{
  "best_metric": 1.3351472616195679,
  "best_model_checkpoint": "./pippa-lora/checkpoint-3500",
  "epoch": 2.7412280701754383,
  "eval_steps": 100,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 2.5e-06,
      "loss": 1.8809,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 5e-06,
      "loss": 1.9142,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 7.5e-06,
      "loss": 2.087,
      "step": 3
    },
    {
      "epoch": 0.0,
      "learning_rate": 1e-05,
      "loss": 1.906,
      "step": 4
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.25e-05,
      "loss": 2.062,
      "step": 5
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.5e-05,
      "loss": 1.9116,
      "step": 6
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.75e-05,
      "loss": 1.7963,
      "step": 7
    },
    {
      "epoch": 0.0,
      "learning_rate": 2e-05,
      "loss": 1.8519,
      "step": 8
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.25e-05,
      "loss": 1.9083,
      "step": 9
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.5e-05,
      "loss": 1.8595,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 1.832,
      "step": 11
    },
    {
      "epoch": 0.01,
      "learning_rate": 3e-05,
      "loss": 2.0534,
      "step": 12
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 1.8395,
      "step": 13
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.5e-05,
      "loss": 2.2138,
      "step": 14
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 1.981,
      "step": 15
    },
    {
      "epoch": 0.01,
      "learning_rate": 4e-05,
      "loss": 2.0866,
      "step": 16
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.25e-05,
      "loss": 1.7202,
      "step": 17
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.5e-05,
      "loss": 1.8915,
      "step": 18
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.75e-05,
      "loss": 1.9513,
      "step": 19
    },
    {
      "epoch": 0.01,
      "learning_rate": 5e-05,
      "loss": 1.9582,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.25e-05,
      "loss": 1.8452,
      "step": 21
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.500000000000001e-05,
      "loss": 1.8175,
      "step": 22
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.7499999999999995e-05,
      "loss": 1.9523,
      "step": 23
    },
    {
      "epoch": 0.01,
      "learning_rate": 6e-05,
      "loss": 1.8199,
      "step": 24
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.25e-05,
      "loss": 1.8729,
      "step": 25
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.500000000000001e-05,
      "loss": 1.77,
      "step": 26
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.750000000000001e-05,
      "loss": 1.7409,
      "step": 27
    },
    {
      "epoch": 0.02,
      "learning_rate": 7e-05,
      "loss": 2.0072,
      "step": 28
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.25e-05,
      "loss": 1.9466,
      "step": 29
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.500000000000001e-05,
      "loss": 1.8767,
      "step": 30
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.75e-05,
      "loss": 1.8352,
      "step": 31
    },
    {
      "epoch": 0.02,
      "learning_rate": 8e-05,
      "loss": 1.7732,
      "step": 32
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.25e-05,
      "loss": 1.7858,
      "step": 33
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.5e-05,
      "loss": 1.6732,
      "step": 34
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.75e-05,
      "loss": 1.903,
      "step": 35
    },
    {
      "epoch": 0.02,
      "learning_rate": 9e-05,
      "loss": 1.8067,
      "step": 36
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.250000000000001e-05,
      "loss": 1.8276,
      "step": 37
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.5e-05,
      "loss": 1.8657,
      "step": 38
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.75e-05,
      "loss": 1.7926,
      "step": 39
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001,
      "loss": 1.6443,
      "step": 40
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999999163780638e-05,
      "loss": 1.757,
      "step": 41
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.99999665512283e-05,
      "loss": 1.7316,
      "step": 42
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999992474027416e-05,
      "loss": 1.8997,
      "step": 43
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.999986620495793e-05,
      "loss": 1.7623,
      "step": 44
    },
    {
      "epoch": 0.02,
      "learning_rate": 9.99997909452992e-05,
      "loss": 1.9351,
      "step": 45
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999969896132315e-05,
      "loss": 1.9325,
      "step": 46
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999959025306054e-05,
      "loss": 2.0134,
      "step": 47
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999946482054772e-05,
      "loss": 1.9623,
      "step": 48
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999932266382668e-05,
      "loss": 1.7813,
      "step": 49
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999916378294493e-05,
      "loss": 1.6981,
      "step": 50
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999898817795563e-05,
      "loss": 1.7212,
      "step": 51
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999879584891754e-05,
      "loss": 1.8679,
      "step": 52
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999858679589496e-05,
      "loss": 1.8856,
      "step": 53
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999836101895783e-05,
      "loss": 1.7105,
      "step": 54
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999811851818165e-05,
      "loss": 1.7958,
      "step": 55
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999785929364756e-05,
      "loss": 1.8583,
      "step": 56
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999758334544226e-05,
      "loss": 1.728,
      "step": 57
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999729067365803e-05,
      "loss": 1.8283,
      "step": 58
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999698127839281e-05,
      "loss": 1.8551,
      "step": 59
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999665515975006e-05,
      "loss": 1.8186,
      "step": 60
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999631231783886e-05,
      "loss": 1.8333,
      "step": 61
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999595275277389e-05,
      "loss": 1.7514,
      "step": 62
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999557646467543e-05,
      "loss": 1.6589,
      "step": 63
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999518345366932e-05,
      "loss": 1.688,
      "step": 64
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999477371988703e-05,
      "loss": 1.8049,
      "step": 65
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999434726346564e-05,
      "loss": 1.6386,
      "step": 66
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999390408454773e-05,
      "loss": 1.6214,
      "step": 67
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999344418328162e-05,
      "loss": 1.7415,
      "step": 68
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999296755982107e-05,
      "loss": 1.5987,
      "step": 69
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999247421432554e-05,
      "loss": 1.721,
      "step": 70
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999196414696004e-05,
      "loss": 1.6221,
      "step": 71
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999143735789518e-05,
      "loss": 1.7561,
      "step": 72
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999089384730715e-05,
      "loss": 1.8831,
      "step": 73
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.999033361537779e-05,
      "loss": 1.8154,
      "step": 74
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.998975666229445e-05,
      "loss": 1.859,
      "step": 75
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.998916298825014e-05,
      "loss": 1.7888,
      "step": 76
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.998855259344342e-05,
      "loss": 1.7651,
      "step": 77
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.998792547807845e-05,
      "loss": 1.6255,
      "step": 78
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.998728164236503e-05,
      "loss": 1.8108,
      "step": 79
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.998662108651848e-05,
      "loss": 1.6713,
      "step": 80
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.998594381075977e-05,
      "loss": 1.7785,
      "step": 81
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.99852498153154e-05,
      "loss": 1.6179,
      "step": 82
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.998453910041756e-05,
      "loss": 1.9004,
      "step": 83
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.998381166630395e-05,
      "loss": 1.7622,
      "step": 84
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.998306751321787e-05,
      "loss": 1.8313,
      "step": 85
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.998230664140826e-05,
      "loss": 1.8524,
      "step": 86
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.998152905112961e-05,
      "loss": 1.7167,
      "step": 87
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.9980734742642e-05,
      "loss": 1.9441,
      "step": 88
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.997992371621113e-05,
      "loss": 1.9783,
      "step": 89
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.997909597210828e-05,
      "loss": 1.7863,
      "step": 90
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.997825151061034e-05,
      "loss": 1.5992,
      "step": 91
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.997739033199974e-05,
      "loss": 1.773,
      "step": 92
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.997651243656451e-05,
      "loss": 1.8229,
      "step": 93
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.997561782459837e-05,
      "loss": 1.6475,
      "step": 94
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.99747064964005e-05,
      "loss": 1.8028,
      "step": 95
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.997377845227576e-05,
      "loss": 1.855,
      "step": 96
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.997283369253453e-05,
      "loss": 1.8379,
      "step": 97
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.997187221749286e-05,
      "loss": 1.6612,
      "step": 98
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.997089402747233e-05,
      "loss": 1.6892,
      "step": 99
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.996989912280014e-05,
      "loss": 1.7313,
      "step": 100
    },
    {
      "epoch": 0.05,
      "eval_loss": 1.7044463157653809,
      "eval_runtime": 36.6189,
      "eval_samples_per_second": 3.85,
      "eval_steps_per_second": 1.939,
      "step": 100
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.996888750380907e-05,
      "loss": 1.7579,
      "step": 101
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.996785917083749e-05,
      "loss": 1.6712,
      "step": 102
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.996681412422937e-05,
      "loss": 1.7144,
      "step": 103
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.996575236433428e-05,
      "loss": 1.6995,
      "step": 104
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.996467389150734e-05,
      "loss": 1.7101,
      "step": 105
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.996357870610931e-05,
      "loss": 1.8103,
      "step": 106
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.99624668085065e-05,
      "loss": 1.7972,
      "step": 107
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.996133819907081e-05,
      "loss": 1.7793,
      "step": 108
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.996019287817979e-05,
      "loss": 1.7286,
      "step": 109
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.99590308462165e-05,
      "loss": 1.7587,
      "step": 110
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.995785210356965e-05,
      "loss": 1.7065,
      "step": 111
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.995665665063349e-05,
      "loss": 1.7103,
      "step": 112
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.995544448780789e-05,
      "loss": 1.5125,
      "step": 113
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.995421561549833e-05,
      "loss": 1.7484,
      "step": 114
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.995297003411581e-05,
      "loss": 1.7892,
      "step": 115
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.9951707744077e-05,
      "loss": 1.8235,
      "step": 116
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.995042874580411e-05,
      "loss": 1.7413,
      "step": 117
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.994913303972494e-05,
      "loss": 1.6547,
      "step": 118
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.994782062627288e-05,
      "loss": 1.6716,
      "step": 119
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.994649150588693e-05,
      "loss": 1.7557,
      "step": 120
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.994514567901167e-05,
      "loss": 1.7555,
      "step": 121
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.994378314609725e-05,
      "loss": 1.6922,
      "step": 122
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.994240390759944e-05,
      "loss": 1.6831,
      "step": 123
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.994100796397954e-05,
      "loss": 1.6498,
      "step": 124
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.993959531570452e-05,
      "loss": 1.7442,
      "step": 125
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.993816596324685e-05,
      "loss": 1.7683,
      "step": 126
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.993671990708468e-05,
      "loss": 1.6529,
      "step": 127
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.993525714770166e-05,
      "loss": 1.7629,
      "step": 128
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.993377768558709e-05,
      "loss": 1.6093,
      "step": 129
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.993228152123581e-05,
      "loss": 1.714,
      "step": 130
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.993076865514828e-05,
      "loss": 1.682,
      "step": 131
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.992923908783055e-05,
      "loss": 1.7414,
      "step": 132
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.992769281979419e-05,
      "loss": 1.7505,
      "step": 133
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.992612985155647e-05,
      "loss": 1.6918,
      "step": 134
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.992455018364016e-05,
      "loss": 1.6127,
      "step": 135
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.992295381657361e-05,
      "loss": 1.5986,
      "step": 136
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.992134075089084e-05,
      "loss": 1.7537,
      "step": 137
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.991971098713136e-05,
      "loss": 1.6213,
      "step": 138
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.991806452584032e-05,
      "loss": 1.6921,
      "step": 139
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.991640136756842e-05,
      "loss": 1.6715,
      "step": 140
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.9914721512872e-05,
      "loss": 1.5912,
      "step": 141
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.991302496231294e-05,
      "loss": 1.7052,
      "step": 142
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.991131171645872e-05,
      "loss": 1.6621,
      "step": 143
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.990958177588236e-05,
      "loss": 1.5554,
      "step": 144
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.990783514116257e-05,
      "loss": 1.596,
      "step": 145
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.990607181288352e-05,
      "loss": 1.4977,
      "step": 146
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.990429179163505e-05,
      "loss": 1.6575,
      "step": 147
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.990249507801256e-05,
      "loss": 1.7554,
      "step": 148
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.990068167261701e-05,
      "loss": 1.7382,
      "step": 149
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.989885157605496e-05,
      "loss": 1.7572,
      "step": 150
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.989700478893858e-05,
      "loss": 1.5932,
      "step": 151
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.989514131188559e-05,
      "loss": 1.7349,
      "step": 152
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.989326114551929e-05,
      "loss": 1.7046,
      "step": 153
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.989136429046856e-05,
      "loss": 1.7366,
      "step": 154
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.98894507473679e-05,
      "loss": 1.7825,
      "step": 155
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.988752051685736e-05,
      "loss": 1.8189,
      "step": 156
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.988557359958257e-05,
      "loss": 1.6414,
      "step": 157
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.988360999619476e-05,
      "loss": 1.6754,
      "step": 158
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.988162970735073e-05,
      "loss": 1.6224,
      "step": 159
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.987963273371286e-05,
      "loss": 1.4985,
      "step": 160
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.987761907594911e-05,
      "loss": 1.4666,
      "step": 161
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.987558873473302e-05,
      "loss": 1.8173,
      "step": 162
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.987354171074372e-05,
      "loss": 1.6869,
      "step": 163
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.987147800466593e-05,
      "loss": 1.6555,
      "step": 164
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.98693976171899e-05,
      "loss": 1.6812,
      "step": 165
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.986730054901153e-05,
      "loss": 1.5679,
      "step": 166
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.986518680083223e-05,
      "loss": 1.6499,
      "step": 167
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.986305637335907e-05,
      "loss": 1.6184,
      "step": 168
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.986090926730461e-05,
      "loss": 1.8161,
      "step": 169
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.985874548338702e-05,
      "loss": 1.6443,
      "step": 170
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.98565650223301e-05,
      "loss": 1.704,
      "step": 171
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.985436788486318e-05,
      "loss": 1.6735,
      "step": 172
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.985215407172115e-05,
      "loss": 1.7004,
      "step": 173
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.984992358364451e-05,
      "loss": 1.7332,
      "step": 174
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.984767642137935e-05,
      "loss": 1.6327,
      "step": 175
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.984541258567731e-05,
      "loss": 1.528,
      "step": 176
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.98431320772956e-05,
      "loss": 1.6471,
      "step": 177
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.984083489699706e-05,
      "loss": 1.6716,
      "step": 178
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.983852104555002e-05,
      "loss": 1.5888,
      "step": 179
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.983619052372848e-05,
      "loss": 1.7457,
      "step": 180
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.983384333231193e-05,
      "loss": 1.5845,
      "step": 181
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.983147947208552e-05,
      "loss": 1.5369,
      "step": 182
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.982909894383987e-05,
      "loss": 1.573,
      "step": 183
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.98267017483713e-05,
      "loss": 1.7446,
      "step": 184
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.982428788648162e-05,
      "loss": 1.6811,
      "step": 185
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.982185735897824e-05,
      "loss": 1.6759,
      "step": 186
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.981941016667413e-05,
      "loss": 1.7232,
      "step": 187
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.981694631038785e-05,
      "loss": 1.6485,
      "step": 188
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.981446579094354e-05,
      "loss": 1.569,
      "step": 189
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.98119686091709e-05,
      "loss": 1.6893,
      "step": 190
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.980945476590518e-05,
      "loss": 1.7607,
      "step": 191
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.980692426198728e-05,
      "loss": 1.6087,
      "step": 192
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.980437709826358e-05,
      "loss": 1.6765,
      "step": 193
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.980181327558609e-05,
      "loss": 1.7362,
      "step": 194
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.979923279481238e-05,
      "loss": 1.4928,
      "step": 195
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.97966356568056e-05,
      "loss": 1.7994,
      "step": 196
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.979402186243443e-05,
      "loss": 1.6853,
      "step": 197
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.979139141257319e-05,
      "loss": 1.5346,
      "step": 198
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.978874430810168e-05,
      "loss": 1.6063,
      "step": 199
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.978608054990539e-05,
      "loss": 1.68,
      "step": 200
    },
    {
      "epoch": 0.11,
      "eval_loss": 1.617618441581726,
      "eval_runtime": 36.7305,
      "eval_samples_per_second": 3.839,
      "eval_steps_per_second": 1.933,
      "step": 200
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.978340013887526e-05,
      "loss": 1.7748,
      "step": 201
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.97807030759079e-05,
      "loss": 1.6407,
      "step": 202
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.97779893619054e-05,
      "loss": 1.5273,
      "step": 203
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.977525899777548e-05,
      "loss": 1.6856,
      "step": 204
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.977251198443143e-05,
      "loss": 1.6027,
      "step": 205
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.976974832279209e-05,
      "loss": 1.8237,
      "step": 206
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.976696801378185e-05,
      "loss": 1.5264,
      "step": 207
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.97641710583307e-05,
      "loss": 1.6243,
      "step": 208
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.976135745737419e-05,
      "loss": 1.7358,
      "step": 209
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.975852721185344e-05,
      "loss": 1.646,
      "step": 210
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.975568032271511e-05,
      "loss": 1.6747,
      "step": 211
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.975281679091148e-05,
      "loss": 1.6404,
      "step": 212
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.974993661740035e-05,
      "loss": 1.7356,
      "step": 213
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.97470398031451e-05,
      "loss": 1.7024,
      "step": 214
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.974412634911467e-05,
      "loss": 1.5152,
      "step": 215
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.974119625628361e-05,
      "loss": 1.6149,
      "step": 216
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.973824952563198e-05,
      "loss": 1.5926,
      "step": 217
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.97352861581454e-05,
      "loss": 1.6799,
      "step": 218
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.973230615481511e-05,
      "loss": 1.5922,
      "step": 219
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.972930951663789e-05,
      "loss": 1.3735,
      "step": 220
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.972629624461605e-05,
      "loss": 1.543,
      "step": 221
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.972326633975752e-05,
      "loss": 1.6413,
      "step": 222
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.972021980307575e-05,
      "loss": 1.5425,
      "step": 223
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.971715663558979e-05,
      "loss": 1.7366,
      "step": 224
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.97140768383242e-05,
      "loss": 1.6097,
      "step": 225
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.971098041230915e-05,
      "loss": 1.5688,
      "step": 226
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.970786735858037e-05,
      "loss": 1.5358,
      "step": 227
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.970473767817911e-05,
      "loss": 1.5674,
      "step": 228
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.970159137215224e-05,
      "loss": 1.6846,
      "step": 229
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.969842844155214e-05,
      "loss": 1.5928,
      "step": 230
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.969524888743677e-05,
      "loss": 1.4993,
      "step": 231
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.969205271086968e-05,
      "loss": 1.5089,
      "step": 232
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.96888399129199e-05,
      "loss": 1.5034,
      "step": 233
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.968561049466214e-05,
      "loss": 1.706,
      "step": 234
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.968236445717654e-05,
      "loss": 1.5533,
      "step": 235
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.967910180154889e-05,
      "loss": 1.4252,
      "step": 236
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.96758225288705e-05,
      "loss": 1.6591,
      "step": 237
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.967252664023824e-05,
      "loss": 1.7582,
      "step": 238
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.966921413675458e-05,
      "loss": 1.4237,
      "step": 239
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.966588501952746e-05,
      "loss": 1.5899,
      "step": 240
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.966253928967047e-05,
      "loss": 1.6682,
      "step": 241
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.965917694830268e-05,
      "loss": 1.4552,
      "step": 242
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.965579799654878e-05,
      "loss": 1.346,
      "step": 243
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.965240243553898e-05,
      "loss": 1.7999,
      "step": 244
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.964899026640905e-05,
      "loss": 1.7168,
      "step": 245
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.964556149030033e-05,
      "loss": 1.4031,
      "step": 246
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.96421161083597e-05,
      "loss": 1.7865,
      "step": 247
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.963865412173957e-05,
      "loss": 1.5786,
      "step": 248
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.963517553159797e-05,
      "loss": 1.636,
      "step": 249
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.963168033909842e-05,
      "loss": 1.6806,
      "step": 250
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.962816854541004e-05,
      "loss": 1.6561,
      "step": 251
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.962464015170747e-05,
      "loss": 1.5236,
      "step": 252
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.962109515917094e-05,
      "loss": 1.7198,
      "step": 253
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.961753356898615e-05,
      "loss": 1.6403,
      "step": 254
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.961395538234445e-05,
      "loss": 1.4961,
      "step": 255
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.961036060044268e-05,
      "loss": 1.6338,
      "step": 256
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.960674922448327e-05,
      "loss": 1.5249,
      "step": 257
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.960312125567419e-05,
      "loss": 1.6002,
      "step": 258
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.959947669522891e-05,
      "loss": 1.5878,
      "step": 259
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.959581554436653e-05,
      "loss": 1.6431,
      "step": 260
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.959213780431164e-05,
      "loss": 1.6643,
      "step": 261
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.95884434762944e-05,
      "loss": 1.7333,
      "step": 262
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.958473256155052e-05,
      "loss": 1.5732,
      "step": 263
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.958100506132127e-05,
      "loss": 1.3707,
      "step": 264
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.957726097685342e-05,
      "loss": 1.7529,
      "step": 265
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.957350030939935e-05,
      "loss": 1.6427,
      "step": 266
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.956972306021695e-05,
      "loss": 1.4937,
      "step": 267
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.956592923056965e-05,
      "loss": 1.5907,
      "step": 268
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.956211882172646e-05,
      "loss": 1.5137,
      "step": 269
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.955829183496189e-05,
      "loss": 1.5936,
      "step": 270
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.955444827155604e-05,
      "loss": 1.7017,
      "step": 271
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.955058813279455e-05,
      "loss": 1.7307,
      "step": 272
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.954671141996855e-05,
      "loss": 1.4443,
      "step": 273
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.954281813437477e-05,
      "loss": 1.5454,
      "step": 274
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.953890827731548e-05,
      "loss": 1.5452,
      "step": 275
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.953498185009846e-05,
      "loss": 1.5515,
      "step": 276
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.953103885403707e-05,
      "loss": 1.8008,
      "step": 277
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.952707929045018e-05,
      "loss": 1.6693,
      "step": 278
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.95231031606622e-05,
      "loss": 1.6526,
      "step": 279
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.951911046600313e-05,
      "loss": 1.6965,
      "step": 280
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.951510120780848e-05,
      "loss": 1.5984,
      "step": 281
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.951107538741927e-05,
      "loss": 1.6076,
      "step": 282
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.950703300618206e-05,
      "loss": 1.6642,
      "step": 283
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.950297406544907e-05,
      "loss": 1.8548,
      "step": 284
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.949889856657788e-05,
      "loss": 1.4139,
      "step": 285
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.949480651093175e-05,
      "loss": 1.7774,
      "step": 286
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.949069789987936e-05,
      "loss": 1.6914,
      "step": 287
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.948657273479507e-05,
      "loss": 1.3921,
      "step": 288
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.948243101705863e-05,
      "loss": 1.5774,
      "step": 289
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.947827274805544e-05,
      "loss": 1.65,
      "step": 290
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.947409792917634e-05,
      "loss": 1.5218,
      "step": 291
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.946990656181781e-05,
      "loss": 1.6733,
      "step": 292
