 2330
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.449877092679075e-08,
      "loss": 0.6191,
      "step": 2331
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.3764945043838185e-08,
      "loss": 0.6372,
      "step": 2332
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.3050162305199288e-08,
      "loss": 0.6073,
      "step": 2333
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.2354424073921512e-08,
      "loss": 0.5926,
      "step": 2334
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.1677731676733584e-08,
      "loss": 0.6076,
      "step": 2335
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.1020086404049946e-08,
      "loss": 0.5765,
      "step": 2336
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.0381489509957432e-08,
      "loss": 0.5883,
      "step": 2337
    },
    {
      "epoch": 2.96,
      "learning_rate": 9.76194221222193e-09,
      "loss": 0.6283,
      "step": 2338
    },
    {
      "epoch": 2.96,
      "learning_rate": 9.161445692283943e-09,
      "loss": 0.6014,
      "step": 2339
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.580001095253032e-09,
      "loss": 0.6108,
      "step": 2340
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.017609529908933e-09,
      "loss": 0.6172,
      "step": 2341
    },
    {
      "epoch": 2.96,
      "learning_rate": 7.474272068698219e-09,
      "loss": 0.5756,
      "step": 2342
    },
    {
      "epoch": 2.96,
      "learning_rate": 6.949989747733199e-09,
      "loss": 0.5994,
      "step": 2343
    },
    {
      "epoch": 2.97,
      "learning_rate": 6.444763566786361e-09,
      "loss": 0.593,
      "step": 2344
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.958594489295921e-09,
      "loss": 0.6244,
      "step": 2345
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.491483442354728e-09,
      "loss": 0.6048,
      "step": 2346
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.0434313167158075e-09,
      "loss": 0.592,
      "step": 2347
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.614438966787926e-09,
      "loss": 0.6006,
      "step": 2348
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.204507210633368e-09,
      "loss": 0.5974,
      "step": 2349
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.8136368299668266e-09,
      "loss": 0.5981,
      "step": 2350
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.441828570155403e-09,
      "loss": 0.6104,
      "step": 2351
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.0890831402130563e-09,
      "loss": 0.6247,
      "step": 2352
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.7554012128050423e-09,
      "loss": 0.6071,
      "step": 2353
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.4407834242412555e-09,
      "loss": 0.5978,
      "step": 2354
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.145230374481777e-09,
      "loss": 0.5828,
      "step": 2355
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.8687426271246646e-09,
      "loss": 0.6091,
      "step": 2356
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.6113207094181626e-09,
      "loss": 0.6097,
      "step": 2357
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.3729651122484922e-09,
      "loss": 0.6005,
      "step": 2358
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.1536762901465103e-09,
      "loss": 0.6099,
      "step": 2359
    },
    {
      "epoch": 2.99,
      "learning_rate": 9.534546612810502e-10,
      "loss": 0.5994,
      "step": 2360
    },
    {
      "epoch": 2.99,
      "learning_rate": 7.723006074644712e-10,
      "loss": 0.5696,
      "step": 2361
    },
    {
      "epoch": 2.99,
      "learning_rate": 6.102144741459981e-10,
      "loss": 0.6089,
      "step": 2362
    },
    {
      "epoch": 2.99,
      "learning_rate": 4.671965704128312e-10,
      "loss": 0.6439,
      "step": 2363
    },
    {
      "epoch": 2.99,
      "learning_rate": 3.432471689934769e-10,
      "loss": 0.6171,
      "step": 2364
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.3836650624997627e-10,
      "loss": 0.5897,
      "step": 2365
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.5255478218456632e-10,
      "loss": 0.5963,
      "step": 2366
    },
    {
      "epoch": 3.0,
      "learning_rate": 8.581216043412888e-11,
      "loss": 0.6201,
      "step": 2367
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.8138768274631476e-11,
      "loss": 0.6036,
      "step": 2368
    },
    {
      "epoch": 3.0,
      "learning_rate": 9.534696614466043e-12,
      "loss": 0.5946,
      "step": 2369
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 0.5967,
      "step": 2370
    },
    {
      "epoch": 3.0,
      "step": 2370,
      "total_flos": 1.2127530440420491e+19,
      "train_loss": 0.7452560326721095,
      "train_runtime": 184850.0057,
      "train_samples_per_second": 1.641,
      "train_steps_per_second": 0.013
    }
  ],
  "max_steps": 2370,
  "num_train_epochs": 3,
  "total_flos": 1.2127530440420491e+19,
  "trial_name": null,
  "trial_params": null
}