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.946569864738178e-05,
      "loss": 1.608,
      "step": 293
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.946147418727574e-05,
      "loss": 1.6465,
      "step": 294
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.945723318291274e-05,
      "loss": 1.5155,
      "step": 295
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.945297563571135e-05,
      "loss": 1.6531,
      "step": 296
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.944870154709562e-05,
      "loss": 1.7634,
      "step": 297
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.944441091849524e-05,
      "loss": 1.4945,
      "step": 298
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.944010375134533e-05,
      "loss": 1.5843,
      "step": 299
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.943578004708663e-05,
      "loss": 1.5642,
      "step": 300
    },
    {
      "epoch": 0.16,
      "eval_loss": 1.55384361743927,
      "eval_runtime": 36.8949,
      "eval_samples_per_second": 3.822,
      "eval_steps_per_second": 1.924,
      "step": 300
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.94314398071653e-05,
      "loss": 1.6959,
      "step": 301
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.942708303303315e-05,
      "loss": 1.6018,
      "step": 302
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.942270972614745e-05,
      "loss": 1.404,
      "step": 303
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.941831988797104e-05,
      "loss": 1.6023,
      "step": 304
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.941391351997222e-05,
      "loss": 1.575,
      "step": 305
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.940949062362491e-05,
      "loss": 1.6563,
      "step": 306
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.940505120040847e-05,
      "loss": 1.3636,
      "step": 307
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.940059525180788e-05,
      "loss": 1.6183,
      "step": 308
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.939612277931358e-05,
      "loss": 1.3996,
      "step": 309
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.939163378442156e-05,
      "loss": 1.5751,
      "step": 310
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.93871282686333e-05,
      "loss": 1.6,
      "step": 311
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.938260623345591e-05,
      "loss": 1.4927,
      "step": 312
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.93780676804019e-05,
      "loss": 1.7072,
      "step": 313
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.937351261098937e-05,
      "loss": 1.4406,
      "step": 314
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.936894102674196e-05,
      "loss": 1.5471,
      "step": 315
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.936435292918877e-05,
      "loss": 1.4668,
      "step": 316
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.93597483198645e-05,
      "loss": 1.686,
      "step": 317
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.93551272003093e-05,
      "loss": 1.5596,
      "step": 318
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.935048957206892e-05,
      "loss": 1.5392,
      "step": 319
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.934583543669453e-05,
      "loss": 1.7112,
      "step": 320
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.934116479574295e-05,
      "loss": 1.595,
      "step": 321
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.93364776507764e-05,
      "loss": 1.5186,
      "step": 322
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.933177400336269e-05,
      "loss": 1.5663,
      "step": 323
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.932705385507515e-05,
      "loss": 1.4623,
      "step": 324
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.932231720749258e-05,
      "loss": 1.6501,
      "step": 325
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.931756406219936e-05,
      "loss": 1.5327,
      "step": 326
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.931279442078534e-05,
      "loss": 1.4443,
      "step": 327
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.930800828484592e-05,
      "loss": 1.5233,
      "step": 328
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.930320565598198e-05,
      "loss": 1.4933,
      "step": 329
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.929838653579997e-05,
      "loss": 1.5529,
      "step": 330
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.92935509259118e-05,
      "loss": 1.7202,
      "step": 331
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.928869882793495e-05,
      "loss": 1.4906,
      "step": 332
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.928383024349236e-05,
      "loss": 1.3411,
      "step": 333
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.927894517421252e-05,
      "loss": 1.511,
      "step": 334
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.927404362172942e-05,
      "loss": 1.7573,
      "step": 335
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.926912558768262e-05,
      "loss": 1.3295,
      "step": 336
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.926419107371707e-05,
      "loss": 1.6602,
      "step": 337
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.925924008148335e-05,
      "loss": 1.749,
      "step": 338
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.925427261263747e-05,
      "loss": 1.377,
      "step": 339
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.924928866884103e-05,
      "loss": 1.5544,
      "step": 340
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.924428825176107e-05,
      "loss": 1.7618,
      "step": 341
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.923927136307019e-05,
      "loss": 1.6318,
      "step": 342
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.923423800444644e-05,
      "loss": 1.7002,
      "step": 343
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.922918817757345e-05,
      "loss": 1.7834,
      "step": 344
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.922412188414032e-05,
      "loss": 1.7808,
      "step": 345
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.921903912584165e-05,
      "loss": 1.5827,
      "step": 346
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.921393990437757e-05,
      "loss": 1.4452,
      "step": 347
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.920882422145372e-05,
      "loss": 1.61,
      "step": 348
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.920369207878121e-05,
      "loss": 1.5658,
      "step": 349
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.919854347807669e-05,
      "loss": 1.4943,
      "step": 350
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.919337842106229e-05,
      "loss": 1.7497,
      "step": 351
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.918819690946567e-05,
      "loss": 1.6391,
      "step": 352
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.918299894502e-05,
      "loss": 1.4223,
      "step": 353
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.917778452946391e-05,
      "loss": 1.6242,
      "step": 354
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.917255366454157e-05,
      "loss": 1.5401,
      "step": 355
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.916730635200264e-05,
      "loss": 1.4488,
      "step": 356
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.916204259360227e-05,
      "loss": 1.6609,
      "step": 357
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.915676239110112e-05,
      "loss": 1.4885,
      "step": 358
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.915146574626538e-05,
      "loss": 1.7408,
      "step": 359
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.914615266086668e-05,
      "loss": 1.6769,
      "step": 360
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.914082313668223e-05,
      "loss": 1.4584,
      "step": 361
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.913547717549463e-05,
      "loss": 1.4677,
      "step": 362
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.913011477909207e-05,
      "loss": 1.6108,
      "step": 363
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.912473594926821e-05,
      "loss": 1.7461,
      "step": 364
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.911934068782219e-05,
      "loss": 1.6484,
      "step": 365
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.911392899655867e-05,
      "loss": 1.4317,
      "step": 366
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.910850087728778e-05,
      "loss": 1.598,
      "step": 367
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.910305633182518e-05,
      "loss": 1.3288,
      "step": 368
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.9097595361992e-05,
      "loss": 1.4594,
      "step": 369
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.909211796961484e-05,
      "loss": 1.6349,
      "step": 370
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.908662415652585e-05,
      "loss": 1.4202,
      "step": 371
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.908111392456262e-05,
      "loss": 1.5376,
      "step": 372
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.90755872755683e-05,
      "loss": 1.4628,
      "step": 373
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.907004421139144e-05,
      "loss": 1.457,
      "step": 374
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.906448473388613e-05,
      "loss": 1.4207,
      "step": 375
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.905890884491195e-05,
      "loss": 1.5277,
      "step": 376
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.9053316546334e-05,
      "loss": 1.3968,
      "step": 377
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.904770784002279e-05,
      "loss": 1.55,
      "step": 378
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.90420827278544e-05,
      "loss": 1.5759,
      "step": 379
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.903644121171035e-05,
      "loss": 1.4236,
      "step": 380
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.903078329347765e-05,
      "loss": 1.4537,
      "step": 381
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.90251089750488e-05,
      "loss": 1.5536,
      "step": 382
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.90194182583218e-05,
      "loss": 1.5428,
      "step": 383
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.901371114520012e-05,
      "loss": 1.4612,
      "step": 384
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.900798763759273e-05,
      "loss": 1.5266,
      "step": 385
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.900224773741406e-05,
      "loss": 1.435,
      "step": 386
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.899649144658406e-05,
      "loss": 1.4723,
      "step": 387
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.899071876702809e-05,
      "loss": 1.6275,
      "step": 388
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.898492970067709e-05,
      "loss": 1.4796,
      "step": 389
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.89791242494674e-05,
      "loss": 1.5126,
      "step": 390
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.89733024153409e-05,
      "loss": 1.4878,
      "step": 391
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.89674642002449e-05,
      "loss": 1.4939,
      "step": 392
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.89616096061322e-05,
      "loss": 1.7362,
      "step": 393
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.895573863496114e-05,
      "loss": 1.4123,
      "step": 394
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.894985128869544e-05,
      "loss": 1.3768,
      "step": 395
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.894394756930436e-05,
      "loss": 1.3157,
      "step": 396
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.893802747876264e-05,
      "loss": 1.5932,
      "step": 397
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.893209101905044e-05,
      "loss": 1.6065,
      "step": 398
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.892613819215346e-05,
      "loss": 1.5144,
      "step": 399
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.892016900006284e-05,
      "loss": 1.6617,
      "step": 400
    },
    {
      "epoch": 0.22,
      "eval_loss": 1.4986377954483032,
      "eval_runtime": 36.767,
      "eval_samples_per_second": 3.835,
      "eval_steps_per_second": 1.931,
      "step": 400
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.891418344477522e-05,
      "loss": 1.5419,
      "step": 401
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.890818152829266e-05,
      "loss": 1.6694,
      "step": 402
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.890216325262276e-05,
      "loss": 1.59,
      "step": 403
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.889612861977853e-05,
      "loss": 1.5786,
      "step": 404
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.889007763177851e-05,
      "loss": 1.4079,
      "step": 405
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.888401029064667e-05,
      "loss": 1.4818,
      "step": 406
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.887792659841244e-05,
      "loss": 1.6226,
      "step": 407
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.887182655711077e-05,
      "loss": 1.4785,
      "step": 408
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.886571016878203e-05,
      "loss": 1.5573,
      "step": 409
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.88595774354721e-05,
      "loss": 1.4523,
      "step": 410
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.885342835923228e-05,
      "loss": 1.5657,
      "step": 411
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.884726294211937e-05,
      "loss": 1.4207,
      "step": 412
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.884108118619562e-05,
      "loss": 1.3178,
      "step": 413
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.883488309352877e-05,
      "loss": 1.3218,
      "step": 414
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.882866866619199e-05,
      "loss": 1.7805,
      "step": 415
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.882243790626393e-05,
      "loss": 1.4812,
      "step": 416
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.881619081582871e-05,
      "loss": 1.7566,
      "step": 417
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.88099273969759e-05,
      "loss": 1.4292,
      "step": 418
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.880364765180055e-05,
      "loss": 1.363,
      "step": 419
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.879735158240313e-05,
      "loss": 1.614,
      "step": 420
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.879103919088963e-05,
      "loss": 1.5891,
      "step": 421
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.878471047937143e-05,
      "loss": 1.5151,
      "step": 422
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.877836544996544e-05,
      "loss": 1.6327,
      "step": 423
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.877200410479399e-05,
      "loss": 1.641,
      "step": 424
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.876562644598486e-05,
      "loss": 1.5764,
      "step": 425
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.87592324756713e-05,
      "loss": 1.4256,
      "step": 426
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.875282219599203e-05,
      "loss": 1.4759,
      "step": 427
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.874639560909117e-05,
      "loss": 1.7108,
      "step": 428
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.87399527171184e-05,
      "loss": 1.468,
      "step": 429
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.873349352222873e-05,
      "loss": 1.3544,
      "step": 430
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.87270180265827e-05,
      "loss": 1.6302,
      "step": 431
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.872052623234632e-05,
      "loss": 1.264,
      "step": 432
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.871401814169094e-05,
      "loss": 1.2577,
      "step": 433
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.87074937567935e-05,
      "loss": 1.4197,
      "step": 434
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.870095307983631e-05,
      "loss": 1.4912,
      "step": 435
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.869439611300713e-05,
      "loss": 1.6923,
      "step": 436
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.86878228584992e-05,
      "loss": 1.5276,
      "step": 437
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.86812333185112e-05,
      "loss": 1.519,
      "step": 438
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.867462749524723e-05,
      "loss": 1.6289,
      "step": 439
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.866800539091688e-05,
      "loss": 1.5821,
      "step": 440
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.866136700773513e-05,
      "loss": 1.6478,
      "step": 441
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.865471234792248e-05,
      "loss": 1.3992,
      "step": 442
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.864804141370479e-05,
      "loss": 1.3251,
      "step": 443
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.864135420731344e-05,
      "loss": 1.4939,
      "step": 444
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.863465073098519e-05,
      "loss": 1.592,
      "step": 445
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.86279309869623e-05,
      "loss": 1.4832,
      "step": 446
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.862119497749242e-05,
      "loss": 1.6815,
      "step": 447
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.861444270482868e-05,
      "loss": 1.5266,
      "step": 448
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.860767417122961e-05,
      "loss": 1.4752,
      "step": 449
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.860088937895923e-05,
      "loss": 1.5003,
      "step": 450
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.859408833028694e-05,
      "loss": 1.3889,
      "step": 451
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.858727102748763e-05,
      "loss": 1.4368,
      "step": 452
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.858043747284158e-05,
      "loss": 1.4785,
      "step": 453
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.857358766863456e-05,
      "loss": 1.436,
      "step": 454
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.856672161715773e-05,
      "loss": 1.448,
      "step": 455
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.85598393207077e-05,
      "loss": 1.5604,
      "step": 456
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.855294078158652e-05,
      "loss": 1.6768,
      "step": 457
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.854602600210165e-05,
      "loss": 1.7228,
      "step": 458
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.853909498456602e-05,
      "loss": 1.5037,
      "step": 459
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.853214773129796e-05,
      "loss": 1.6755,
      "step": 460
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.852518424462125e-05,
      "loss": 1.5016,
      "step": 461
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.851820452686507e-05,
      "loss": 1.4891,
      "step": 462
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.851120858036407e-05,
      "loss": 1.7245,
      "step": 463
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.850419640745831e-05,
      "loss": 1.4918,
      "step": 464
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.849716801049324e-05,
      "loss": 1.5573,
      "step": 465
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.849012339181983e-05,
      "loss": 1.6531,
      "step": 466
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.848306255379438e-05,
      "loss": 1.6636,
      "step": 467
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.847598549877866e-05,
      "loss": 1.5337,
      "step": 468
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.846889222913986e-05,
      "loss": 1.4775,
      "step": 469
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.846178274725059e-05,
      "loss": 1.4483,
      "step": 470
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.84546570554889e-05,
      "loss": 1.379,
      "step": 471
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.844751515623824e-05,
      "loss": 1.5167,
      "step": 472
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.844035705188747e-05,
      "loss": 1.5387,
      "step": 473
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.843318274483089e-05,
      "loss": 1.7508,
      "step": 474
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.842599223746825e-05,
      "loss": 1.4962,
      "step": 475
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.841878553220465e-05,
      "loss": 1.6946,
      "step": 476
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.841156263145066e-05,
      "loss": 1.668,
      "step": 477
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.840432353762225e-05,
      "loss": 1.7026,
      "step": 478
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.839706825314081e-05,
      "loss": 1.5847,
      "step": 479
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.838979678043315e-05,
      "loss": 1.5309,
      "step": 480
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.838250912193147e-05,
      "loss": 1.6372,
      "step": 481
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.837520528007341e-05,
      "loss": 1.3924,
      "step": 482
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.836788525730202e-05,
      "loss": 1.3683,
      "step": 483
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.836054905606578e-05,
      "loss": 1.3838,
      "step": 484
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.835319667881852e-05,
      "loss": 1.2594,
      "step": 485
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.834582812801954e-05,
      "loss": 1.5784,
      "step": 486
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.833844340613351e-05,
      "loss": 1.5738,
      "step": 487
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.833104251563056e-05,
      "loss": 1.4493,
      "step": 488
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.832362545898618e-05,
      "loss": 1.5019,
      "step": 489
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.83161922386813e-05,
      "loss": 1.2788,
      "step": 490
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.830874285720222e-05,
      "loss": 1.4136,
      "step": 491
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.830127731704066e-05,
      "loss": 1.7904,
      "step": 492
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.829379562069379e-05,
      "loss": 1.4728,
      "step": 493
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.828629777066411e-05,
      "loss": 1.4144,
      "step": 494
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.827878376945958e-05,
      "loss": 1.4691,
      "step": 495
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.827125361959353e-05,
      "loss": 1.5393,
      "step": 496
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.82637073235847e-05,
      "loss": 1.6571,
      "step": 497
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.825614488395724e-05,
      "loss": 1.8122,
      "step": 498
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.824856630324068e-05,
      "loss": 1.6513,
      "step": 499
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.824097158397001e-05,
      "loss": 1.4733,
      "step": 500
    },
    {
      "epoch": 0.27,
      "eval_loss": 1.4722751379013062,
      "eval_runtime": 36.7781,
      "eval_samples_per_second": 3.834,
      "eval_steps_per_second": 1.93,
      "step": 500
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.823336072868551e-05,
      "loss": 1.5385,
      "step": 501
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.822573373993295e-05,
      "loss": 1.3659,
      "step": 502
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.821809062026346e-05,
      "loss": 1.624,
      "step": 503
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.821043137223356e-05,
      "loss": 1.713,
      "step": 504
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.820275599840519e-05,
      "loss": 1.5789,
      "step": 505
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.819506450134566e-05,
      "loss": 1.4229,
      "step": 506
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.818735688362767e-05,
      "loss": 1.6025,
      "step": 507
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.817963314782934e-05,
      "loss": 1.5751,
      "step": 508
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.817189329653416e-05,
      "loss": 1.3054,
      "step": 509
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.816413733233103e-05,
      "loss": 1.4614,
      "step": 510
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.81563652578142e-05,
      "loss": 1.581,
      "step": 511
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.814857707558335e-05,
      "loss": 1.5658,
      "step": 512
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.814077278824352e-05,
      "loss": 1.5364,
      "step": 513
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.813295239840516e-05,
      "loss": 1.4701,
      "step": 514
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.81251159086841e-05,
      "loss": 1.5498,
      "step": 515
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.811726332170153e-05,
      "loss": 1.5782,
      "step": 516
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.810939464008404e-05,
      "loss": 1.4876,
      "step": 517
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.810150986646363e-05,
      "loss": 1.3724,
      "step": 518
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.809360900347765e-05,
      "loss": 1.5996,
      "step": 519
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.808569205376884e-05,
      "loss": 1.4485,
      "step": 520
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.807775901998532e-05,
      "loss": 1.6323,
      "step": 521
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.806980990478062e-05,
      "loss": 1.7685,
      "step": 522
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.80618447108136e-05,
      "loss": 1.3104,
      "step": 523
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.80538634407485e-05,
      "loss": 1.7373,
      "step": 524
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.804586609725499e-05,
      "loss": 1.6309,
      "step": 525
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.803785268300806e-05,
      "loss": 1.5738,
      "step": 526
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.802982320068811e-05,
      "loss": 1.4459,
      "step": 527
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.802177765298091e-05,
      "loss": 1.3339,
      "step": 528
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.801371604257758e-05,
      "loss": 1.4136,
      "step": 529
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.800563837217464e-05,
      "loss": 1.3621,
      "step": 530
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.799754464447397e-05,
      "loss": 1.5002,
      "step": 531
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.798943486218284e-05,
      "loss": 1.634,
      "step": 532
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.798130902801384e-05,
      "loss": 1.5725,
      "step": 533
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.797316714468498e-05,
      "loss": 1.3882,
      "step": 534
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.796500921491963e-05,
      "loss": 1.3056,
      "step": 535
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.795683524144649e-05,
      "loss": 1.4078,
      "step": 536
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.794864522699967e-05,
      "loss": 1.2367,
      "step": 537
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.794043917431863e-05,
      "loss": 1.6724,
      "step": 538
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.793221708614821e-05,
      "loss": 1.4024,
      "step": 539
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.792397896523858e-05,
      "loss": 1.4046,
      "step": 540
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.791572481434528e-05,
      "loss": 1.2647,
      "step": 541
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.790745463622924e-05,
      "loss": 1.6655,
      "step": 542
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.789916843365672e-05,
      "loss": 1.4923,
      "step": 543
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.789086620939936e-05,
      "loss": 1.4124,
      "step": 544
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.788254796623415e-05,
      "loss": 1.2791,
      "step": 545
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.787421370694345e-05,
      "loss": 1.4361,
      "step": 546
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.786586343431496e-05,
      "loss": 1.565,
      "step": 547
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.785749715114177e-05,
      "loss": 1.6277,
      "step": 548
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.784911486022224e-05,
      "loss": 1.2493,
      "step": 549
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.784071656436017e-05,
      "loss": 1.4295,
      "step": 550
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.783230226636472e-05,
      "loss": 1.1658,
      "step": 551
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.782387196905034e-05,
      "loss": 1.6212,
      "step": 552
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.781542567523686e-05,
      "loss": 1.4246,
      "step": 553
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.780696338774949e-05,
      "loss": 1.2565,
      "step": 554
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.77984851094187e-05,
      "loss": 1.72,
      "step": 555
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.778999084308043e-05,
      "loss": 1.5483,
      "step": 556
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.778148059157589e-05,
      "loss": 1.2621,
      "step": 557
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.777295435775165e-05,
      "loss": 1.4483,
      "step": 558
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.776441214445961e-05,
      "loss": 1.3818,
      "step": 559
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.775585395455708e-05,
      "loss": 1.3014,
      "step": 560
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.774727979090663e-05,
      "loss": 1.6783,
      "step": 561
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.773868965637624e-05,
      "loss": 1.4269,
      "step": 562
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.773008355383921e-05,
      "loss": 1.3423,
      "step": 563
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.772146148617414e-05,
      "loss": 1.4432,
      "step": 564
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.771282345626505e-05,
      "loss": 1.4501,
      "step": 565
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.770416946700121e-05,
      "loss": 1.5259,
      "step": 566
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.769549952127731e-05,
      "loss": 1.4745,
      "step": 567
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.76868136219933e-05,
      "loss": 1.4865,
      "step": 568
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.767811177205455e-05,
      "loss": 1.4951,
      "step": 569
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.766939397437171e-05,
      "loss": 1.4829,
      "step": 570
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.766066023186076e-05,
      "loss": 1.4125,
      "step": 571
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.765191054744305e-05,
      "loss": 1.4172,
      "step": 572
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.764314492404523e-05,
      "loss": 1.1917,
      "step": 573
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.763436336459931e-05,
      "loss": 1.4866,
      "step": 574
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.762556587204257e-05,
      "loss": 1.2856,
      "step": 575
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.761675244931772e-05,
      "loss": 1.4103,
      "step": 576
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.76079230993727e-05,
      "loss": 1.386,
      "step": 577
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.759907782516085e-05,
      "loss": 1.5233,
      "step": 578
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.759021662964079e-05,
      "loss": 1.5288,
      "step": 579
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.758133951577648e-05,
      "loss": 1.5925,
      "step": 580
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.75724464865372e-05,
      "loss": 1.4235,
      "step": 581
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.756353754489758e-05,
      "loss": 1.5297,
      "step": 582
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.755461269383753e-05,
      "loss": 1.4286,
      "step": 583
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.754567193634232e-05,
      "loss": 1.5984,
      "step": 584
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.753671527540252e-05,
      "loss": 1.1895,
      "step": 585
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.752774271401402e-05,
      "loss": 1.4767,
      "step": 586
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.751875425517802e-05,
      "loss": 1.4235,
      "step": 587
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.750974990190106e-05,
      "loss": 1.3751,
      "step": 588
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.750072965719501e-05,
      "loss": 1.4335,
      "step": 589
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.749169352407699e-05,
      "loss": 1.5916,
      "step": 590
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.748264150556949e-05,
      "loss": 1.4927,
      "step": 591
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.747357360470033e-05,
      "loss": 1.5581,
      "step": 592
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.746448982450256e-05,
      "loss": 1.6105,
      "step": 593
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.745539016801462e-05,
      "loss": 1.599,
      "step": 594
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.744627463828025e-05,
      "loss": 1.6301,
      "step": 595
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.743714323834844e-05,
      "loss": 1.6018,
      "step": 596
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.742799597127357e-05,
      "loss": 1.499,
      "step": 597
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.741883284011527e-05,
      "loss": 1.2309,
      "step": 598
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.74096538479385e-05,
      "loss": 1.3185,
      "step": 599
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.740045899781352e-05,
      "loss": 1.4916,
      "step": 600
    },
    {
      "epoch": 0.33,
      "eval_loss": 1.4427075386047363,
      "eval_runtime": 36.8478,
      "eval_samples_per_second": 3.827,
      "eval_steps_per_second": 1.927,
      "step": 600
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.73912482928159e-05,
      "loss": 1.8063,
      "step": 601
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.738202173602651e-05,
      "loss": 1.5728,
      "step": 602
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.737277933053151e-05,
      "loss": 1.3874,
      "step": 603
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.736352107942236e-05,
      "loss": 1.6993,
      "step": 604
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.735424698579586e-05,
      "loss": 1.5198,
      "step": 605
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.734495705275406e-05,
      "loss": 1.2906,
      "step": 606
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.733565128340434e-05,
      "loss": 1.3701,
      "step": 607
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.732632968085936e-05,
      "loss": 1.3148,
      "step": 608
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.73169922482371e-05,
      "loss": 1.5968,
      "step": 609
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.730763898866078e-05,
      "loss": 1.5244,
      "step": 610
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.729826990525897e-05,
      "loss": 1.3418,
      "step": 611
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.728888500116551e-05,
      "loss": 1.4939,
      "step": 612
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.727948427951956e-05,
      "loss": 1.4774,
      "step": 613
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.727006774346551e-05,
      "loss": 1.6881,
      "step": 614
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.726063539615309e-05,
      "loss": 1.5334,
      "step": 615
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.725118724073731e-05,
      "loss": 1.4223,
      "step": 616
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.724172328037847e-05,
      "loss": 1.5708,
      "step": 617
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.723224351824213e-05,
      "loss": 1.6845,
      "step": 618
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.722274795749916e-05,
      "loss": 1.3943,
      "step": 619
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.721323660132571e-05,
      "loss": 1.3402,
      "step": 620
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.720370945290321e-05,
      "loss": 1.4511,
      "step": 621
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.719416651541839e-05,
      "loss": 1.3799,
      "step": 622
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.718460779206322e-05,
      "loss": 1.4327,
      "step": 623
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.717503328603498e-05,
      "loss": 1.4914,
      "step": 624
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.716544300053625e-05,
      "loss": 1.4686,
      "step": 625
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.715583693877484e-05,
      "loss": 1.1474,
      "step": 626
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.714621510396386e-05,
      "loss": 1.43,
      "step": 627
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.713657749932172e-05,
      "loss": 1.6726,
      "step": 628
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.712692412807204e-05,
      "loss": 1.5083,
      "step": 629
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.71172549934438e-05,
      "loss": 1.4269,
      "step": 630
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.710757009867117e-05,
      "loss": 1.5713,
      "step": 631
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.709786944699364e-05,
      "loss": 1.1851,
      "step": 632
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.708815304165597e-05,
      "loss": 1.6094,
      "step": 633
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.707842088590816e-05,
      "loss": 1.4428,
      "step": 634
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.706867298300552e-05,
      "loss": 1.4716,
      "step": 635
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.705890933620858e-05,
      "loss": 1.3122,
      "step": 636
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.704912994878316e-05,
      "loss": 1.6551,
      "step": 637
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.703933482400038e-05,
      "loss": 1.43,
      "step": 638
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.702952396513656e-05,
      "loss": 1.5217,
      "step": 639
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.701969737547331e-05,
      "loss": 1.6894,
      "step": 640
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.700985505829753e-05,
      "loss": 1.5085,
      "step": 641
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.699999701690133e-05,
      "loss": 1.4107,
      "step": 642
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.69901232545821e-05,
      "loss": 1.5065,
      "step": 643
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.69802337746425e-05,
      "loss": 1.2372,
      "step": 644
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.697032858039045e-05,
      "loss": 1.3157,
      "step": 645
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.696040767513911e-05,
      "loss": 1.5557,
      "step": 646
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.69504710622069e-05,
      "loss": 1.489,
      "step": 647
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.694051874491748e-05,
      "loss": 1.2018,
      "step": 648
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.693055072659982e-05,
      "loss": 1.553,
      "step": 649
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.692056701058805e-05,
      "loss": 1.461,
      "step": 650
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.691056760022163e-05,
      "loss": 1.2479,
      "step": 651
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.690055249884524e-05,
      "loss": 1.6044,
      "step": 652
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.68905217098088e-05,
      "loss": 1.5106,
      "step": 653
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.688047523646748e-05,
      "loss": 1.3433,
      "step": 654
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.687041308218173e-05,
      "loss": 1.4593,
      "step": 655
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.686033525031719e-05,
      "loss": 1.3379,
      "step": 656
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.685024174424478e-05,
      "loss": 1.4891,
      "step": 657
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.684013256734065e-05,
      "loss": 1.4017,
      "step": 658
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.68300077229862e-05,
      "loss": 1.4275,
      "step": 659
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.681986721456806e-05,
      "loss": 1.4821,
      "step": 660
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.680971104547811e-05,
      "loss": 1.5315,
      "step": 661
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.679953921911349e-05,
      "loss": 1.5456,
      "step": 662
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.67893517388765e-05,
      "loss": 1.3524,
      "step": 663
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.677914860817476e-05,
      "loss": 1.685,
      "step": 664
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.67689298304211e-05,
      "loss": 1.5883,
      "step": 665
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.675869540903355e-05,
      "loss": 1.3907,
      "step": 666
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.674844534743542e-05,
      "loss": 1.3913,
      "step": 667
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.67381796490552e-05,
      "loss": 1.4642,
      "step": 668
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.672789831732668e-05,
      "loss": 1.3643,
      "step": 669
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.671760135568881e-05,
      "loss": 1.5848,
      "step": 670
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.670728876758582e-05,
      "loss": 1.367,
      "step": 671
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.669696055646713e-05,
      "loss": 1.63,
      "step": 672
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.66866167257874e-05,
      "loss": 1.4155,
      "step": 673
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.667625727900652e-05,
      "loss": 1.5518,
      "step": 674
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.666588221958959e-05,
      "loss": 1.4966,
      "step": 675
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.665549155100694e-05,
      "loss": 1.1819,
      "step": 676
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.664508527673415e-05,
      "loss": 1.1972,
      "step": 677
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.663466340025194e-05,
      "loss": 1.4923,
      "step": 678
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.662422592504635e-05,
      "loss": 1.2743,
      "step": 679
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.661377285460855e-05,
      "loss": 1.8538,
      "step": 680
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.660330419243498e-05,
      "loss": 1.7904,
      "step": 681
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.659281994202729e-05,
      "loss": 1.4875,
      "step": 682
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.65823201068923e-05,
      "loss": 1.4073,
      "step": 683
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.657180469054213e-05,
      "loss": 1.5877,
      "step": 684
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.656127369649399e-05,
      "loss": 1.4882,
      "step": 685
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.655072712827043e-05,
      "loss": 1.361,
      "step": 686
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.65401649893991e-05,
      "loss": 1.4815,
      "step": 687
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.652958728341296e-05,
      "loss": 1.4289,
      "step": 688
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.651899401385007e-05,
      "loss": 1.4219,
      "step": 689
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.650838518425377e-05,
      "loss": 1.7062,
      "step": 690
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.649776079817258e-05,
      "loss": 1.3703,
      "step": 691
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.648712085916026e-05,
      "loss": 1.314,
      "step": 692
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.647646537077568e-05,
      "loss": 1.6359,
      "step": 693
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.6465794336583e-05,
      "loss": 1.3489,
      "step": 694
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.645510776015156e-05,
      "loss": 1.4972,
      "step": 695
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.644440564505588e-05,
      "loss": 1.3124,
      "step": 696
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.643368799487569e-05,
      "loss": 1.3759,
      "step": 697
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.642295481319588e-05,
      "loss": 1.5464,
      "step": 698
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.641220610360662e-05,
      "loss": 1.6827,
      "step": 699
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.64014418697032e-05,
      "loss": 1.5036,
      "step": 700
    },
    {
      "epoch": 0.38,
      "eval_loss": 1.4270912408828735,
      "eval_runtime": 36.6963,
      "eval_samples_per_second": 3.842,
      "eval_steps_per_second": 1.935,
      "step": 700
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.639066211508611e-05,
      "loss": 1.4752,
      "step": 701
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.637986684336107e-05,
      "loss": 1.3483,
      "step": 702
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.636905605813894e-05,
      "loss": 1.5509,
      "step": 703
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.635822976303581e-05,
      "loss": 1.2147,
      "step": 704
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.634738796167295e-05,
      "loss": 1.1906,
      "step": 705
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.63365306576768e-05,
      "loss": 1.5194,
      "step": 706
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.6325657854679e-05,
      "loss": 1.5629,
      "step": 707
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.631476955631635e-05,
      "loss": 1.3198,
      "step": 708
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.630386576623087e-05,
      "loss": 1.4998,
      "step": 709
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.629294648806976e-05,
      "loss": 1.479,
      "step": 710
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.628201172548535e-05,
      "loss": 1.4496,
      "step": 711
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.627106148213522e-05,
      "loss": 1.435,
      "step": 712
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.626009576168206e-05,
      "loss": 1.4386,
      "step": 713
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.624911456779378e-05,
      "loss": 1.3032,
      "step": 714
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.623811790414346e-05,
      "loss": 1.4093,
      "step": 715
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.622710577440936e-05,
      "loss": 1.3006,
      "step": 716
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.621607818227486e-05,
      "loss": 1.3545,
      "step": 717
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.620503513142861e-05,
      "loss": 1.5209,
      "step": 718
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.619397662556435e-05,
      "loss": 1.6819,
      "step": 719
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.6182902668381e-05,
      "loss": 1.3767,
      "step": 720
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.617181326358268e-05,
      "loss": 1.4189,
      "step": 721
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.616070841487865e-05,
      "loss": 1.6011,
      "step": 722
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.614958812598336e-05,
      "loss": 1.3735,
      "step": 723
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.61384524006164e-05,
      "loss": 1.5289,
      "step": 724
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.612730124250254e-05,
      "loss": 1.5457,
      "step": 725
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.61161346553717e-05,
      "loss": 1.4667,
      "step": 726
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.610495264295896e-05,
      "loss": 1.4116,
      "step": 727
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.609375520900459e-05,
      "loss": 1.4389,
      "step": 728
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.608254235725396e-05,
      "loss": 1.4747,
      "step": 729
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.607131409145766e-05,
      "loss": 1.4785,
      "step": 730
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.606007041537141e-05,
      "loss": 1.5348,
      "step": 731
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.604881133275605e-05,
      "loss": 1.4784,
      "step": 732
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.603753684737764e-05,
      "loss": 1.5852,
      "step": 733
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.602624696300733e-05,
      "loss": 1.5752,
      "step": 734
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.601494168342147e-05,
      "loss": 1.5627,
      "step": 735
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.600362101240152e-05,
      "loss": 1.4923,
      "step": 736
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.599228495373411e-05,
      "loss": 1.6381,
      "step": 737
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.598093351121103e-05,
      "loss": 1.4893,
      "step": 738
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.596956668862917e-05,
      "loss": 1.5766,
      "step": 739
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.595818448979061e-05,
      "loss": 1.5097,
      "step": 740
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.594678691850256e-05,
      "loss": 1.4205,
      "step": 741
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.593537397857736e-05,
      "loss": 1.5191,
      "step": 742
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.592394567383249e-05,
      "loss": 1.2992,
      "step": 743
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.59125020080906e-05,
      "loss": 1.4063,
      "step": 744
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.590104298517944e-05,
      "loss": 1.4152,
      "step": 745
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.588956860893191e-05,
      "loss": 1.5419,
      "step": 746
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.587807888318605e-05,
      "loss": 1.4363,
      "step": 747
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.586657381178505e-05,
      "loss": 1.529,
      "step": 748
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.58550533985772e-05,
      "loss": 1.3535,
      "step": 749
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.584351764741594e-05,
      "loss": 1.3485,
      "step": 750
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.583196656215983e-05,
      "loss": 1.3991,
      "step": 751
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.582040014667258e-05,
      "loss": 1.5006,
      "step": 752
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.5808818404823e-05,
      "loss": 1.5599,
      "step": 753
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.579722134048506e-05,
      "loss": 1.5587,
      "step": 754
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.57856089575378e-05,
      "loss": 1.7304,
      "step": 755
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.577398125986547e-05,
      "loss": 1.5014,
      "step": 756
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.576233825135735e-05,
      "loss": 1.2779,
      "step": 757
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.575067993590791e-05,
      "loss": 1.419,
      "step": 758
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.57390063174167e-05,
      "loss": 1.4133,
      "step": 759
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.572731739978839e-05,
      "loss": 1.6145,
      "step": 760
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.571561318693283e-05,
      "loss": 1.5294,
      "step": 761
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.570389368276486e-05,
      "loss": 1.2207,
      "step": 762
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.569215889120459e-05,
      "loss": 1.3521,
      "step": 763
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.56804088161771e-05,
      "loss": 1.3679,
      "step": 764
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.566864346161268e-05,
      "loss": 1.3888,
      "step": 765
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.565686283144669e-05,
      "loss": 1.2173,
      "step": 766
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.564506692961961e-05,
      "loss": 1.5328,
      "step": 767
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.563325576007701e-05,
      "loss": 1.4986,
      "step": 768
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.56214293267696e-05,
      "loss": 1.4195,
      "step": 769
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.560958763365316e-05,
      "loss": 1.3024,
      "step": 770
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.55977306846886e-05,
      "loss": 1.5882,
      "step": 771
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.558585848384194e-05,
      "loss": 1.3811,
      "step": 772
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.557397103508425e-05,
      "loss": 1.5454,
      "step": 773
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.556206834239177e-05,
      "loss": 1.2483,
      "step": 774
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.555015040974578e-05,
      "loss": 1.4561,
      "step": 775
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.553821724113268e-05,
      "loss": 1.3138,
      "step": 776
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.5526268840544e-05,
      "loss": 1.22,
      "step": 777
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.55143052119763e-05,
      "loss": 1.4946,
      "step": 778
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.550232635943131e-05,
      "loss": 1.2528,
      "step": 779
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.549033228691575e-05,
      "loss": 1.5111,
      "step": 780
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.547832299844153e-05,
      "loss": 1.3643,
      "step": 781
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.546629849802562e-05,
      "loss": 1.404,
      "step": 782
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.545425878969003e-05,
      "loss": 1.429,
      "step": 783
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.544220387746192e-05,
      "loss": 1.5545,
      "step": 784
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.543013376537349e-05,
      "loss": 1.3721,
      "step": 785
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.541804845746208e-05,
      "loss": 1.6218,
      "step": 786
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.540594795777005e-05,
      "loss": 1.3407,
      "step": 787
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.539383227034487e-05,
      "loss": 1.1125,
      "step": 788
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.53817013992391e-05,
      "loss": 1.4596,
      "step": 789
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.536955534851037e-05,
      "loss": 1.3413,
      "step": 790
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.535739412222137e-05,
      "loss": 1.4717,
      "step": 791
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.534521772443988e-05,
      "loss": 1.2931,
      "step": 792
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.533302615923879e-05,
      "loss": 1.2282,
      "step": 793
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.532081943069597e-05,
      "loss": 1.3767,
      "step": 794
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.530859754289447e-05,
      "loss": 1.3381,
      "step": 795
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.529636049992234e-05,
      "loss": 1.3597,
      "step": 796
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.528410830587274e-05,
      "loss": 1.6438,
      "step": 797
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.527184096484385e-05,
      "loss": 1.4676,
      "step": 798
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.525955848093898e-05,
      "loss": 1.6144,
      "step": 799
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.524726085826644e-05,
      "loss": 1.2385,
      "step": 800
    },
    {
      "epoch": 0.44,
      "eval_loss": 1.4108761548995972,
      "eval_runtime": 36.7593,
      "eval_samples_per_second": 3.836,
      "eval_steps_per_second": 1.931,
      "step": 800
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.523494810093965e-05,
      "loss": 1.409,
      "step": 801
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.522262021307707e-05,
      "loss": 1.1556,
      "step": 802
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.521027719880222e-05,
      "loss": 1.2249,
      "step": 803
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.519791906224371e-05,
      "loss": 1.556,
      "step": 804
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.518554580753516e-05,
      "loss": 1.3799,
      "step": 805
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.51731574388153e-05,
      "loss": 1.6302,
      "step": 806
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.516075396022786e-05,
      "loss": 1.6061,
      "step": 807
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.514833537592166e-05,
      "loss": 1.4205,
      "step": 808
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.513590169005058e-05,
      "loss": 1.3647,
      "step": 809
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.51234529067735e-05,
      "loss": 1.166,
      "step": 810
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.511098903025443e-05,
      "loss": 1.7086,
      "step": 811
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.509851006466234e-05,
      "loss": 1.5236,
      "step": 812
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.508601601417133e-05,
      "loss": 1.6145,
      "step": 813
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.50735068829605e-05,
      "loss": 1.3944,
      "step": 814
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.506098267521397e-05,
      "loss": 1.3782,
      "step": 815
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.504844339512095e-05,
      "loss": 1.4339,
      "step": 816
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.50358890468757e-05,
      "loss": 1.3564,
      "step": 817
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.502331963467748e-05,
      "loss": 1.1868,
      "step": 818
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.501073516273058e-05,
      "loss": 1.4752,
      "step": 819
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.499813563524438e-05,
      "loss": 1.5265,
      "step": 820
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.498552105643327e-05,
      "loss": 1.4095,
      "step": 821
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.497289143051664e-05,
      "loss": 1.6006,
      "step": 822
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.4960246761719e-05,
      "loss": 1.1724,
      "step": 823
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.494758705426978e-05,
      "loss": 1.6322,
      "step": 824
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.493491231240353e-05,
      "loss": 1.5915,
      "step": 825
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.492222254035977e-05,
      "loss": 1.4657,
      "step": 826
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.490951774238309e-05,
      "loss": 1.5854,
      "step": 827
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.48967979227231e-05,
      "loss": 1.493,
      "step": 828
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.488406308563441e-05,
      "loss": 1.4395,
      "step": 829
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.487131323537668e-05,
      "loss": 1.3752,
      "step": 830
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.485854837621455e-05,
      "loss": 1.1705,
      "step": 831
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.484576851241773e-05,
      "loss": 1.4337,
      "step": 832
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.483297364826094e-05,
      "loss": 1.4482,
      "step": 833
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.482016378802388e-05,
      "loss": 1.5758,
      "step": 834
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.48073389359913e-05,
      "loss": 1.4157,
      "step": 835
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.479449909645296e-05,
      "loss": 1.5657,
      "step": 836
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.478164427370363e-05,
      "loss": 1.2117,
      "step": 837
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.476877447204308e-05,
      "loss": 1.3389,
      "step": 838
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.475588969577611e-05,
      "loss": 1.3639,
      "step": 839
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.474298994921251e-05,
      "loss": 1.4776,
      "step": 840
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.473007523666711e-05,
      "loss": 1.3218,
      "step": 841
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.471714556245969e-05,
      "loss": 1.4894,
      "step": 842
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.470420093091512e-05,
      "loss": 1.3036,
      "step": 843
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.469124134636316e-05,
      "loss": 1.3324,
      "step": 844
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.467826681313864e-05,
      "loss": 1.2482,
      "step": 845
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.466527733558144e-05,
      "loss": 1.2368,
      "step": 846
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.465227291803632e-05,
      "loss": 1.3426,
      "step": 847
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.463925356485313e-05,
      "loss": 1.4441,
      "step": 848
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.462621928038667e-05,
      "loss": 1.4244,
      "step": 849
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.461317006899675e-05,
      "loss": 1.3798,
      "step": 850
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.460010593504816e-05,
      "loss": 1.7388,
      "step": 851
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.458702688291073e-05,
      "loss": 1.344,
      "step": 852
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.457393291695921e-05,
      "loss": 1.3237,
      "step": 853
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.456082404157338e-05,
      "loss": 1.6278,
      "step": 854
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.454770026113799e-05,
      "loss": 1.3602,
      "step": 855
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.45345615800428e-05,
      "loss": 1.2669,
      "step": 856
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.452140800268253e-05,
      "loss": 1.344,
      "step": 857
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.45082395334569e-05,
      "loss": 1.2855,
      "step": 858
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.449505617677057e-05,
      "loss": 1.5536,
      "step": 859
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.448185793703326e-05,
      "loss": 1.5212,
      "step": 860
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.446864481865959e-05,
      "loss": 1.2306,
      "step": 861
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.445541682606916e-05,
      "loss": 1.4159,
      "step": 862
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.444217396368662e-05,
      "loss": 1.3117,
      "step": 863
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.442891623594153e-05,
      "loss": 1.5532,
      "step": 864
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.441564364726843e-05,
      "loss": 1.6038,
      "step": 865
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.440235620210683e-05,
      "loss": 1.516,
      "step": 866
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.438905390490123e-05,
      "loss": 1.5231,
      "step": 867
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.437573676010109e-05,
      "loss": 1.1848,
      "step": 868
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.436240477216083e-05,
      "loss": 1.4511,
      "step": 869
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.434905794553983e-05,
      "loss": 1.2814,
      "step": 870
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.433569628470245e-05,
      "loss": 1.4011,
      "step": 871
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.432231979411798e-05,
      "loss": 1.2911,
      "step": 872
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.430892847826071e-05,
      "loss": 1.5617,
      "step": 873
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.429552234160987e-05,
      "loss": 1.1851,
      "step": 874
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.428210138864965e-05,
      "loss": 1.5199,
      "step": 875
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.426866562386917e-05,
      "loss": 1.4251,
      "step": 876
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.425521505176258e-05,
      "loss": 1.3895,
      "step": 877
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.424174967682888e-05,
      "loss": 1.4713,
      "step": 878
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.422826950357211e-05,
      "loss": 1.5795,
      "step": 879
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.421477453650118e-05,
      "loss": 1.4363,
      "step": 880
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.420126478013003e-05,
      "loss": 1.4376,
      "step": 881
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.41877402389775e-05,
      "loss": 1.4122,
      "step": 882
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.417420091756738e-05,
      "loss": 1.4208,
      "step": 883
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.41606468204284e-05,
      "loss": 1.2931,
      "step": 884
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.414707795209425e-05,
      "loss": 1.189,
      "step": 885
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.413349431710353e-05,
      "loss": 1.3412,
      "step": 886
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.411989591999982e-05,
      "loss": 1.4887,
      "step": 887
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.410628276533163e-05,
      "loss": 1.4574,
      "step": 888
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.409265485765235e-05,
      "loss": 1.3857,
      "step": 889
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.407901220152038e-05,
      "loss": 1.4062,
      "step": 890
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.406535480149903e-05,
      "loss": 1.3591,
      "step": 891
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.40516826621565e-05,
      "loss": 1.37,
      "step": 892
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.403799578806599e-05,
      "loss": 1.4833,
      "step": 893
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.402429418380554e-05,
      "loss": 1.1785,
      "step": 894
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.401057785395822e-05,
      "loss": 1.559,
      "step": 895
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.399684680311196e-05,
      "loss": 1.5577,
      "step": 896
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.398310103585961e-05,
      "loss": 1.5023,
      "step": 897
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.3969340556799e-05,
      "loss": 1.5057,
      "step": 898
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.395556537053278e-05,
      "loss": 1.1581,
      "step": 899
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.394177548166864e-05,
      "loss": 1.4094,
      "step": 900
    },
    {
      "epoch": 0.49,
      "eval_loss": 1.3968260288238525,
      "eval_runtime": 36.8023,
      "eval_samples_per_second": 3.831,
      "eval_steps_per_second": 1.929,
      "step": 900
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.39279708948191e-05,
      "loss": 1.3184,
      "step": 901
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.391415161460162e-05,
      "loss": 1.3996,
      "step": 902
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.39003176456386e-05,
      "loss": 1.2233,
      "step": 903
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.388646899255733e-05,
      "loss": 1.4305,
      "step": 904
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.387260565999e-05,
      "loss": 1.6054,
      "step": 905
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.385872765257373e-05,
      "loss": 1.3534,
      "step": 906
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.384483497495055e-05,
      "loss": 1.6175,
      "step": 907
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.38309276317674e-05,
      "loss": 1.5101,
      "step": 908
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.381700562767609e-05,
      "loss": 1.2654,
      "step": 909
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.380306896733336e-05,
      "loss": 1.4425,
      "step": 910
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.378911765540088e-05,
      "loss": 1.429,
      "step": 911
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.377515169654518e-05,
      "loss": 1.512,
      "step": 912
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.376117109543769e-05,
      "loss": 1.3809,
      "step": 913
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.374717585675476e-05,
      "loss": 1.381,
      "step": 914
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.373316598517763e-05,
      "loss": 1.4734,
      "step": 915
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.371914148539243e-05,
      "loss": 1.5823,
      "step": 916
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.370510236209017e-05,
      "loss": 1.4174,
      "step": 917
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.369104861996677e-05,
      "loss": 1.2308,
      "step": 918
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.367698026372305e-05,
      "loss": 1.3034,
      "step": 919
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.366289729806468e-05,
      "loss": 1.5157,
      "step": 920
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.364879972770227e-05,
      "loss": 1.3952,
      "step": 921
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.363468755735123e-05,
      "loss": 1.4611,
      "step": 922
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.362056079173197e-05,
      "loss": 1.6004,
      "step": 923
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.360641943556969e-05,
      "loss": 1.5868,
      "step": 924
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.359226349359449e-05,
      "loss": 1.6087,
      "step": 925
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.357809297054139e-05,
      "loss": 1.2811,
      "step": 926
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.356390787115023e-05,
      "loss": 1.2932,
      "step": 927
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.354970820016576e-05,
      "loss": 1.4741,
      "step": 928
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.353549396233758e-05,
      "loss": 1.2006,
      "step": 929
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.352126516242021e-05,
      "loss": 1.4901,
      "step": 930
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.3507021805173e-05,
      "loss": 1.1914,
      "step": 931
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.349276389536017e-05,
      "loss": 1.6753,
      "step": 932
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.347849143775082e-05,
      "loss": 1.4431,
      "step": 933
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.34642044371189e-05,
      "loss": 1.4558,
      "step": 934
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.344990289824325e-05,
      "loss": 1.5534,
      "step": 935
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.343558682590756e-05,
      "loss": 1.3927,
      "step": 936
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.342125622490037e-05,
      "loss": 1.426,
      "step": 937
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.340691110001511e-05,
      "loss": 1.2195,
      "step": 938
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.339255145605003e-05,
      "loss": 1.24,
      "step": 939
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.337817729780827e-05,
      "loss": 1.1934,
      "step": 940
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.336378863009778e-05,
      "loss": 1.2773,
      "step": 941
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.334938545773142e-05,
      "loss": 1.2583,
      "step": 942
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.333496778552689e-05,
      "loss": 1.5928,
      "step": 943
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.332053561830669e-05,
      "loss": 1.2163,
      "step": 944
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.330608896089821e-05,
      "loss": 1.4654,
      "step": 945
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.329162781813369e-05,
      "loss": 1.6089,
      "step": 946
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.327715219485021e-05,
      "loss": 1.441,
      "step": 947
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.326266209588966e-05,
      "loss": 1.3098,
      "step": 948
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.324815752609884e-05,
      "loss": 1.2131,
      "step": 949
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.323363849032933e-05,
      "loss": 1.4135,
      "step": 950
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.321910499343756e-05,
      "loss": 1.5087,
      "step": 951
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.320455704028481e-05,
      "loss": 1.3965,
      "step": 952
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.318999463573721e-05,
      "loss": 1.4109,
      "step": 953
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.31754177846657e-05,
      "loss": 1.5512,
      "step": 954
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.316082649194606e-05,
      "loss": 1.2591,
      "step": 955
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.314622076245887e-05,
      "loss": 1.2841,
      "step": 956
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.313160060108961e-05,
      "loss": 1.3919,
      "step": 957
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.311696601272852e-05,
      "loss": 1.0574,
      "step": 958
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.310231700227068e-05,
      "loss": 1.4227,
      "step": 959
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.308765357461604e-05,
      "loss": 1.3592,
      "step": 960
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.307297573466931e-05,
      "loss": 1.2357,
      "step": 961
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.305828348734005e-05,
      "loss": 1.2483,
      "step": 962
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.304357683754264e-05,
      "loss": 1.1654,
      "step": 963
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.302885579019627e-05,
      "loss": 1.5328,
      "step": 964
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.301412035022496e-05,
      "loss": 1.3868,
      "step": 965
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.299937052255752e-05,
      "loss": 1.3587,
      "step": 966
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.29846063121276e-05,
      "loss": 1.5776,
      "step": 967
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.296982772387365e-05,
      "loss": 1.5431,
      "step": 968
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.295503476273892e-05,
      "loss": 1.334,
      "step": 969
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.294022743367145e-05,
      "loss": 1.4464,
      "step": 970
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.292540574162416e-05,
      "loss": 1.2866,
      "step": 971
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.291056969155469e-05,
      "loss": 1.6356,
      "step": 972
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.289571928842553e-05,
      "loss": 1.4209,
      "step": 973
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.288085453720394e-05,
      "loss": 1.5093,
      "step": 974
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.286597544286202e-05,
      "loss": 1.3069,
      "step": 975
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.285108201037662e-05,
      "loss": 1.4576,
      "step": 976
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.283617424472943e-05,
      "loss": 1.2648,
      "step": 977
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.282125215090694e-05,
      "loss": 1.3521,
      "step": 978
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.280631573390033e-05,
      "loss": 1.4213,
      "step": 979
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.279136499870574e-05,
      "loss": 1.3894,
      "step": 980
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.277639995032394e-05,
      "loss": 1.4639,
      "step": 981
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.276142059376057e-05,
      "loss": 1.472,
      "step": 982
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.274642693402606e-05,
      "loss": 1.646,
      "step": 983
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.27314189761356e-05,
      "loss": 1.5777,
      "step": 984
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.271639672510917e-05,
      "loss": 1.4036,
      "step": 985
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.270136018597151e-05,
      "loss": 1.3505,
      "step": 986
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.268630936375217e-05,
      "loss": 1.476,
      "step": 987
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.267124426348548e-05,
      "loss": 1.5432,
      "step": 988
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.26561648902105e-05,
      "loss": 1.6045,
      "step": 989
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.264107124897113e-05,
      "loss": 1.3145,
      "step": 990
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.262596334481598e-05,
      "loss": 1.4727,
      "step": 991
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.261084118279847e-05,
      "loss": 1.4644,
      "step": 992
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.259570476797678e-05,
      "loss": 1.0788,
      "step": 993
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.258055410541385e-05,
      "loss": 1.3453,
      "step": 994
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.256538920017738e-05,
      "loss": 1.2995,
      "step": 995
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.255021005733989e-05,
      "loss": 1.4042,
      "step": 996
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.253501668197857e-05,
      "loss": 1.3593,
      "step": 997
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.251980907917544e-05,
      "loss": 1.3195,
      "step": 998
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.250458725401725e-05,
      "loss": 1.5335,
      "step": 999
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.24893512115955e-05,
      "loss": 1.4042,
      "step": 1000
    },
    {
      "epoch": 0.55,
      "eval_loss": 1.384839415550232,
      "eval_runtime": 36.8767,
      "eval_samples_per_second": 3.824,
      "eval_steps_per_second": 1.925,
      "step": 1000
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.24741009570065e-05,
      "loss": 1.5173,
      "step": 1001
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.245883649535123e-05,
      "loss": 1.4023,
      "step": 1002
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.244355783173548e-05,
      "loss": 1.4732,
      "step": 1003
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.24282649712698e-05,
      "loss": 1.3597,
      "step": 1004
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.241295791906943e-05,
      "loss": 1.5777,
      "step": 1005
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.239763668025439e-05,
      "loss": 1.3677,
      "step": 1006
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.238230125994949e-05,
      "loss": 1.639,
      "step": 1007
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.236695166328419e-05,
      "loss": 1.6893,
      "step": 1008
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.235158789539278e-05,
      "loss": 1.3066,
      "step": 1009
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.233620996141421e-05,
      "loss": 1.3495,
      "step": 1010
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.232081786649225e-05,
      "loss": 2.0538,
      "step": 1011
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.230541161577535e-05,
      "loss": 1.3891,
      "step": 1012
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.228999121441672e-05,
      "loss": 1.5153,
      "step": 1013
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.227455666757429e-05,
      "loss": 1.4505,
      "step": 1014
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.225910798041073e-05,
      "loss": 1.5768,
      "step": 1015
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.224364515809343e-05,
      "loss": 1.2983,
      "step": 1016
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.222816820579452e-05,
      "loss": 1.5059,
      "step": 1017
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.221267712869084e-05,
      "loss": 1.3279,
      "step": 1018
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.219717193196397e-05,
      "loss": 1.5534,
      "step": 1019
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.218165262080023e-05,
      "loss": 1.1876,
      "step": 1020
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.216611920039062e-05,
      "loss": 1.4348,
      "step": 1021
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.215057167593087e-05,
      "loss": 1.422,
      "step": 1022
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.213501005262147e-05,
      "loss": 1.16,
      "step": 1023
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.211943433566755e-05,
      "loss": 1.216,
      "step": 1024
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.2103844530279e-05,
      "loss": 1.2264,
      "step": 1025
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.208824064167047e-05,
      "loss": 1.518,
      "step": 1026
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.207262267506122e-05,
      "loss": 1.537,
      "step": 1027
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.205699063567527e-05,
      "loss": 1.1935,
      "step": 1028
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.204134452874137e-05,
      "loss": 1.1592,
      "step": 1029
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.202568435949294e-05,
      "loss": 1.5657,
      "step": 1030
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.201001013316812e-05,
      "loss": 1.4023,
      "step": 1031
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.199432185500973e-05,
      "loss": 1.4504,
      "step": 1032
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.197861953026532e-05,
      "loss": 1.4372,
      "step": 1033
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.196290316418711e-05,
      "loss": 1.4757,
      "step": 1034
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.194717276203208e-05,
      "loss": 1.3044,
      "step": 1035
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.193142832906179e-05,
      "loss": 1.263,
      "step": 1036
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.191566987054261e-05,
      "loss": 1.6228,
      "step": 1037
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.189989739174552e-05,
      "loss": 1.2383,
      "step": 1038
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.188411089794624e-05,
      "loss": 1.3616,
      "step": 1039
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.186831039442514e-05,
      "loss": 1.2849,
      "step": 1040
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.185249588646732e-05,
      "loss": 1.3404,
      "step": 1041
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.183666737936251e-05,
      "loss": 1.2606,
      "step": 1042
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.182082487840519e-05,
      "loss": 1.7579,
      "step": 1043
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.180496838889445e-05,
      "loss": 1.3049,
      "step": 1044
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.17890979161341e-05,
      "loss": 1.384,
      "step": 1045
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.177321346543262e-05,
      "loss": 1.2359,
      "step": 1046
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.175731504210316e-05,
      "loss": 1.4374,
      "step": 1047
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.174140265146356e-05,
      "loss": 1.6802,
      "step": 1048
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.172547629883632e-05,
      "loss": 1.492,
      "step": 1049
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.170953598954859e-05,
      "loss": 1.1791,
      "step": 1050
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.169358172893221e-05,
      "loss": 1.3224,
      "step": 1051
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.167761352232371e-05,
      "loss": 1.4983,
      "step": 1052
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.166163137506423e-05,
      "loss": 1.4214,
      "step": 1053
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.164563529249963e-05,
      "loss": 1.3126,
      "step": 1054
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.162962527998038e-05,
      "loss": 1.5933,
      "step": 1055
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.161360134286166e-05,
      "loss": 1.2852,
      "step": 1056
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.159756348650324e-05,
      "loss": 1.5764,
      "step": 1057
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.158151171626961e-05,
      "loss": 1.5588,
      "step": 1058
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.15654460375299e-05,
      "loss": 1.3055,
      "step": 1059
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.154936645565787e-05,
      "loss": 1.4728,
      "step": 1060
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.153327297603195e-05,
      "loss": 1.3558,
      "step": 1061
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.15171656040352e-05,
      "loss": 1.4781,
      "step": 1062
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.150104434505535e-05,
      "loss": 1.5669,
      "step": 1063
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.148490920448477e-05,
      "loss": 1.2851,
      "step": 1064
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.146876018772044e-05,
      "loss": 1.1777,
      "step": 1065
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.145259730016402e-05,
      "loss": 1.2743,
      "step": 1066
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.143642054722183e-05,
      "loss": 1.4441,
      "step": 1067
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.142022993430475e-05,
      "loss": 1.3422,
      "step": 1068
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.140402546682836e-05,
      "loss": 1.2159,
      "step": 1069
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.138780715021285e-05,
      "loss": 1.5294,
      "step": 1070
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.137157498988306e-05,
      "loss": 1.2654,
      "step": 1071
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.135532899126844e-05,
      "loss": 1.3753,
      "step": 1072
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.133906915980308e-05,
      "loss": 1.2395,
      "step": 1073
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.13227955009257e-05,
      "loss": 1.4715,
      "step": 1074
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.130650802007963e-05,
      "loss": 1.3072,
      "step": 1075
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.129020672271283e-05,
      "loss": 1.5101,
      "step": 1076
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.12738916142779e-05,
      "loss": 1.4501,
      "step": 1077
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.125756270023203e-05,
      "loss": 1.3527,
      "step": 1078
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.124121998603704e-05,
      "loss": 1.6469,
      "step": 1079
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.122486347715937e-05,
      "loss": 1.4204,
      "step": 1080
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.12084931790701e-05,
      "loss": 1.4826,
      "step": 1081
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.119210909724485e-05,
      "loss": 1.3207,
      "step": 1082
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.117571123716391e-05,
      "loss": 1.3103,
      "step": 1083
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.115929960431217e-05,
      "loss": 1.3407,
      "step": 1084
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.114287420417913e-05,
      "loss": 1.3765,
      "step": 1085
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.112643504225886e-05,
      "loss": 1.2456,
      "step": 1086
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.110998212405008e-05,
      "loss": 1.4621,
      "step": 1087
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.109351545505607e-05,
      "loss": 1.5254,
      "step": 1088
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.107703504078475e-05,
      "loss": 1.2145,
      "step": 1089
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.10605408867486e-05,
      "loss": 1.4058,
      "step": 1090
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.104403299846471e-05,
      "loss": 1.3062,
      "step": 1091
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.10275113814548e-05,
      "loss": 1.4311,
      "step": 1092
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.101097604124508e-05,
      "loss": 1.3034,
      "step": 1093
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.099442698336649e-05,
      "loss": 1.3926,
      "step": 1094
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.097786421335446e-05,
      "loss": 1.3116,
      "step": 1095
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.096128773674902e-05,
      "loss": 1.5093,
      "step": 1096
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.094469755909483e-05,
      "loss": 1.2885,
      "step": 1097
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.092809368594107e-05,
      "loss": 1.2197,
      "step": 1098
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.091147612284154e-05,
      "loss": 1.3578,
      "step": 1099
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.08948448753546e-05,
      "loss": 1.3946,
      "step": 1100
    },
    {
      "epoch": 0.6,
      "eval_loss": 1.3771336078643799,
      "eval_runtime": 36.9035,
      "eval_samples_per_second": 3.821,
      "eval_steps_per_second": 1.924,
      "step": 1100
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.087819994904324e-05,
      "loss": 1.748,
      "step": 1101
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.086154134947494e-05,
      "loss": 1.4167,
      "step": 1102
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.084486908222182e-05,
      "loss": 1.3661,
      "step": 1103
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.082818315286055e-05,
      "loss": 1.2332,
      "step": 1104
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.081148356697236e-05,
      "loss": 1.4766,
      "step": 1105
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.079477033014306e-05,
      "loss": 1.4262,
      "step": 1106
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.077804344796302e-05,
      "loss": 1.3386,
      "step": 1107
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.076130292602717e-05,
      "loss": 1.3211,
      "step": 1108
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.074454876993504e-05,
      "loss": 1.4328,
      "step": 1109
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.072778098529064e-05,
      "loss": 1.1087,
      "step": 1110
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.071099957770264e-05,
      "loss": 1.3823,
      "step": 1111
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.069420455278419e-05,
      "loss": 1.7021,
      "step": 1112
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.067739591615301e-05,
      "loss": 1.3176,
      "step": 1113
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.06605736734314e-05,
      "loss": 1.3502,
      "step": 1114
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.064373783024619e-05,
      "loss": 1.4063,
      "step": 1115
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.062688839222877e-05,
      "loss": 1.7037,
      "step": 1116
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.061002536501506e-05,
      "loss": 1.4425,
      "step": 1117
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.059314875424553e-05,
      "loss": 1.5067,
      "step": 1118
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.057625856556522e-05,
      "loss": 1.3296,
      "step": 1119
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.055935480462367e-05,
      "loss": 1.2994,
      "step": 1120
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.0542437477075e-05,
      "loss": 1.2041,
      "step": 1121
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.052550658857783e-05,
      "loss": 1.4281,
      "step": 1122
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.050856214479533e-05,
      "loss": 1.6111,
      "step": 1123
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.049160415139525e-05,
      "loss": 1.2628,
      "step": 1124
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.04746326140498e-05,
      "loss": 1.5526,
      "step": 1125
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.045764753843575e-05,
      "loss": 1.2943,
      "step": 1126
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.044064893023441e-05,
      "loss": 1.6136,
      "step": 1127
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.042363679513158e-05,
      "loss": 1.212,
      "step": 1128
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.040661113881765e-05,
      "loss": 1.471,
      "step": 1129
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.038957196698748e-05,
      "loss": 1.6176,
      "step": 1130
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.037251928534043e-05,
      "loss": 1.3584,
      "step": 1131
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.035545309958046e-05,
      "loss": 1.4925,
      "step": 1132
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.0338373415416e-05,
      "loss": 1.521,
      "step": 1133
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.032128023855994e-05,
      "loss": 1.4746,
      "step": 1134
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.03041735747298e-05,
      "loss": 1.5895,
      "step": 1135
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.028705342964753e-05,
      "loss": 1.0965,
      "step": 1136
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.026991980903959e-05,
      "loss": 1.3186,
      "step": 1137
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.025277271863699e-05,
      "loss": 1.5158,
      "step": 1138
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.023561216417519e-05,
      "loss": 1.4753,
      "step": 1139
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.021843815139423e-05,
      "loss": 1.5086,
      "step": 1140
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.020125068603858e-05,
      "loss": 1.1635,
      "step": 1141
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.018404977385723e-05,
      "loss": 1.3392,
      "step": 1142
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.016683542060368e-05,
      "loss": 1.3288,
      "step": 1143
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.014960763203592e-05,
      "loss": 1.5322,
      "step": 1144
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.013236641391647e-05,
      "loss": 1.398,
      "step": 1145
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.011511177201225e-05,
      "loss": 1.3598,
      "step": 1146
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.009784371209475e-05,
      "loss": 1.4684,
      "step": 1147
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.008056223993993e-05,
      "loss": 1.6218,
      "step": 1148
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.006326736132822e-05,
      "loss": 1.2508,
      "step": 1149
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.004595908204456e-05,
      "loss": 1.3228,
      "step": 1150
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.002863740787835e-05,
      "loss": 1.3351,
      "step": 1151
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.001130234462347e-05,
      "loss": 1.3702,
      "step": 1152
    },
    {
      "epoch": 0.63,
      "learning_rate": 8.99939538980783e-05,
      "loss": 1.2463,
      "step": 1153
    },
    {
      "epoch": 0.63,
      "learning_rate": 8.997659207404566e-05,
      "loss": 1.3602,
      "step": 1154
    },
    {
      "epoch": 0.63,
      "learning_rate": 8.995921687833289e-05,
      "loss": 1.3852,
      "step": 1155
    },
    {
      "epoch": 0.63,
      "learning_rate": 8.994182831675176e-05,
      "loss": 1.5222,
      "step": 1156
    },
    {
      "epoch": 0.63,
      "learning_rate": 8.992442639511856e-05,
      "loss": 1.3915,
      "step": 1157
    },
    {
      "epoch": 0.63,
      "learning_rate": 8.990701111925399e-05,
      "loss": 1.3468,
      "step": 1158
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.988958249498327e-05,
      "loss": 1.2885,
      "step": 1159
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.987214052813604e-05,
      "loss": 1.5726,
      "step": 1160
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.985468522454645e-05,
      "loss": 1.2292,
      "step": 1161
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.983721659005305e-05,
      "loss": 1.5092,
      "step": 1162
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.98197346304989e-05,
      "loss": 1.3191,
      "step": 1163
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.980223935173153e-05,
      "loss": 1.4734,
      "step": 1164
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.978473075960285e-05,
      "loss": 1.4559,
      "step": 1165
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.97672088599693e-05,
      "loss": 1.4137,
      "step": 1166
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.974967365869173e-05,
      "loss": 1.6625,
      "step": 1167
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.973212516163545e-05,
      "loss": 1.2993,
      "step": 1168
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.971456337467024e-05,
      "loss": 1.432,
      "step": 1169
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.969698830367024e-05,
      "loss": 1.1423,
      "step": 1170
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.967939995451416e-05,
      "loss": 1.4361,
      "step": 1171
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.966179833308506e-05,
      "loss": 1.249,
      "step": 1172
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.964418344527047e-05,
      "loss": 1.3598,
      "step": 1173
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.962655529696236e-05,
      "loss": 1.4405,
      "step": 1174
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.960891389405712e-05,
      "loss": 1.3278,
      "step": 1175
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.959125924245559e-05,
      "loss": 1.3722,
      "step": 1176
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.957359134806302e-05,
      "loss": 1.1758,
      "step": 1177
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.955591021678913e-05,
      "loss": 1.3478,
      "step": 1178
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.953821585454803e-05,
      "loss": 1.4414,
      "step": 1179
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.952050826725826e-05,
      "loss": 1.4023,
      "step": 1180
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.95027874608428e-05,
      "loss": 1.3108,
      "step": 1181
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.948505344122904e-05,
      "loss": 1.3868,
      "step": 1182
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.946730621434877e-05,
      "loss": 1.633,
      "step": 1183
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.944954578613827e-05,
      "loss": 1.1175,
      "step": 1184
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.943177216253812e-05,
      "loss": 1.1517,
      "step": 1185
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.941398534949345e-05,
      "loss": 1.206,
      "step": 1186
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.939618535295368e-05,
      "loss": 1.3396,
      "step": 1187
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.937837217887273e-05,
      "loss": 1.5007,
      "step": 1188
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.936054583320884e-05,
      "loss": 1.4411,
      "step": 1189
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.934270632192474e-05,
      "loss": 1.2795,
      "step": 1190
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.932485365098752e-05,
      "loss": 1.3769,
      "step": 1191
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.930698782636867e-05,
      "loss": 1.4882,
      "step": 1192
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.928910885404411e-05,
      "loss": 1.4823,
      "step": 1193
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.927121673999411e-05,
      "loss": 1.2508,
      "step": 1194
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.925331149020339e-05,
      "loss": 1.5334,
      "step": 1195
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.9235393110661e-05,
      "loss": 1.4646,
      "step": 1196
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.921746160736046e-05,
      "loss": 1.4672,
      "step": 1197
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.919951698629962e-05,
      "loss": 1.2901,
      "step": 1198
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.918155925348075e-05,
      "loss": 1.4212,
      "step": 1199
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.916358841491046e-05,
      "loss": 1.2523,
      "step": 1200
    },
    {
      "epoch": 0.66,
      "eval_loss": 1.3691751956939697,
      "eval_runtime": 36.9064,
      "eval_samples_per_second": 3.82,
      "eval_steps_per_second": 1.924,
      "step": 1200
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.914560447659978e-05,
      "loss": 1.5167,
      "step": 1201
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.912760744456415e-05,
      "loss": 1.1834,
      "step": 1202
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.910959732482333e-05,
      "loss": 1.6289,
      "step": 1203
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.90915741234015e-05,
      "loss": 1.4629,
      "step": 1204
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.907353784632717e-05,
      "loss": 1.4562,
      "step": 1205
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.90554884996333e-05,
      "loss": 1.4629,
      "step": 1206
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.903742608935714e-05,
      "loss": 1.536,
      "step": 1207
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.901935062154034e-05,
      "loss": 1.4759,
      "step": 1208
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.900126210222894e-05,
      "loss": 1.197,
      "step": 1209
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.898316053747334e-05,
      "loss": 1.5119,
      "step": 1210
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.896504593332827e-05,
      "loss": 1.0428,
      "step": 1211
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.894691829585285e-05,
      "loss": 1.3722,
      "step": 1212
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.892877763111056e-05,
      "loss": 1.5563,
      "step": 1213
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.89106239451692e-05,
      "loss": 1.1695,
      "step": 1214
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.8892457244101e-05,
      "loss": 1.144,
      "step": 1215
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.887427753398248e-05,
      "loss": 1.1856,
      "step": 1216
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.885608482089451e-05,
      "loss": 1.3873,
      "step": 1217
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.883787911092235e-05,
      "loss": 1.4795,
      "step": 1218
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.881966041015559e-05,
      "loss": 1.2607,
      "step": 1219
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.880142872468815e-05,
      "loss": 1.094,
      "step": 1220
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.87831840606183e-05,
      "loss": 1.6404,
      "step": 1221
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.876492642404869e-05,
      "loss": 1.1459,
      "step": 1222
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.874665582108625e-05,
      "loss": 0.9399,
      "step": 1223
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.872837225784226e-05,
      "loss": 1.2831,
      "step": 1224
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.871007574043237e-05,
      "loss": 1.3696,
      "step": 1225
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.869176627497654e-05,
      "loss": 1.5,
      "step": 1226
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.867344386759906e-05,
      "loss": 1.4983,
      "step": 1227
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.865510852442854e-05,
      "loss": 1.4904,
      "step": 1228
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.863676025159793e-05,
      "loss": 1.3333,
      "step": 1229
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.861839905524452e-05,
      "loss": 1.3232,
      "step": 1230
    },
    {
      "epoch": 0.67,
      "learning_rate": 8.860002494150988e-05,
      "loss": 1.4308,
      "step": 1231
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.858163791653994e-05,
      "loss": 1.4426,
      "step": 1232
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.856323798648494e-05,
      "loss": 1.4245,
      "step": 1233
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.854482515749943e-05,
      "loss": 1.691,
      "step": 1234
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.852639943574225e-05,
      "loss": 1.4547,
      "step": 1235
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.85079608273766e-05,
      "loss": 1.4453,
      "step": 1236
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.848950933856998e-05,
      "loss": 1.3603,
      "step": 1237
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.847104497549417e-05,
      "loss": 1.2526,
      "step": 1238
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.845256774432527e-05,
      "loss": 1.3324,
      "step": 1239
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.84340776512437e-05,
      "loss": 1.3989,
      "step": 1240
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.841557470243416e-05,
      "loss": 1.3431,
      "step": 1241
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.839705890408564e-05,
      "loss": 1.1767,
      "step": 1242
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.837853026239152e-05,
      "loss": 1.5477,
      "step": 1243
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.835998878354931e-05,
      "loss": 1.4905,
      "step": 1244
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.834143447376096e-05,
      "loss": 1.3825,
      "step": 1245
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.832286733923266e-05,
      "loss": 1.6241,
      "step": 1246
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.830428738617489e-05,
      "loss": 1.8487,
      "step": 1247
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.828569462080238e-05,
      "loss": 1.2949,
      "step": 1248
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.826708904933423e-05,
      "loss": 1.2304,
      "step": 1249
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.824847067799374e-05,
      "loss": 1.3122,
      "step": 1250
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.822983951300855e-05,
      "loss": 1.4431,
      "step": 1251
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.821119556061055e-05,
      "loss": 1.1996,
      "step": 1252
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.819253882703591e-05,
      "loss": 1.5836,
      "step": 1253
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.817386931852507e-05,
      "loss": 1.2332,
      "step": 1254
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.815518704132276e-05,
      "loss": 1.5345,
      "step": 1255
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.813649200167799e-05,
      "loss": 1.1864,
      "step": 1256
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.811778420584399e-05,
      "loss": 1.3666,
      "step": 1257
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.809906366007832e-05,
      "loss": 1.4968,
      "step": 1258
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.808033037064273e-05,
      "loss": 1.2667,
      "step": 1259
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.806158434380333e-05,
      "loss": 1.3871,
      "step": 1260
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.804282558583041e-05,
      "loss": 1.5239,
      "step": 1261
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.802405410299854e-05,
      "loss": 1.44,
      "step": 1262
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.800526990158655e-05,
      "loss": 1.4118,
      "step": 1263
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.798647298787754e-05,
      "loss": 1.5034,
      "step": 1264
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.796766336815883e-05,
      "loss": 1.2579,
      "step": 1265
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.7948841048722e-05,
      "loss": 1.4044,
      "step": 1266
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.793000603586292e-05,
      "loss": 1.2475,
      "step": 1267
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.791115833588164e-05,
      "loss": 1.4576,
      "step": 1268
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.789229795508251e-05,
      "loss": 1.3182,
      "step": 1269
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.787342489977409e-05,
      "loss": 1.3418,
      "step": 1270
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.785453917626915e-05,
      "loss": 1.4359,
      "step": 1271
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.783564079088477e-05,
      "loss": 1.3881,
      "step": 1272
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.781672974994221e-05,
      "loss": 1.3815,
      "step": 1273
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.7797806059767e-05,
      "loss": 1.4785,
      "step": 1274
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.777886972668889e-05,
      "loss": 1.2328,
      "step": 1275
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.775992075704182e-05,
      "loss": 1.1147,
      "step": 1276
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.774095915716401e-05,
      "loss": 1.4431,
      "step": 1277
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.772198493339784e-05,
      "loss": 1.3362,
      "step": 1278
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.770299809209004e-05,
      "loss": 1.3783,
      "step": 1279
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.76839986395914e-05,
      "loss": 1.3849,
      "step": 1280
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.766498658225703e-05,
      "loss": 1.5028,
      "step": 1281
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.764596192644623e-05,
      "loss": 1.4,
      "step": 1282
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.762692467852253e-05,
      "loss": 1.2857,
      "step": 1283
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.760787484485363e-05,
      "loss": 1.393,
      "step": 1284
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.758881243181147e-05,
      "loss": 1.4945,
      "step": 1285
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.756973744577221e-05,
      "loss": 1.57,
      "step": 1286
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.755064989311618e-05,
      "loss": 1.318,
      "step": 1287
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.753154978022795e-05,
      "loss": 1.3036,
      "step": 1288
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.751243711349628e-05,
      "loss": 1.4148,
      "step": 1289
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.749331189931409e-05,
      "loss": 1.5207,
      "step": 1290
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.747417414407855e-05,
      "loss": 1.2803,
      "step": 1291
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.7455023854191e-05,
      "loss": 1.3997,
      "step": 1292
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.743586103605698e-05,
      "loss": 1.301,
      "step": 1293
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.741668569608622e-05,
      "loss": 1.5071,
      "step": 1294
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.739749784069264e-05,
      "loss": 1.4687,
      "step": 1295
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.737829747629432e-05,
      "loss": 1.6001,
      "step": 1296
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.735908460931358e-05,
      "loss": 1.0613,
      "step": 1297
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.733985924617687e-05,
      "loss": 1.3222,
      "step": 1298
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.732062139331483e-05,
      "loss": 1.6002,
      "step": 1299
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.73013710571623e-05,
      "loss": 1.2932,
      "step": 1300
    },
    {
      "epoch": 0.71,
      "eval_loss": 1.3647781610488892,
      "eval_runtime": 36.8915,
      "eval_samples_per_second": 3.822,
      "eval_steps_per_second": 1.925,
      "step": 1300
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.728210824415827e-05,
      "loss": 1.0652,
      "step": 1301
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.726283296074596e-05,
      "loss": 1.5468,
      "step": 1302
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.724354521337264e-05,
      "loss": 1.3485,
      "step": 1303
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.722424500848987e-05,
      "loss": 1.275,
      "step": 1304
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.720493235255333e-05,
      "loss": 1.2041,
      "step": 1305
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.718560725202288e-05,
      "loss": 1.4242,
      "step": 1306
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.716626971336248e-05,
      "loss": 1.3271,
      "step": 1307
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.714691974304035e-05,
      "loss": 1.4929,
      "step": 1308
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.71275573475288e-05,
      "loss": 1.3621,
      "step": 1309
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.71081825333043e-05,
      "loss": 1.2402,
      "step": 1310
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.708879530684751e-05,
      "loss": 1.5312,
      "step": 1311
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.706939567464321e-05,
      "loss": 1.3971,
      "step": 1312
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.704998364318035e-05,
      "loss": 1.5908,
      "step": 1313
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.7030559218952e-05,
      "loss": 1.137,
      "step": 1314
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.701112240845539e-05,
      "loss": 1.3276,
      "step": 1315
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.699167321819191e-05,
      "loss": 1.1859,
      "step": 1316
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.697221165466707e-05,
      "loss": 1.2808,
      "step": 1317
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.695273772439052e-05,
      "loss": 1.443,
      "step": 1318
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.693325143387606e-05,
      "loss": 1.3168,
      "step": 1319
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.691375278964162e-05,
      "loss": 1.2272,
      "step": 1320
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.689424179820923e-05,
      "loss": 1.2386,
      "step": 1321
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.68747184661051e-05,
      "loss": 1.3734,
      "step": 1322
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.685518279985954e-05,
      "loss": 1.4561,
      "step": 1323
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.6835634806007e-05,
      "loss": 1.3355,
      "step": 1324
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.681607449108603e-05,
      "loss": 1.2502,
      "step": 1325
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.679650186163932e-05,
      "loss": 1.5251,
      "step": 1326
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.677691692421368e-05,
      "loss": 1.2511,
      "step": 1327
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.675731968536002e-05,
      "loss": 1.4981,
      "step": 1328
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.673771015163339e-05,
      "loss": 0.9665,
      "step": 1329
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.671808832959294e-05,
      "loss": 1.1603,
      "step": 1330
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.669845422580191e-05,
      "loss": 1.0915,
      "step": 1331
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.66788078468277e-05,
      "loss": 1.4325,
      "step": 1332
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.665914919924175e-05,
      "loss": 1.5095,
      "step": 1333
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.663947828961963e-05,
      "loss": 1.3491,
      "step": 1334
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.661979512454104e-05,
      "loss": 1.4016,
      "step": 1335
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.660009971058978e-05,
      "loss": 1.1411,
      "step": 1336
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.658039205435369e-05,
      "loss": 1.2492,
      "step": 1337
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.656067216242474e-05,
      "loss": 1.3803,
      "step": 1338
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.654094004139899e-05,
      "loss": 1.8103,
      "step": 1339
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.652119569787663e-05,
      "loss": 1.3484,
      "step": 1340
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.650143913846186e-05,
      "loss": 1.2641,
      "step": 1341
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.648167036976302e-05,
      "loss": 1.3491,
      "step": 1342
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.646188939839252e-05,
      "loss": 1.4852,
      "step": 1343
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.644209623096686e-05,
      "loss": 1.3375,
      "step": 1344
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.64222908741066e-05,
      "loss": 1.3609,
      "step": 1345
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.64024733344364e-05,
      "loss": 1.3838,
      "step": 1346
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.6382643618585e-05,
      "loss": 1.2436,
      "step": 1347
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.636280173318515e-05,
      "loss": 1.5688,
      "step": 1348
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.634294768487375e-05,
      "loss": 1.4802,
      "step": 1349
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.632308148029173e-05,
      "loss": 1.4417,
      "step": 1350
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.630320312608411e-05,
      "loss": 1.2505,
      "step": 1351
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.628331262889991e-05,
      "loss": 1.2601,
      "step": 1352
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.62634099953923e-05,
      "loss": 1.2984,
      "step": 1353
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.624349523221847e-05,
      "loss": 1.2796,
      "step": 1354
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.622356834603961e-05,
      "loss": 1.4448,
      "step": 1355
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.620362934352109e-05,
      "loss": 1.4312,
      "step": 1356
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.61836782313322e-05,
      "loss": 1.3668,
      "step": 1357
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.616371501614638e-05,
      "loss": 1.3455,
      "step": 1358
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.614373970464108e-05,
      "loss": 1.1842,
      "step": 1359
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.612375230349778e-05,
      "loss": 1.3202,
      "step": 1360
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.610375281940204e-05,
      "loss": 1.4735,
      "step": 1361
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.608374125904341e-05,
      "loss": 1.4877,
      "step": 1362
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.606371762911556e-05,
      "loss": 1.6568,
      "step": 1363
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.604368193631611e-05,
      "loss": 1.5026,
      "step": 1364
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.602363418734676e-05,
      "loss": 1.3766,
      "step": 1365
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.600357438891325e-05,
      "loss": 1.3497,
      "step": 1366
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.598350254772531e-05,
      "loss": 1.6687,
      "step": 1367
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.596341867049677e-05,
      "loss": 1.4173,
      "step": 1368
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.594332276394539e-05,
      "loss": 1.3051,
      "step": 1369
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.592321483479305e-05,
      "loss": 1.4397,
      "step": 1370
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.590309488976558e-05,
      "loss": 1.248,
      "step": 1371
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.588296293559286e-05,
      "loss": 1.3414,
      "step": 1372
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.586281897900877e-05,
      "loss": 1.3538,
      "step": 1373
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.584266302675124e-05,
      "loss": 1.4986,
      "step": 1374
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.582249508556217e-05,
      "loss": 1.5249,
      "step": 1375
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.58023151621875e-05,
      "loss": 1.2847,
      "step": 1376
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.578212326337715e-05,
      "loss": 1.2832,
      "step": 1377
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.576191939588509e-05,
      "loss": 1.4069,
      "step": 1378
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.574170356646925e-05,
      "loss": 1.432,
      "step": 1379
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.572147578189159e-05,
      "loss": 1.0847,
      "step": 1380
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.570123604891804e-05,
      "loss": 1.2856,
      "step": 1381
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.568098437431854e-05,
      "loss": 1.5063,
      "step": 1382
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.566072076486704e-05,
      "loss": 1.5753,
      "step": 1383
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.564044522734147e-05,
      "loss": 1.5486,
      "step": 1384
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.562015776852373e-05,
      "loss": 1.3689,
      "step": 1385
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.559985839519975e-05,
      "loss": 1.1437,
      "step": 1386
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.557954711415941e-05,
      "loss": 1.1885,
      "step": 1387
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.555922393219658e-05,
      "loss": 1.4003,
      "step": 1388
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.553888885610914e-05,
      "loss": 1.5752,
      "step": 1389
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.551854189269888e-05,
      "loss": 1.2865,
      "step": 1390
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.549818304877164e-05,
      "loss": 1.3903,
      "step": 1391
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.54778123311372e-05,
      "loss": 1.3444,
      "step": 1392
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.545742974660933e-05,
      "loss": 1.193,
      "step": 1393
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.543703530200572e-05,
      "loss": 1.642,
      "step": 1394
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.541662900414808e-05,
      "loss": 1.4944,
      "step": 1395
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.539621085986208e-05,
      "loss": 1.4039,
      "step": 1396
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.537578087597731e-05,
      "loss": 1.2408,
      "step": 1397
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.535533905932738e-05,
      "loss": 1.5909,
      "step": 1398
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.533488541674981e-05,
      "loss": 1.327,
      "step": 1399
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.531441995508609e-05,
      "loss": 1.346,
      "step": 1400
    },
    {
      "epoch": 0.77,
      "eval_loss": 1.3608676195144653,
      "eval_runtime": 36.8832,
      "eval_samples_per_second": 3.823,
      "eval_steps_per_second": 1.925,
      "step": 1400
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.529394268118168e-05,
      "loss": 1.2402,
      "step": 1401
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.527345360188596e-05,
      "loss": 1.3763,
      "step": 1402
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.525295272405227e-05,
      "loss": 1.5156,
      "step": 1403
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.523244005453795e-05,
      "loss": 1.4222,
      "step": 1404
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.521191560020417e-05,
      "loss": 1.3776,
      "step": 1405
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.519137936791617e-05,
      "loss": 1.348,
      "step": 1406
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.517083136454303e-05,
      "loss": 1.2601,
      "step": 1407
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.515027159695781e-05,
      "loss": 1.3362,
      "step": 1408
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.51297000720375e-05,
      "loss": 1.6127,
      "step": 1409
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.510911679666304e-05,
      "loss": 1.2267,
      "step": 1410
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.508852177771927e-05,
      "loss": 1.4029,
      "step": 1411
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.506791502209496e-05,
      "loss": 1.5504,
      "step": 1412
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.504729653668284e-05,
      "loss": 1.4521,
      "step": 1413
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.502666632837954e-05,
      "loss": 1.1612,
      "step": 1414
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.50060244040856e-05,
      "loss": 1.2095,
      "step": 1415
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.498537077070548e-05,
      "loss": 1.262,
      "step": 1416
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.496470543514759e-05,
      "loss": 1.6734,
      "step": 1417
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.494402840432423e-05,
      "loss": 1.4934,
      "step": 1418
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.492333968515159e-05,
      "loss": 1.3871,
      "step": 1419
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.490263928454983e-05,
      "loss": 1.4878,
      "step": 1420
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.488192720944294e-05,
      "loss": 1.1228,
      "step": 1421
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.486120346675888e-05,
      "loss": 1.2864,
      "step": 1422
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.484046806342948e-05,
      "loss": 1.3066,
      "step": 1423
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.481972100639049e-05,
      "loss": 1.1283,
      "step": 1424
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.479896230258153e-05,
      "loss": 1.4781,
      "step": 1425
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.477819195894614e-05,
      "loss": 1.5192,
      "step": 1426
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.475740998243174e-05,
      "loss": 1.4034,
      "step": 1427
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.473661637998966e-05,
      "loss": 1.3184,
      "step": 1428
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.471581115857509e-05,
      "loss": 1.4036,
      "step": 1429
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.469499432514712e-05,
      "loss": 1.4177,
      "step": 1430
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.467416588666875e-05,
      "loss": 1.3309,
      "step": 1431
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.465332585010682e-05,
      "loss": 1.3865,
      "step": 1432
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.463247422243206e-05,
      "loss": 1.4129,
      "step": 1433
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.46116110106191e-05,
      "loss": 1.4222,
      "step": 1434
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.45907362216464e-05,
      "loss": 1.215,
      "step": 1435
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.456984986249636e-05,
      "loss": 1.3565,
      "step": 1436
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.454895194015518e-05,
      "loss": 1.1328,
      "step": 1437
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.452804246161299e-05,
      "loss": 1.4693,
      "step": 1438
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.450712143386372e-05,
      "loss": 1.3119,
      "step": 1439
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.448618886390522e-05,
      "loss": 1.5348,
      "step": 1440
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.446524475873915e-05,
      "loss": 1.4128,
      "step": 1441
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.444428912537111e-05,
      "loss": 1.4029,
      "step": 1442
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.442332197081046e-05,
      "loss": 1.2715,
      "step": 1443
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.440234330207047e-05,
      "loss": 1.5166,
      "step": 1444
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.438135312616823e-05,
      "loss": 1.2437,
      "step": 1445
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.436035145012473e-05,
      "loss": 1.4107,
      "step": 1446
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.433933828096473e-05,
      "loss": 1.3158,
      "step": 1447
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.431831362571691e-05,
      "loss": 1.2641,
      "step": 1448
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.429727749141377e-05,
      "loss": 1.1259,
      "step": 1449
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.427622988509161e-05,
      "loss": 0.8333,
      "step": 1450
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.42551708137906e-05,
      "loss": 1.3185,
      "step": 1451
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.423410028455474e-05,
      "loss": 1.4789,
      "step": 1452
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.421301830443189e-05,
      "loss": 1.5746,
      "step": 1453
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.419192488047369e-05,
      "loss": 1.6012,
      "step": 1454
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.417082001973564e-05,
      "loss": 1.4087,
      "step": 1455
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.414970372927704e-05,
      "loss": 0.9521,
      "step": 1456
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.412857601616107e-05,
      "loss": 1.3099,
      "step": 1457
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.410743688745464e-05,
      "loss": 1.3754,
      "step": 1458
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.408628635022857e-05,
      "loss": 1.2651,
      "step": 1459
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.406512441155745e-05,
      "loss": 1.6379,
      "step": 1460
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.404395107851967e-05,
      "loss": 1.3654,
      "step": 1461
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.402276635819746e-05,
      "loss": 1.5104,
      "step": 1462
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.400157025767685e-05,
      "loss": 1.349,
      "step": 1463
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.398036278404767e-05,
      "loss": 1.3065,
      "step": 1464
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.395914394440358e-05,
      "loss": 1.205,
      "step": 1465
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.393791374584202e-05,
      "loss": 1.3844,
      "step": 1466
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.391667219546418e-05,
      "loss": 1.3526,
      "step": 1467
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.389541930037516e-05,
      "loss": 1.4392,
      "step": 1468
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.387415506768377e-05,
      "loss": 1.4547,
      "step": 1469
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.385287950450263e-05,
      "loss": 1.2119,
      "step": 1470
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.383159261794817e-05,
      "loss": 1.4494,
      "step": 1471
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.38102944151406e-05,
      "loss": 1.3156,
      "step": 1472
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.378898490320387e-05,
      "loss": 1.5232,
      "step": 1473
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.376766408926576e-05,
      "loss": 1.4241,
      "step": 1474
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.374633198045785e-05,
      "loss": 1.5593,
      "step": 1475
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.372498858391544e-05,
      "loss": 1.1971,
      "step": 1476
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.370363390677765e-05,
      "loss": 1.1821,
      "step": 1477
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.368226795618737e-05,
      "loss": 1.2036,
      "step": 1478
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.366089073929121e-05,
      "loss": 1.4014,
      "step": 1479
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.363950226323963e-05,
      "loss": 1.1974,
      "step": 1480
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.361810253518679e-05,
      "loss": 1.4443,
      "step": 1481
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.359669156229062e-05,
      "loss": 1.3722,
      "step": 1482
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.357526935171288e-05,
      "loss": 1.2627,
      "step": 1483
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.355383591061898e-05,
      "loss": 1.5364,
      "step": 1484
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.353239124617818e-05,
      "loss": 1.3108,
      "step": 1485
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.351093536556344e-05,
      "loss": 1.3319,
      "step": 1486
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.34894682759515e-05,
      "loss": 1.365,
      "step": 1487
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.346798998452282e-05,
      "loss": 1.3239,
      "step": 1488
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.344650049846166e-05,
      "loss": 1.3227,
      "step": 1489
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.342499982495595e-05,
      "loss": 1.5481,
      "step": 1490
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.340348797119743e-05,
      "loss": 1.3427,
      "step": 1491
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.338196494438153e-05,
      "loss": 1.5732,
      "step": 1492
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.336043075170746e-05,
      "loss": 1.4169,
      "step": 1493
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.333888540037813e-05,
      "loss": 1.2777,
      "step": 1494
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.33173288976002e-05,
      "loss": 1.168,
      "step": 1495
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.329576125058406e-05,
      "loss": 1.3843,
      "step": 1496
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.327418246654382e-05,
      "loss": 1.4524,
      "step": 1497
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.325259255269731e-05,
      "loss": 1.4855,
      "step": 1498
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.32309915162661e-05,
      "loss": 1.5154,
      "step": 1499
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.320937936447548e-05,
      "loss": 1.1163,
      "step": 1500
    },
    {
      "epoch": 0.82,
      "eval_loss": 1.3565027713775635,
      "eval_runtime": 36.7572,
      "eval_samples_per_second": 3.836,
      "eval_steps_per_second": 1.932,
      "step": 1500
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.318775610455444e-05,
      "loss": 1.3007,
      "step": 1501
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.316612174373571e-05,
      "loss": 1.1855,
      "step": 1502
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.314447628925568e-05,
      "loss": 1.2163,
      "step": 1503
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.312281974835452e-05,
      "loss": 1.3295,
      "step": 1504
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.310115212827608e-05,
      "loss": 1.1217,
      "step": 1505
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.307947343626791e-05,
      "loss": 1.3889,
      "step": 1506
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.305778367958126e-05,
      "loss": 1.4899,
      "step": 1507
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.303608286547109e-05,
      "loss": 1.2408,
      "step": 1508
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.301437100119605e-05,
      "loss": 1.3762,
      "step": 1509
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.29926480940185e-05,
      "loss": 1.1186,
      "step": 1510
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.297091415120448e-05,
      "loss": 1.3886,
      "step": 1511
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.294916918002376e-05,
      "loss": 1.1553,
      "step": 1512
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.292741318774971e-05,
      "loss": 1.4741,
      "step": 1513
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.290564618165948e-05,
      "loss": 1.5721,
      "step": 1514
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.288386816903387e-05,
      "loss": 1.4381,
      "step": 1515
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.286207915715733e-05,
      "loss": 1.4033,
      "step": 1516
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.284027915331806e-05,
      "loss": 1.5207,
      "step": 1517
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.281846816480784e-05,
      "loss": 1.5913,
      "step": 1518
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.279664619892222e-05,
      "loss": 1.5134,
      "step": 1519
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.277481326296038e-05,
      "loss": 1.6983,
      "step": 1520
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.275296936422514e-05,
      "loss": 1.4251,
      "step": 1521
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.273111451002305e-05,
      "loss": 1.4919,
      "step": 1522
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.270924870766426e-05,
      "loss": 1.4349,
      "step": 1523
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.268737196446264e-05,
      "loss": 1.3514,
      "step": 1524
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.266548428773566e-05,
      "loss": 1.3787,
      "step": 1525
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.26435856848045e-05,
      "loss": 1.4563,
      "step": 1526
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.262167616299398e-05,
      "loss": 1.4732,
      "step": 1527
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.259975572963257e-05,
      "loss": 1.3999,
      "step": 1528
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.257782439205238e-05,
      "loss": 0.9667,
      "step": 1529
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.255588215758916e-05,
      "loss": 1.0957,
      "step": 1530
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.253392903358233e-05,
      "loss": 1.2929,
      "step": 1531
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.251196502737494e-05,
      "loss": 1.424,
      "step": 1532
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.248999014631368e-05,
      "loss": 1.3442,
      "step": 1533
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.246800439774889e-05,
      "loss": 1.2627,
      "step": 1534
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.24460077890345e-05,
      "loss": 1.23,
      "step": 1535
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.242400032752813e-05,
      "loss": 1.3209,
      "step": 1536
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.240198202059102e-05,
      "loss": 1.3161,
      "step": 1537
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.237995287558802e-05,
      "loss": 1.5957,
      "step": 1538
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.235791289988757e-05,
      "loss": 1.6315,
      "step": 1539
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.233586210086181e-05,
      "loss": 1.0535,
      "step": 1540
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.231380048588646e-05,
      "loss": 1.4862,
      "step": 1541
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.229172806234085e-05,
      "loss": 1.3469,
      "step": 1542
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.226964483760792e-05,
      "loss": 1.3342,
      "step": 1543
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.224755081907427e-05,
      "loss": 1.4645,
      "step": 1544
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.222544601413005e-05,
      "loss": 1.3196,
      "step": 1545
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.220333043016908e-05,
      "loss": 1.5113,
      "step": 1546
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.218120407458871e-05,
      "loss": 1.461,
      "step": 1547
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.215906695478996e-05,
      "loss": 1.3335,
      "step": 1548
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.213691907817742e-05,
      "loss": 1.3128,
      "step": 1549
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.21147604521593e-05,
      "loss": 1.458,
      "step": 1550
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.209259108414737e-05,
      "loss": 1.1729,
      "step": 1551
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.2070410981557e-05,
      "loss": 1.5137,
      "step": 1552
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.204822015180718e-05,
      "loss": 1.3117,
      "step": 1553
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.20260186023205e-05,
      "loss": 1.4225,
      "step": 1554
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.200380634052303e-05,
      "loss": 1.5778,
      "step": 1555
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.198158337384457e-05,
      "loss": 1.2495,
      "step": 1556
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.19593497097184e-05,
      "loss": 1.2573,
      "step": 1557
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.19371053555814e-05,
      "loss": 1.4969,
      "step": 1558
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.191485031887404e-05,
      "loss": 1.2208,
      "step": 1559
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.189258460704038e-05,
      "loss": 1.5786,
      "step": 1560
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.187030822752801e-05,
      "loss": 1.3743,
      "step": 1561
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.184802118778812e-05,
      "loss": 1.4122,
      "step": 1562
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.182572349527542e-05,
      "loss": 1.2908,
      "step": 1563
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.180341515744824e-05,
      "loss": 1.2936,
      "step": 1564
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.178109618176844e-05,
      "loss": 1.1441,
      "step": 1565
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.175876657570143e-05,
      "loss": 1.571,
      "step": 1566
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.173642634671621e-05,
      "loss": 1.228,
      "step": 1567
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.171407550228532e-05,
      "loss": 1.5654,
      "step": 1568
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.16917140498848e-05,
      "loss": 1.6108,
      "step": 1569
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.166934199699433e-05,
      "loss": 1.3438,
      "step": 1570
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.164695935109707e-05,
      "loss": 1.428,
      "step": 1571
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.162456611967973e-05,
      "loss": 1.4085,
      "step": 1572
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.160216231023258e-05,
      "loss": 1.4559,
      "step": 1573
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.157974793024943e-05,
      "loss": 1.4101,
      "step": 1574
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.155732298722759e-05,
      "loss": 1.4942,
      "step": 1575
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.153488748866796e-05,
      "loss": 1.2717,
      "step": 1576
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.15124414420749e-05,
      "loss": 1.5733,
      "step": 1577
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.148998485495638e-05,
      "loss": 1.5379,
      "step": 1578
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.146751773482383e-05,
      "loss": 1.2596,
      "step": 1579
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.144504008919222e-05,
      "loss": 1.1619,
      "step": 1580
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.142255192558008e-05,
      "loss": 1.3947,
      "step": 1581
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.140005325150939e-05,
      "loss": 1.2783,
      "step": 1582
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.137754407450568e-05,
      "loss": 1.4894,
      "step": 1583
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.135502440209804e-05,
      "loss": 1.2947,
      "step": 1584
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.133249424181897e-05,
      "loss": 1.2713,
      "step": 1585
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.130995360120454e-05,
      "loss": 1.3917,
      "step": 1586
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.128740248779436e-05,
      "loss": 1.3993,
      "step": 1587
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.126484090913147e-05,
      "loss": 1.3165,
      "step": 1588
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.124226887276243e-05,
      "loss": 1.3462,
      "step": 1589
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.121968638623733e-05,
      "loss": 1.3473,
      "step": 1590
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.119709345710973e-05,
      "loss": 2.0555,
      "step": 1591
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.117449009293668e-05,
      "loss": 1.354,
      "step": 1592
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.115187630127874e-05,
      "loss": 1.2401,
      "step": 1593
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.112925208969995e-05,
      "loss": 1.1532,
      "step": 1594
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.110661746576783e-05,
      "loss": 1.3554,
      "step": 1595
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.108397243705335e-05,
      "loss": 1.5368,
      "step": 1596
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.106131701113104e-05,
      "loss": 1.2612,
      "step": 1597
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.103865119557884e-05,
      "loss": 1.294,
      "step": 1598
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.10159749979782e-05,
      "loss": 1.4354,
      "step": 1599
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.0993288425914e-05,
      "loss": 1.4656,
      "step": 1600
    },
    {
      "epoch": 0.88,
      "eval_loss": 1.349539875984192,
      "eval_runtime": 36.7546,
      "eval_samples_per_second": 3.836,
      "eval_steps_per_second": 1.932,
      "step": 1600
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.097059148697468e-05,
      "loss": 1.3552,
      "step": 1601
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.094788418875201e-05,
      "loss": 1.4237,
      "step": 1602
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.092516653884137e-05,
      "loss": 1.4067,
      "step": 1603
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.090243854484149e-05,
      "loss": 1.4237,
      "step": 1604
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.087970021435465e-05,
      "loss": 1.5264,
      "step": 1605
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.08569515549865e-05,
      "loss": 1.3481,
      "step": 1606
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.083419257434622e-05,
      "loss": 1.1441,
      "step": 1607
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.081142328004637e-05,
      "loss": 1.3389,
      "step": 1608
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.078864367970304e-05,
      "loss": 1.335,
      "step": 1609
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.076585378093571e-05,
      "loss": 1.4537,
      "step": 1610
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.074305359136732e-05,
      "loss": 1.4492,
      "step": 1611
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.072024311862426e-05,
      "loss": 1.1223,
      "step": 1612
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.069742237033635e-05,
      "loss": 1.2358,
      "step": 1613
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.067459135413686e-05,
      "loss": 1.219,
      "step": 1614
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.065175007766248e-05,
      "loss": 1.5461,
      "step": 1615
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.062889854855333e-05,
      "loss": 1.3462,
      "step": 1616
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.060603677445297e-05,
      "loss": 1.0643,
      "step": 1617
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.058316476300838e-05,
      "loss": 1.1816,
      "step": 1618
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.056028252186998e-05,
      "loss": 1.3463,
      "step": 1619
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.053739005869157e-05,
      "loss": 1.5719,
      "step": 1620
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.051448738113046e-05,
      "loss": 1.5353,
      "step": 1621
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.049157449684723e-05,
      "loss": 1.2569,
      "step": 1622
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.046865141350604e-05,
      "loss": 1.3229,
      "step": 1623
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.044571813877431e-05,
      "loss": 1.4288,
      "step": 1624
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.042277468032299e-05,
      "loss": 1.1178,
      "step": 1625
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.039982104582638e-05,
      "loss": 1.6745,
      "step": 1626
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.037685724296217e-05,
      "loss": 1.8162,
      "step": 1627
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.035388327941147e-05,
      "loss": 1.4811,
      "step": 1628
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.03308991628588e-05,
      "loss": 1.2445,
      "step": 1629
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.030790490099207e-05,
      "loss": 1.4695,
      "step": 1630
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.028490050150257e-05,
      "loss": 1.261,
      "step": 1631
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.0261885972085e-05,
      "loss": 1.265,
      "step": 1632
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.023886132043742e-05,
      "loss": 1.2588,
      "step": 1633
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.021582655426129e-05,
      "loss": 1.3223,
      "step": 1634
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.01927816812615e-05,
      "loss": 1.1426,
      "step": 1635
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.016972670914624e-05,
      "loss": 1.3395,
      "step": 1636
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.014666164562712e-05,
      "loss": 1.3517,
      "step": 1637
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.012358649841912e-05,
      "loss": 1.3321,
      "step": 1638
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.01005012752406e-05,
      "loss": 1.2848,
      "step": 1639
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.00774059838133e-05,
      "loss": 1.3877,
      "step": 1640
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.005430063186228e-05,
      "loss": 1.5788,
      "step": 1641
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.003118522711601e-05,
      "loss": 1.2315,
      "step": 1642
    },
    {
      "epoch": 0.9,
      "learning_rate": 8.000805977730633e-05,
      "loss": 1.3056,
      "step": 1643
    },
    {
      "epoch": 0.9,
      "learning_rate": 7.998492429016837e-05,
      "loss": 1.4762,
      "step": 1644
    },
    {
      "epoch": 0.9,
      "learning_rate": 7.996177877344072e-05,
      "loss": 1.4585,
      "step": 1645
    },
    {
      "epoch": 0.9,
      "learning_rate": 7.993862323486525e-05,
      "loss": 1.5119,
      "step": 1646
    },
    {
      "epoch": 0.9,
      "learning_rate": 7.99154576821872e-05,
      "loss": 1.5201,
      "step": 1647
    },
    {
      "epoch": 0.9,
      "learning_rate": 7.989228212315516e-05,
      "loss": 1.2479,
      "step": 1648
    },
    {
      "epoch": 0.9,
      "learning_rate": 7.98690965655211e-05,
      "loss": 1.1894,
      "step": 1649
    },
    {
      "epoch": 0.9,
      "learning_rate": 7.984590101704025e-05,
      "loss": 1.4378,
      "step": 1650
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.982269548547129e-05,
      "loss": 1.1332,
      "step": 1651
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.979947997857617e-05,
      "loss": 1.3217,
      "step": 1652
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.97762545041202e-05,
      "loss": 1.477,
      "step": 1653
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.975301906987198e-05,
      "loss": 1.2517,
      "step": 1654
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.97297736836035e-05,
      "loss": 1.22,
      "step": 1655
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.970651835309009e-05,
      "loss": 1.4012,
      "step": 1656
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.968325308611031e-05,
      "loss": 1.3052,
      "step": 1657
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.965997789044615e-05,
      "loss": 1.2546,
      "step": 1658
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.963669277388288e-05,
      "loss": 1.5679,
      "step": 1659
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.961339774420907e-05,
      "loss": 1.3919,
      "step": 1660
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.959009280921661e-05,
      "loss": 1.469,
      "step": 1661
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.956677797670074e-05,
      "loss": 1.2912,
      "step": 1662
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.954345325445996e-05,
      "loss": 1.3299,
      "step": 1663
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.952011865029614e-05,
      "loss": 1.5038,
      "step": 1664
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.94967741720144e-05,
      "loss": 1.1415,
      "step": 1665
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.947341982742315e-05,
      "loss": 1.5228,
      "step": 1666
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.945005562433419e-05,
      "loss": 1.3466,
      "step": 1667
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.942668157056254e-05,
      "loss": 1.7454,
      "step": 1668
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.940329767392651e-05,
      "loss": 1.429,
      "step": 1669
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.937990394224774e-05,
      "loss": 1.1641,
      "step": 1670
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.935650038335117e-05,
      "loss": 1.283,
      "step": 1671
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.933308700506497e-05,
      "loss": 1.1952,
      "step": 1672
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.930966381522064e-05,
      "loss": 1.4498,
      "step": 1673
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.928623082165297e-05,
      "loss": 1.6252,
      "step": 1674
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.926278803219997e-05,
      "loss": 1.4741,
      "step": 1675
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.923933545470301e-05,
      "loss": 1.2081,
      "step": 1676
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.921587309700665e-05,
      "loss": 1.5404,
      "step": 1677
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.919240096695877e-05,
      "loss": 1.3352,
      "step": 1678
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.916891907241053e-05,
      "loss": 1.3773,
      "step": 1679
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.914542742121633e-05,
      "loss": 1.5853,
      "step": 1680
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.912192602123383e-05,
      "loss": 1.5302,
      "step": 1681
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.909841488032395e-05,
      "loss": 1.1502,
      "step": 1682
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.90748940063509e-05,
      "loss": 1.4492,
      "step": 1683
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.905136340718211e-05,
      "loss": 1.3651,
      "step": 1684
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.90278230906883e-05,
      "loss": 1.3831,
      "step": 1685
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.900427306474339e-05,
      "loss": 1.5398,
      "step": 1686
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.898071333722458e-05,
      "loss": 1.4065,
      "step": 1687
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.895714391601232e-05,
      "loss": 1.319,
      "step": 1688
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.89335648089903e-05,
      "loss": 1.2097,
      "step": 1689
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.890997602404541e-05,
      "loss": 1.3745,
      "step": 1690
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.888637756906785e-05,
      "loss": 1.5959,
      "step": 1691
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.886276945195099e-05,
      "loss": 1.2685,
      "step": 1692
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.883915168059144e-05,
      "loss": 1.1695,
      "step": 1693
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.881552426288909e-05,
      "loss": 1.2246,
      "step": 1694
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.879188720674702e-05,
      "loss": 1.4211,
      "step": 1695
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.876824052007149e-05,
      "loss": 1.2886,
      "step": 1696
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.874458421077209e-05,
      "loss": 1.4347,
      "step": 1697
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.872091828676151e-05,
      "loss": 1.2476,
      "step": 1698
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.869724275595575e-05,
      "loss": 1.2151,
      "step": 1699
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.867355762627397e-05,
      "loss": 1.2698,
      "step": 1700
    },
    {
      "epoch": 0.93,
      "eval_loss": 1.3484092950820923,
      "eval_runtime": 36.6615,
      "eval_samples_per_second": 3.846,
      "eval_steps_per_second": 1.937,
      "step": 1700
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.864986290563856e-05,
      "loss": 1.1259,
      "step": 1701
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.86261586019751e-05,
      "loss": 1.5961,
      "step": 1702
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.860244472321242e-05,
      "loss": 1.0896,
      "step": 1703
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.857872127728248e-05,
      "loss": 1.3534,
      "step": 1704
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.85549882721205e-05,
      "loss": 1.4195,
      "step": 1705
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.853124571566491e-05,
      "loss": 1.4944,
      "step": 1706
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.850749361585727e-05,
      "loss": 1.4193,
      "step": 1707
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.848373198064236e-05,
      "loss": 1.421,
      "step": 1708
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.845996081796817e-05,
      "loss": 1.45,
      "step": 1709
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.843618013578586e-05,
      "loss": 1.5272,
      "step": 1710
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.84123899420498e-05,
      "loss": 1.2384,
      "step": 1711
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.838859024471748e-05,
      "loss": 1.1173,
      "step": 1712
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.836478105174961e-05,
      "loss": 1.6394,
      "step": 1713
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.83409623711101e-05,
      "loss": 1.2715,
      "step": 1714
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.831713421076601e-05,
      "loss": 1.2305,
      "step": 1715
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.829329657868753e-05,
      "loss": 1.3169,
      "step": 1716
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.82694494828481e-05,
      "loss": 1.3656,
      "step": 1717
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.824559293122423e-05,
      "loss": 1.4981,
      "step": 1718
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.822172693179568e-05,
      "loss": 1.5595,
      "step": 1719
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.819785149254532e-05,
      "loss": 1.6228,
      "step": 1720
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.817396662145921e-05,
      "loss": 1.2149,
      "step": 1721
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.815007232652654e-05,
      "loss": 1.5753,
      "step": 1722
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.812616861573966e-05,
      "loss": 1.1868,
      "step": 1723
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.810225549709404e-05,
      "loss": 1.2207,
      "step": 1724
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.807833297858838e-05,
      "loss": 1.3239,
      "step": 1725
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.805440106822442e-05,
      "loss": 1.3502,
      "step": 1726
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.80304597740071e-05,
      "loss": 1.6119,
      "step": 1727
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.800650910394449e-05,
      "loss": 1.165,
      "step": 1728
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.798254906604782e-05,
      "loss": 1.4029,
      "step": 1729
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.795857966833142e-05,
      "loss": 1.4175,
      "step": 1730
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.793460091881273e-05,
      "loss": 1.3314,
      "step": 1731
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.791061282551237e-05,
      "loss": 1.3254,
      "step": 1732
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.788661539645407e-05,
      "loss": 1.3946,
      "step": 1733
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.786260863966468e-05,
      "loss": 1.4439,
      "step": 1734
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.783859256317416e-05,
      "loss": 1.659,
      "step": 1735
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.781456717501557e-05,
      "loss": 1.2757,
      "step": 1736
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.779053248322511e-05,
      "loss": 1.4224,
      "step": 1737
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.776648849584214e-05,
      "loss": 1.4263,
      "step": 1738
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.774243522090902e-05,
      "loss": 1.3555,
      "step": 1739
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.771837266647131e-05,
      "loss": 1.2742,
      "step": 1740
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.769430084057763e-05,
      "loss": 1.4396,
      "step": 1741
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.767021975127972e-05,
      "loss": 1.1526,
      "step": 1742
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.764612940663239e-05,
      "loss": 1.352,
      "step": 1743
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.762202981469357e-05,
      "loss": 1.3641,
      "step": 1744
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.75979209835243e-05,
      "loss": 1.4456,
      "step": 1745
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.757380292118865e-05,
      "loss": 1.5297,
      "step": 1746
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.754967563575386e-05,
      "loss": 1.3547,
      "step": 1747
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.752553913529018e-05,
      "loss": 1.3191,
      "step": 1748
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.750139342787099e-05,
      "loss": 1.414,
      "step": 1749
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.747723852157271e-05,
      "loss": 1.1959,
      "step": 1750
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.745307442447489e-05,
      "loss": 1.3144,
      "step": 1751
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.74289011446601e-05,
      "loss": 1.3419,
      "step": 1752
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.740471869021404e-05,
      "loss": 1.2754,
      "step": 1753
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.73805270692254e-05,
      "loss": 1.3597,
      "step": 1754
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.735632628978601e-05,
      "loss": 1.56,
      "step": 1755
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.733211635999072e-05,
      "loss": 1.2595,
      "step": 1756
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.730789728793746e-05,
      "loss": 1.1341,
      "step": 1757
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.728366908172722e-05,
      "loss": 1.3435,
      "step": 1758
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.725943174946403e-05,
      "loss": 1.6414,
      "step": 1759
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.7235185299255e-05,
      "loss": 1.4056,
      "step": 1760
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.721092973921022e-05,
      "loss": 1.3158,
      "step": 1761
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.718666507744294e-05,
      "loss": 1.4503,
      "step": 1762
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.716239132206934e-05,
      "loss": 1.2474,
      "step": 1763
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.713810848120872e-05,
      "loss": 1.3043,
      "step": 1764
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.71138165629834e-05,
      "loss": 1.4612,
      "step": 1765
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.708951557551873e-05,
      "loss": 1.3134,
      "step": 1766
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.706520552694305e-05,
      "loss": 1.5413,
      "step": 1767
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.704088642538782e-05,
      "loss": 1.4014,
      "step": 1768
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.701655827898746e-05,
      "loss": 1.3453,
      "step": 1769
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.699222109587945e-05,
      "loss": 1.6987,
      "step": 1770
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.696787488420426e-05,
      "loss": 1.5512,
      "step": 1771
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.694351965210542e-05,
      "loss": 1.6868,
      "step": 1772
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.691915540772945e-05,
      "loss": 1.428,
      "step": 1773
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.689478215922588e-05,
      "loss": 1.4222,
      "step": 1774
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.687039991474729e-05,
      "loss": 1.2547,
      "step": 1775
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.68460086824492e-05,
      "loss": 1.3963,
      "step": 1776
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.682160847049021e-05,
      "loss": 1.5228,
      "step": 1777
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.679719928703188e-05,
      "loss": 1.3395,
      "step": 1778
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.67727811402388e-05,
      "loss": 1.482,
      "step": 1779
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.674835403827852e-05,
      "loss": 1.4609,
      "step": 1780
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.67239179893216e-05,
      "loss": 1.0545,
      "step": 1781
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.669947300154163e-05,
      "loss": 1.3301,
      "step": 1782
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.667501908311514e-05,
      "loss": 1.3976,
      "step": 1783
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.665055624222166e-05,
      "loss": 1.5219,
      "step": 1784
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.662608448704374e-05,
      "loss": 1.3393,
      "step": 1785
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.660160382576683e-05,
      "loss": 1.2625,
      "step": 1786
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.657711426657946e-05,
      "loss": 1.2448,
      "step": 1787
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.655261581767305e-05,
      "loss": 1.266,
      "step": 1788
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.652810848724208e-05,
      "loss": 1.522,
      "step": 1789
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.650359228348389e-05,
      "loss": 1.2638,
      "step": 1790
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.64790672145989e-05,
      "loss": 1.6299,
      "step": 1791
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.645453328879042e-05,
      "loss": 1.6544,
      "step": 1792
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.642999051426476e-05,
      "loss": 1.3719,
      "step": 1793
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.640543889923118e-05,
      "loss": 1.2812,
      "step": 1794
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.638087845190186e-05,
      "loss": 1.3421,
      "step": 1795
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.635630918049201e-05,
      "loss": 1.2548,
      "step": 1796
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.633173109321973e-05,
      "loss": 1.4544,
      "step": 1797
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.63071441983061e-05,
      "loss": 1.3392,
      "step": 1798
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.628254850397512e-05,
      "loss": 1.4693,
      "step": 1799
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.625794401845377e-05,
      "loss": 1.2019,
      "step": 1800
    },
    {
      "epoch": 0.99,
      "eval_loss": 1.3454393148422241,
      "eval_runtime": 36.8015,
      "eval_samples_per_second": 3.831,
      "eval_steps_per_second": 1.929,
      "step": 1800
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.623333074997193e-05,
      "loss": 1.3376,
      "step": 1801
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.620870870676242e-05,
      "loss": 1.3397,
      "step": 1802
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.618407789706105e-05,
      "loss": 1.4293,
      "step": 1803
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.61594383291065e-05,
      "loss": 1.2768,
      "step": 1804
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.613479001114042e-05,
      "loss": 1.5274,
      "step": 1805
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.611013295140736e-05,
      "loss": 1.4023,
      "step": 1806
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.608546715815481e-05,
      "loss": 1.2233,
      "step": 1807
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.606079263963317e-05,
      "loss": 1.4963,
      "step": 1808
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.603610940409578e-05,
      "loss": 1.3298,
      "step": 1809
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.601141745979884e-05,
      "loss": 1.1963,
      "step": 1810
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.598671681500153e-05,
      "loss": 1.327,
      "step": 1811
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.596200747796593e-05,
      "loss": 1.4308,
      "step": 1812
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.593728945695697e-05,
      "loss": 1.1957,
      "step": 1813
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.591256276024257e-05,
      "loss": 1.212,
      "step": 1814
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.588782739609345e-05,
      "loss": 1.3104,
      "step": 1815
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.586308337278336e-05,
      "loss": 1.2984,
      "step": 1816
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.583833069858881e-05,
      "loss": 1.427,
      "step": 1817
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.581356938178929e-05,
      "loss": 1.4612,
      "step": 1818
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.578879943066716e-05,
      "loss": 1.419,
      "step": 1819
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.576402085350764e-05,
      "loss": 1.2376,
      "step": 1820
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.573923365859891e-05,
      "loss": 1.1369,
      "step": 1821
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.571443785423196e-05,
      "loss": 1.3701,
      "step": 1822
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.568963344870067e-05,
      "loss": 1.3263,
      "step": 1823
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.566482045030179e-05,
      "loss": 1.1364,
      "step": 1824
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.563999886733501e-05,
      "loss": 1.2897,
      "step": 1825
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.561516870810281e-05,
      "loss": 1.4776,
      "step": 1826
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.55903299809106e-05,
      "loss": 1.1265,
      "step": 1827
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.556548269406663e-05,
      "loss": 1.1859,
      "step": 1828
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.554062685588199e-05,
      "loss": 1.4968,
      "step": 1829
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.551576247467067e-05,
      "loss": 1.133,
      "step": 1830
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.54908895587495e-05,
      "loss": 0.9842,
      "step": 1831
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.546600811643816e-05,
      "loss": 1.2121,
      "step": 1832
    },
    {
      "epoch": 1.0,
      "learning_rate": 7.544111815605918e-05,
      "loss": 1.3423,
      "step": 1833
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.541621968593798e-05,
      "loss": 1.0883,
      "step": 1834
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.539131271440274e-05,
      "loss": 1.2931,
      "step": 1835
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.536639724978458e-05,
      "loss": 1.1642,
      "step": 1836
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.534147330041741e-05,
      "loss": 1.1672,
      "step": 1837
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.531654087463796e-05,
      "loss": 1.0992,
      "step": 1838
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.529159998078586e-05,
      "loss": 1.2465,
      "step": 1839
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.52666506272035e-05,
      "loss": 1.1226,
      "step": 1840
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.524169282223615e-05,
      "loss": 1.1497,
      "step": 1841
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.521672657423188e-05,
      "loss": 1.398,
      "step": 1842
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.519175189154161e-05,
      "loss": 1.3212,
      "step": 1843
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.516676878251907e-05,
      "loss": 1.2198,
      "step": 1844
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.514177725552077e-05,
      "loss": 1.1385,
      "step": 1845
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.51167773189061e-05,
      "loss": 1.2369,
      "step": 1846
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.509176898103723e-05,
      "loss": 0.9645,
      "step": 1847
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.50667522502791e-05,
      "loss": 1.4042,
      "step": 1848
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.504172713499957e-05,
      "loss": 1.2384,
      "step": 1849
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.501669364356918e-05,
      "loss": 1.3425,
      "step": 1850
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.499165178436133e-05,
      "loss": 1.3239,
      "step": 1851
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.496660156575226e-05,
      "loss": 1.3759,
      "step": 1852
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.494154299612091e-05,
      "loss": 1.1857,
      "step": 1853
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.491647608384909e-05,
      "loss": 1.2442,
      "step": 1854
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.489140083732137e-05,
      "loss": 1.3326,
      "step": 1855
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.486631726492512e-05,
      "loss": 1.4563,
      "step": 1856
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.484122537505047e-05,
      "loss": 1.1068,
      "step": 1857
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.481612517609037e-05,
      "loss": 1.3343,
      "step": 1858
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.479101667644051e-05,
      "loss": 1.1544,
      "step": 1859
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.476589988449939e-05,
      "loss": 1.4227,
      "step": 1860
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.474077480866825e-05,
      "loss": 1.3957,
      "step": 1861
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.471564145735115e-05,
      "loss": 1.1668,
      "step": 1862
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.469049983895486e-05,
      "loss": 1.2857,
      "step": 1863
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.466534996188897e-05,
      "loss": 1.1046,
      "step": 1864
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.464019183456576e-05,
      "loss": 1.1524,
      "step": 1865
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.461502546540037e-05,
      "loss": 1.4496,
      "step": 1866
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.45898508628106e-05,
      "loss": 1.1105,
      "step": 1867
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.45646680352171e-05,
      "loss": 1.3147,
      "step": 1868
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.453947699104315e-05,
      "loss": 1.2427,
      "step": 1869
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.451427773871489e-05,
      "loss": 1.3249,
      "step": 1870
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.448907028666113e-05,
      "loss": 1.2477,
      "step": 1871
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.446385464331348e-05,
      "loss": 1.164,
      "step": 1872
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.443863081710627e-05,
      "loss": 1.3284,
      "step": 1873
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.441339881647653e-05,
      "loss": 1.2005,
      "step": 1874
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.438815864986407e-05,
      "loss": 1.4375,
      "step": 1875
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.436291032571142e-05,
      "loss": 1.4836,
      "step": 1876
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.433765385246383e-05,
      "loss": 1.257,
      "step": 1877
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.431238923856929e-05,
      "loss": 1.2912,
      "step": 1878
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.42871164924785e-05,
      "loss": 1.3032,
      "step": 1879
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.426183562264488e-05,
      "loss": 1.32,
      "step": 1880
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.423654663752455e-05,
      "loss": 1.2733,
      "step": 1881
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.421124954557641e-05,
      "loss": 1.2181,
      "step": 1882
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.4185944355262e-05,
      "loss": 1.2911,
      "step": 1883
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.41606310750456e-05,
      "loss": 1.2043,
      "step": 1884
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.413530971339418e-05,
      "loss": 1.4496,
      "step": 1885
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.410998027877744e-05,
      "loss": 1.3992,
      "step": 1886
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.408464277966778e-05,
      "loss": 1.2145,
      "step": 1887
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.405929722454026e-05,
      "loss": 1.2775,
      "step": 1888
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.403394362187266e-05,
      "loss": 1.2674,
      "step": 1889
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.400858198014544e-05,
      "loss": 1.2462,
      "step": 1890
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.398321230784178e-05,
      "loss": 1.2725,
      "step": 1891
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.395783461344754e-05,
      "loss": 1.2227,
      "step": 1892
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.39324489054512e-05,
      "loss": 1.1482,
      "step": 1893
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.3907055192344e-05,
      "loss": 1.1507,
      "step": 1894
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.388165348261982e-05,
      "loss": 1.3993,
      "step": 1895
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.385624378477521e-05,
      "loss": 1.2578,
      "step": 1896
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.383082610730942e-05,
      "loss": 1.3152,
      "step": 1897
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.380540045872433e-05,
      "loss": 1.2384,
      "step": 1898
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.377996684752453e-05,
      "loss": 1.4918,
      "step": 1899
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.375452528221722e-05,
      "loss": 1.3685,
      "step": 1900
    },
    {
      "epoch": 1.04,
      "eval_loss": 1.3476861715316772,
      "eval_runtime": 36.7044,
      "eval_samples_per_second": 3.841,
      "eval_steps_per_second": 1.934,
      "step": 1900
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.372907577131235e-05,
      "loss": 1.2849,
      "step": 1901
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.370361832332242e-05,
      "loss": 1.1056,
      "step": 1902
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.367815294676262e-05,
      "loss": 1.3851,
      "step": 1903
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.365267965015086e-05,
      "loss": 1.3417,
      "step": 1904
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.362719844200762e-05,
      "loss": 1.3328,
      "step": 1905
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.360170933085604e-05,
      "loss": 1.3508,
      "step": 1906
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.357621232522194e-05,
      "loss": 1.4106,
      "step": 1907
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.355070743363374e-05,
      "loss": 1.1451,
      "step": 1908
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.352519466462251e-05,
      "loss": 1.2966,
      "step": 1909
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.349967402672196e-05,
      "loss": 1.2441,
      "step": 1910
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.347414552846845e-05,
      "loss": 1.2872,
      "step": 1911
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.344860917840091e-05,
      "loss": 1.2004,
      "step": 1912
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.342306498506098e-05,
      "loss": 1.4764,
      "step": 1913
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.339751295699287e-05,
      "loss": 0.8894,
      "step": 1914
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.33719531027434e-05,
      "loss": 1.2003,
      "step": 1915
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.334638543086203e-05,
      "loss": 1.4263,
      "step": 1916
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.332080994990086e-05,
      "loss": 1.3327,
      "step": 1917
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.329522666841456e-05,
      "loss": 1.3522,
      "step": 1918
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.326963559496039e-05,
      "loss": 1.3232,
      "step": 1919
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.324403673809831e-05,
      "loss": 1.1301,
      "step": 1920
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.321843010639077e-05,
      "loss": 1.2476,
      "step": 1921
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.319281570840292e-05,
      "loss": 1.5011,
      "step": 1922
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.316719355270244e-05,
      "loss": 1.3001,
      "step": 1923
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.314156364785962e-05,
      "loss": 1.1566,
      "step": 1924
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.311592600244738e-05,
      "loss": 1.3169,
      "step": 1925
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.309028062504116e-05,
      "loss": 1.3109,
      "step": 1926
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.306462752421903e-05,
      "loss": 1.2392,
      "step": 1927
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.303896670856167e-05,
      "loss": 1.2304,
      "step": 1928
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.301329818665228e-05,
      "loss": 1.1993,
      "step": 1929
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.298762196707668e-05,
      "loss": 1.4047,
      "step": 1930
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.296193805842322e-05,
      "loss": 1.2411,
      "step": 1931
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.29362464692829e-05,
      "loss": 1.3097,
      "step": 1932
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.291054720824922e-05,
      "loss": 1.4239,
      "step": 1933
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.288484028391825e-05,
      "loss": 1.5578,
      "step": 1934
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.285912570488867e-05,
      "loss": 1.1908,
      "step": 1935
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.283340347976166e-05,
      "loss": 1.1053,
      "step": 1936
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.280767361714103e-05,
      "loss": 1.2089,
      "step": 1937
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.278193612563307e-05,
      "loss": 1.2974,
      "step": 1938
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.275619101384667e-05,
      "loss": 1.2921,
      "step": 1939
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.273043829039325e-05,
      "loss": 1.2124,
      "step": 1940
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.270467796388679e-05,
      "loss": 1.0153,
      "step": 1941
    },
    {
      "epoch": 1.06,
      "learning_rate": 7.26789100429438e-05,
      "loss": 1.4842,
      "step": 1942
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.265313453618332e-05,
      "loss": 1.462,
      "step": 1943
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.262735145222696e-05,
      "loss": 1.2904,
      "step": 1944
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.260156079969883e-05,
      "loss": 1.1925,
      "step": 1945
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.257576258722559e-05,
      "loss": 1.1499,
      "step": 1946
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.254995682343641e-05,
      "loss": 1.3543,
      "step": 1947
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.252414351696306e-05,
      "loss": 1.0513,
      "step": 1948
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.24983226764397e-05,
      "loss": 1.0865,
      "step": 1949
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.247249431050314e-05,
      "loss": 1.5022,
      "step": 1950
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.244665842779263e-05,
      "loss": 1.3777,
      "step": 1951
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.242081503694995e-05,
      "loss": 1.3257,
      "step": 1952
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.239496414661942e-05,
      "loss": 1.4793,
      "step": 1953
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.236910576544782e-05,
      "loss": 0.9378,
      "step": 1954
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.234323990208448e-05,
      "loss": 1.2822,
      "step": 1955
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.23173665651812e-05,
      "loss": 1.4368,
      "step": 1956
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.229148576339231e-05,
      "loss": 1.3087,
      "step": 1957
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.226559750537462e-05,
      "loss": 1.3217,
      "step": 1958
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.223970179978742e-05,
      "loss": 1.1486,
      "step": 1959
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.22137986552925e-05,
      "loss": 1.1555,
      "step": 1960
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.218788808055418e-05,
      "loss": 1.2651,
      "step": 1961
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.216197008423919e-05,
      "loss": 1.3005,
      "step": 1962
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.213604467501683e-05,
      "loss": 1.2213,
      "step": 1963
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.211011186155878e-05,
      "loss": 1.4142,
      "step": 1964
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.208417165253928e-05,
      "loss": 1.1717,
      "step": 1965
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.205822405663499e-05,
      "loss": 1.409,
      "step": 1966
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.203226908252508e-05,
      "loss": 1.2878,
      "step": 1967
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.200630673889117e-05,
      "loss": 1.3593,
      "step": 1968
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.198033703441734e-05,
      "loss": 1.2159,
      "step": 1969
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.195435997779014e-05,
      "loss": 1.3363,
      "step": 1970
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.192837557769859e-05,
      "loss": 1.101,
      "step": 1971
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.190238384283412e-05,
      "loss": 1.1188,
      "step": 1972
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.187638478189068e-05,
      "loss": 1.2245,
      "step": 1973
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.185037840356461e-05,
      "loss": 1.0693,
      "step": 1974
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.182436471655475e-05,
      "loss": 1.583,
      "step": 1975
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.179834372956236e-05,
      "loss": 1.2625,
      "step": 1976
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.177231545129111e-05,
      "loss": 1.276,
      "step": 1977
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.174627989044716e-05,
      "loss": 1.3221,
      "step": 1978
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.172023705573909e-05,
      "loss": 1.3294,
      "step": 1979
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.169418695587791e-05,
      "loss": 1.4082,
      "step": 1980
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.166812959957705e-05,
      "loss": 1.0566,
      "step": 1981
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.164206499555239e-05,
      "loss": 1.4655,
      "step": 1982
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.16159931525222e-05,
      "loss": 1.2589,
      "step": 1983
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.15899140792072e-05,
      "loss": 1.2295,
      "step": 1984
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.156382778433052e-05,
      "loss": 1.2818,
      "step": 1985
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.153773427661774e-05,
      "loss": 1.0575,
      "step": 1986
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.151163356479676e-05,
      "loss": 1.1081,
      "step": 1987
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.1485525657598e-05,
      "loss": 1.2758,
      "step": 1988
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.145941056375419e-05,
      "loss": 1.3975,
      "step": 1989
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.143328829200053e-05,
      "loss": 1.282,
      "step": 1990
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.140715885107462e-05,
      "loss": 1.2889,
      "step": 1991
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.13810222497164e-05,
      "loss": 1.2282,
      "step": 1992
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.135487849666827e-05,
      "loss": 0.9097,
      "step": 1993
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.132872760067497e-05,
      "loss": 1.2533,
      "step": 1994
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.130256957048369e-05,
      "loss": 1.3636,
      "step": 1995
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.127640441484394e-05,
      "loss": 1.1598,
      "step": 1996
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.125023214250766e-05,
      "loss": 1.2016,
      "step": 1997
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.122405276222914e-05,
      "loss": 1.5196,
      "step": 1998
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.119786628276507e-05,
      "loss": 1.0944,
      "step": 1999
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.117167271287453e-05,
      "loss": 1.2248,
      "step": 2000
    },
    {
      "epoch": 1.1,
      "eval_loss": 1.348763108253479,
      "eval_runtime": 36.7425,
      "eval_samples_per_second": 3.838,
      "eval_steps_per_second": 1.932,
      "step": 2000
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.114547206131889e-05,
      "loss": 1.0631,
      "step": 2001
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.1119264336862e-05,
      "loss": 1.3375,
      "step": 2002
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.109304954827e-05,
      "loss": 1.2464,
      "step": 2003
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.106682770431144e-05,
      "loss": 1.0455,
      "step": 2004
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.104059881375716e-05,
      "loss": 1.1286,
      "step": 2005
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.101436288538043e-05,
      "loss": 1.3214,
      "step": 2006
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.098811992795687e-05,
      "loss": 1.3458,
      "step": 2007
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.096186995026439e-05,
      "loss": 1.1727,
      "step": 2008
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.093561296108328e-05,
      "loss": 1.2775,
      "step": 2009
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.090934896919621e-05,
      "loss": 1.4289,
      "step": 2010
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.088307798338816e-05,
      "loss": 1.3783,
      "step": 2011
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.085680001244644e-05,
      "loss": 1.431,
      "step": 2012
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.083051506516071e-05,
      "loss": 1.3293,
      "step": 2013
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.080422315032298e-05,
      "loss": 1.1266,
      "step": 2014
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.077792427672756e-05,
      "loss": 1.3274,
      "step": 2015
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.07516184531711e-05,
      "loss": 1.0875,
      "step": 2016
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.072530568845258e-05,
      "loss": 1.3488,
      "step": 2017
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.069898599137328e-05,
      "loss": 1.1445,
      "step": 2018
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.067265937073683e-05,
      "loss": 1.4278,
      "step": 2019
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.064632583534917e-05,
      "loss": 1.394,
      "step": 2020
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.061998539401853e-05,
      "loss": 1.2558,
      "step": 2021
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.059363805555549e-05,
      "loss": 1.1547,
      "step": 2022
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.05672838287729e-05,
      "loss": 1.1065,
      "step": 2023
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.054092272248589e-05,
      "loss": 1.2016,
      "step": 2024
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.051455474551199e-05,
      "loss": 1.2926,
      "step": 2025
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.048817990667092e-05,
      "loss": 1.1139,
      "step": 2026
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.046179821478473e-05,
      "loss": 1.3604,
      "step": 2027
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.043540967867782e-05,
      "loss": 1.4203,
      "step": 2028
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.04090143071768e-05,
      "loss": 1.4314,
      "step": 2029
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.03826121091106e-05,
      "loss": 1.2345,
      "step": 2030
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.035620309331044e-05,
      "loss": 1.1463,
      "step": 2031
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.032978726860981e-05,
      "loss": 1.2674,
      "step": 2032
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.030336464384449e-05,
      "loss": 0.8144,
      "step": 2033
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.027693522785251e-05,
      "loss": 1.3494,
      "step": 2034
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.025049902947418e-05,
      "loss": 1.3821,
      "step": 2035
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.022405605755208e-05,
      "loss": 1.254,
      "step": 2036
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.01976063209311e-05,
      "loss": 1.4026,
      "step": 2037
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.017114982845833e-05,
      "loss": 1.2241,
      "step": 2038
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.014468658898312e-05,
      "loss": 1.3305,
      "step": 2039
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.011821661135713e-05,
      "loss": 1.3601,
      "step": 2040
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.009173990443422e-05,
      "loss": 1.3735,
      "step": 2041
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.006525647707054e-05,
      "loss": 1.1683,
      "step": 2042
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.003876633812446e-05,
      "loss": 1.3056,
      "step": 2043
    },
    {
      "epoch": 1.12,
      "learning_rate": 7.001226949645662e-05,
      "loss": 1.3352,
      "step": 2044
    },
    {
      "epoch": 1.12,
      "learning_rate": 6.998576596092988e-05,
      "loss": 1.2654,
      "step": 2045
    },
    {
      "epoch": 1.12,
      "learning_rate": 6.995925574040936e-05,
      "loss": 1.3322,
      "step": 2046
    },
    {
      "epoch": 1.12,
      "learning_rate": 6.99327388437624e-05,
      "loss": 1.445,
      "step": 2047
    },
    {
      "epoch": 1.12,
      "learning_rate": 6.990621527985856e-05,
      "loss": 1.1191,
      "step": 2048
    },
    {
      "epoch": 1.12,
      "learning_rate": 6.987968505756966e-05,
      "loss": 1.4963,
      "step": 2049
    },
    {
      "epoch": 1.12,
      "learning_rate": 6.985314818576972e-05,
      "loss": 1.319,
      "step": 2050
    },
    {
      "epoch": 1.12,
      "learning_rate": 6.982660467333502e-05,
      "loss": 1.1708,
      "step": 2051
    },
    {
      "epoch": 1.12,
      "learning_rate": 6.980005452914405e-05,
      "loss": 1.211,
      "step": 2052
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.977349776207746e-05,
      "loss": 1.3631,
      "step": 2053
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.97469343810182e-05,
      "loss": 1.0555,
      "step": 2054
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.972036439485139e-05,
      "loss": 1.3492,
      "step": 2055
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.969378781246436e-05,
      "loss": 1.3611,
      "step": 2056
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.966720464274666e-05,
      "loss": 1.2869,
      "step": 2057
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.964061489459002e-05,
      "loss": 1.205,
      "step": 2058
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.96140185768884e-05,
      "loss": 1.2401,
      "step": 2059
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.958741569853793e-05,
      "loss": 1.2978,
      "step": 2060
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.956080626843695e-05,
      "loss": 1.1099,
      "step": 2061
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.9534190295486e-05,
      "loss": 1.2999,
      "step": 2062
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.950756778858776e-05,
      "loss": 1.4197,
      "step": 2063
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.948093875664718e-05,
      "loss": 1.2007,
      "step": 2064
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.945430320857133e-05,
      "loss": 1.0811,
      "step": 2065
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.942766115326945e-05,
      "loss": 1.1694,
      "step": 2066
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.9401012599653e-05,
      "loss": 1.0374,
      "step": 2067
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.93743575566356e-05,
      "loss": 1.1147,
      "step": 2068
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.934769603313304e-05,
      "loss": 0.9588,
      "step": 2069
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.932102803806324e-05,
      "loss": 1.1738,
      "step": 2070
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.929435358034638e-05,
      "loss": 1.5169,
      "step": 2071
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.926767266890465e-05,
      "loss": 1.1267,
      "step": 2072
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.924098531266256e-05,
      "loss": 1.4917,
      "step": 2073
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.921429152054668e-05,
      "loss": 1.1995,
      "step": 2074
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.918759130148575e-05,
      "loss": 1.0696,
      "step": 2075
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.916088466441068e-05,
      "loss": 1.3199,
      "step": 2076
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.91341716182545e-05,
      "loss": 1.1223,
      "step": 2077
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.91074521719524e-05,
      "loss": 1.1932,
      "step": 2078
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.90807263344417e-05,
      "loss": 1.1025,
      "step": 2079
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.905399411466189e-05,
      "loss": 1.166,
      "step": 2080
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.902725552155456e-05,
      "loss": 1.2279,
      "step": 2081
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.900051056406341e-05,
      "loss": 1.2644,
      "step": 2082
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.897375925113435e-05,
      "loss": 1.4775,
      "step": 2083
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.894700159171534e-05,
      "loss": 1.3792,
      "step": 2084
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.89202375947565e-05,
      "loss": 1.2601,
      "step": 2085
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.889346726921004e-05,
      "loss": 1.568,
      "step": 2086
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.886669062403031e-05,
      "loss": 1.1658,
      "step": 2087
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.883990766817377e-05,
      "loss": 1.1258,
      "step": 2088
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.881311841059901e-05,
      "loss": 1.0731,
      "step": 2089
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.87863228602667e-05,
      "loss": 1.1642,
      "step": 2090
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.87595210261396e-05,
      "loss": 1.2542,
      "step": 2091
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.873271291718264e-05,
      "loss": 1.2609,
      "step": 2092
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.870589854236276e-05,
      "loss": 1.1566,
      "step": 2093
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.867907791064905e-05,
      "loss": 1.032,
      "step": 2094
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.865225103101271e-05,
      "loss": 1.3121,
      "step": 2095
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.862541791242698e-05,
      "loss": 1.0902,
      "step": 2096
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.859857856386721e-05,
      "loss": 1.2954,
      "step": 2097
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.857173299431083e-05,
      "loss": 1.4884,
      "step": 2098
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.854488121273736e-05,
      "loss": 1.2613,
      "step": 2099
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.851802322812839e-05,
      "loss": 1.2162,
      "step": 2100
    },
    {
      "epoch": 1.15,
      "eval_loss": 1.3479008674621582,
      "eval_runtime": 36.7155,
      "eval_samples_per_second": 3.84,
      "eval_steps_per_second": 1.934,
      "step": 2100
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.849115904946759e-05,
      "loss": 1.1745,
      "step": 2101
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.846428868574069e-05,
      "loss": 1.5115,
      "step": 2102
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.843741214593551e-05,
      "loss": 1.2555,
      "step": 2103
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.84105294390419e-05,
      "loss": 1.329,
      "step": 2104
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.838364057405184e-05,
      "loss": 1.0863,
      "step": 2105
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.835674555995928e-05,
      "loss": 1.0669,
      "step": 2106
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.83298444057603e-05,
      "loss": 1.028,
      "step": 2107
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.8302937120453e-05,
      "loss": 1.3061,
      "step": 2108
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.827602371303752e-05,
      "loss": 1.1962,
      "step": 2109
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.82491041925161e-05,
      "loss": 1.4089,
      "step": 2110
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.822217856789295e-05,
      "loss": 1.4123,
      "step": 2111
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.819524684817438e-05,
      "loss": 1.0844,
      "step": 2112
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.816830904236872e-05,
      "loss": 1.2808,
      "step": 2113
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.814136515948636e-05,
      "loss": 1.1474,
      "step": 2114
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.811441520853966e-05,
      "loss": 1.3019,
      "step": 2115
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.808745919854307e-05,
      "loss": 1.2967,
      "step": 2116
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.806049713851302e-05,
      "loss": 1.2057,
      "step": 2117
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.803352903746803e-05,
      "loss": 0.8483,
      "step": 2118
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.800655490442857e-05,
      "loss": 1.1229,
      "step": 2119
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.797957474841716e-05,
      "loss": 1.2843,
      "step": 2120
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.795258857845833e-05,
      "loss": 0.938,
      "step": 2121
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.792559640357863e-05,
      "loss": 1.2704,
      "step": 2122
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.789859823280661e-05,
      "loss": 1.1772,
      "step": 2123
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.787159407517285e-05,
      "loss": 1.1393,
      "step": 2124
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.784458393970987e-05,
      "loss": 1.4645,
      "step": 2125
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.781756783545225e-05,
      "loss": 1.3439,
      "step": 2126
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.779054577143653e-05,
      "loss": 1.0305,
      "step": 2127
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.776351775670129e-05,
      "loss": 1.4142,
      "step": 2128
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.773648380028704e-05,
      "loss": 1.4421,
      "step": 2129
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.770944391123633e-05,
      "loss": 1.3012,
      "step": 2130
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.768239809859365e-05,
      "loss": 0.9921,
      "step": 2131
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.76553463714055e-05,
      "loss": 1.0523,
      "step": 2132
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.762828873872036e-05,
      "loss": 1.3069,
      "step": 2133
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.760122520958867e-05,
      "loss": 1.2377,
      "step": 2134
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.757415579306284e-05,
      "loss": 1.0461,
      "step": 2135
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.754708049819728e-05,
      "loss": 1.3355,
      "step": 2136
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.751999933404834e-05,
      "loss": 1.4551,
      "step": 2137
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.749291230967431e-05,
      "loss": 1.2859,
      "step": 2138
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.746581943413549e-05,
      "loss": 1.9511,
      "step": 2139
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.743872071649411e-05,
      "loss": 1.4625,
      "step": 2140
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.741161616581437e-05,
      "loss": 1.424,
      "step": 2141
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.73845057911624e-05,
      "loss": 1.1822,
      "step": 2142
    },
    {
      "epoch": 1.17,
      "learning_rate": 6.735738960160629e-05,
      "loss": 1.3432,
      "step": 2143
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.733026760621607e-05,
      "loss": 1.2525,
      "step": 2144
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.73031398140637e-05,
      "loss": 1.3376,
      "step": 2145
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.727600623422314e-05,
      "loss": 1.4395,
      "step": 2146
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.72488668757702e-05,
      "loss": 1.2807,
      "step": 2147
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.722172174778267e-05,
      "loss": 1.1805,
      "step": 2148
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.719457085934026e-05,
      "loss": 1.1186,
      "step": 2149
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.716741421952464e-05,
      "loss": 1.2273,
      "step": 2150
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.714025183741932e-05,
      "loss": 1.1251,
      "step": 2151
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.711308372210983e-05,
      "loss": 1.3716,
      "step": 2152
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.708590988268355e-05,
      "loss": 1.332,
      "step": 2153
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.70587303282298e-05,
      "loss": 1.1294,
      "step": 2154
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.703154506783982e-05,
      "loss": 1.3738,
      "step": 2155
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.700435411060673e-05,
      "loss": 1.1072,
      "step": 2156
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.697715746562559e-05,
      "loss": 1.1073,
      "step": 2157
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.69499551419933e-05,
      "loss": 1.2855,
      "step": 2158
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.692274714880874e-05,
      "loss": 1.2748,
      "step": 2159
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.689553349517268e-05,
      "loss": 1.2712,
      "step": 2160
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.686831419018769e-05,
      "loss": 1.0735,
      "step": 2161
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.684108924295834e-05,
      "loss": 1.3622,
      "step": 2162
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.681385866259101e-05,
      "loss": 1.2658,
      "step": 2163
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.678662245819402e-05,
      "loss": 1.5756,
      "step": 2164
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.675938063887752e-05,
      "loss": 1.1528,
      "step": 2165
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.67321332137536e-05,
      "loss": 1.2596,
      "step": 2166
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.670488019193619e-05,
      "loss": 1.5093,
      "step": 2167
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.667762158254104e-05,
      "loss": 1.3243,
      "step": 2168
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.665035739468586e-05,
      "loss": 0.9888,
      "step": 2169
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.662308763749019e-05,
      "loss": 1.3073,
      "step": 2170
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.659581232007542e-05,
      "loss": 1.0161,
      "step": 2171
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.65685314515648e-05,
      "loss": 1.4992,
      "step": 2172
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.654124504108345e-05,
      "loss": 1.1679,
      "step": 2173
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.651395309775837e-05,
      "loss": 1.4385,
      "step": 2174
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.648665563071834e-05,
      "loss": 1.2912,
      "step": 2175
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.645935264909404e-05,
      "loss": 1.2783,
      "step": 2176
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.643204416201799e-05,
      "loss": 1.5898,
      "step": 2177
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.640473017862454e-05,
      "loss": 1.2471,
      "step": 2178
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.63774107080499e-05,
      "loss": 1.3461,
      "step": 2179
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.635008575943208e-05,
      "loss": 1.3001,
      "step": 2180
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.632275534191093e-05,
      "loss": 1.07,
      "step": 2181
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.629541946462817e-05,
      "loss": 1.3998,
      "step": 2182
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.626807813672727e-05,
      "loss": 1.2868,
      "step": 2183
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.624073136735363e-05,
      "loss": 1.1855,
      "step": 2184
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.621337916565436e-05,
      "loss": 1.1517,
      "step": 2185
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.618602154077844e-05,
      "loss": 1.2665,
      "step": 2186
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.61586585018767e-05,
      "loss": 1.0492,
      "step": 2187
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.613129005810169e-05,
      "loss": 1.3421,
      "step": 2188
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.610391621860786e-05,
      "loss": 1.13,
      "step": 2189
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.60765369925514e-05,
      "loss": 1.3044,
      "step": 2190
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.604915238909032e-05,
      "loss": 1.2455,
      "step": 2191
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.602176241738449e-05,
      "loss": 1.2429,
      "step": 2192
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.599436708659546e-05,
      "loss": 1.1952,
      "step": 2193
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.596696640588663e-05,
      "loss": 1.4524,
      "step": 2194
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.593956038442321e-05,
      "loss": 1.5541,
      "step": 2195
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.59121490313722e-05,
      "loss": 1.0816,
      "step": 2196
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.588473235590233e-05,
      "loss": 1.1696,
      "step": 2197
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.585731036718417e-05,
      "loss": 1.1999,
      "step": 2198
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.582988307439e-05,
      "loss": 1.4793,
      "step": 2199
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.580245048669395e-05,
      "loss": 1.0443,
      "step": 2200
    },
    {
      "epoch": 1.21,
      "eval_loss": 1.3491225242614746,
      "eval_runtime": 36.7193,
      "eval_samples_per_second": 3.84,
      "eval_steps_per_second": 1.934,
      "step": 2200
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.577501261327186e-05,
      "loss": 1.3931,
      "step": 2201
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.574756946330138e-05,
      "loss": 1.277,
      "step": 2202
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.57201210459619e-05,
      "loss": 1.2596,
      "step": 2203
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.569266737043458e-05,
      "loss": 1.4368,
      "step": 2204
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.566520844590233e-05,
      "loss": 1.4313,
      "step": 2205
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.563774428154983e-05,
      "loss": 0.9877,
      "step": 2206
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.561027488656352e-05,
      "loss": 1.2626,
      "step": 2207
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.558280027013154e-05,
      "loss": 1.3287,
      "step": 2208
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.555532044144385e-05,
      "loss": 1.5839,
      "step": 2209
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.552783540969212e-05,
      "loss": 1.2137,
      "step": 2210
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.550034518406972e-05,
      "loss": 1.1982,
      "step": 2211
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.547284977377182e-05,
      "loss": 1.2632,
      "step": 2212
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.544534918799529e-05,
      "loss": 1.218,
      "step": 2213
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.541784343593873e-05,
      "loss": 1.088,
      "step": 2214
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.539033252680249e-05,
      "loss": 1.4661,
      "step": 2215
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.536281646978862e-05,
      "loss": 1.153,
      "step": 2216
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.533529527410092e-05,
      "loss": 1.2439,
      "step": 2217
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.530776894894489e-05,
      "loss": 1.2323,
      "step": 2218
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.528023750352772e-05,
      "loss": 1.4843,
      "step": 2219
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.525270094705838e-05,
      "loss": 0.9544,
      "step": 2220
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.522515928874749e-05,
      "loss": 1.4689,
      "step": 2221
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.51976125378074e-05,
      "loss": 1.0806,
      "step": 2222
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.517006070345217e-05,
      "loss": 1.2547,
      "step": 2223
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.514250379489753e-05,
      "loss": 1.2603,
      "step": 2224
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.511494182136096e-05,
      "loss": 1.1177,
      "step": 2225
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.508737479206157e-05,
      "loss": 1.3423,
      "step": 2226
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.505980271622022e-05,
      "loss": 1.2891,
      "step": 2227
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.503222560305941e-05,
      "loss": 1.3647,
      "step": 2228
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.500464346180335e-05,
      "loss": 1.2465,
      "step": 2229
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.497705630167793e-05,
      "loss": 1.4714,
      "step": 2230
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.494946413191072e-05,
      "loss": 1.0874,
      "step": 2231
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.492186696173097e-05,
      "loss": 1.3325,
      "step": 2232
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.489426480036958e-05,
      "loss": 1.3075,
      "step": 2233
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.486665765705915e-05,
      "loss": 1.4355,
      "step": 2234
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.483904554103392e-05,
      "loss": 1.2911,
      "step": 2235
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.481142846152982e-05,
      "loss": 1.1282,
      "step": 2236
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.47838064277844e-05,
      "loss": 1.3707,
      "step": 2237
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.475617944903693e-05,
      "loss": 1.5228,
      "step": 2238
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.472854753452824e-05,
      "loss": 1.0322,
      "step": 2239
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.47009106935009e-05,
      "loss": 1.3598,
      "step": 2240
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.467326893519911e-05,
      "loss": 1.5192,
      "step": 2241
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.464562226886866e-05,
      "loss": 1.2289,
      "step": 2242
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.461797070375706e-05,
      "loss": 1.4236,
      "step": 2243
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.459031424911339e-05,
      "loss": 1.2129,
      "step": 2244
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.456265291418841e-05,
      "loss": 1.2668,
      "step": 2245
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.453498670823449e-05,
      "loss": 1.1587,
      "step": 2246
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.450731564050565e-05,
      "loss": 1.4367,
      "step": 2247
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.447963972025751e-05,
      "loss": 1.1714,
      "step": 2248
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.445195895674734e-05,
      "loss": 1.23,
      "step": 2249
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.4424273359234e-05,
      "loss": 1.3988,
      "step": 2250
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.439658293697798e-05,
      "loss": 1.1727,
      "step": 2251
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.436888769924142e-05,
      "loss": 1.4,
      "step": 2252
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.4341187655288e-05,
      "loss": 1.3259,
      "step": 2253
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.431348281438306e-05,
      "loss": 1.3139,
      "step": 2254
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.428577318579353e-05,
      "loss": 1.3244,
      "step": 2255
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.425805877878793e-05,
      "loss": 1.2935,
      "step": 2256
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.423033960263642e-05,
      "loss": 1.0602,
      "step": 2257
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.42026156666107e-05,
      "loss": 1.2755,
      "step": 2258
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.417488697998409e-05,
      "loss": 1.5266,
      "step": 2259
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.414715355203148e-05,
      "loss": 1.3378,
      "step": 2260
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.41194153920294e-05,
      "loss": 1.535,
      "step": 2261
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.409167250925589e-05,
      "loss": 1.0573,
      "step": 2262
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.406392491299065e-05,
      "loss": 1.217,
      "step": 2263
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.403617261251484e-05,
      "loss": 1.1618,
      "step": 2264
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.400841561711133e-05,
      "loss": 1.4243,
      "step": 2265
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.398065393606445e-05,
      "loss": 1.3887,
      "step": 2266
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.395288757866017e-05,
      "loss": 1.4644,
      "step": 2267
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.392511655418599e-05,
      "loss": 1.2544,
      "step": 2268
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.389734087193097e-05,
      "loss": 1.2246,
      "step": 2269
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.386956054118574e-05,
      "loss": 1.1736,
      "step": 2270
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.384177557124247e-05,
      "loss": 1.6672,
      "step": 2271
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.381398597139492e-05,
      "loss": 1.292,
      "step": 2272
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.378619175093834e-05,
      "loss": 1.0467,
      "step": 2273
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.375839291916957e-05,
      "loss": 1.4755,
      "step": 2274
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.373058948538698e-05,
      "loss": 1.4855,
      "step": 2275
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.370278145889048e-05,
      "loss": 1.5068,
      "step": 2276
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.36749688489815e-05,
      "loss": 1.331,
      "step": 2277
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.364715166496303e-05,
      "loss": 1.4006,
      "step": 2278
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.361932991613958e-05,
      "loss": 1.7102,
      "step": 2279
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.359150361181715e-05,
      "loss": 1.3883,
      "step": 2280
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.356367276130335e-05,
      "loss": 1.3194,
      "step": 2281
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.353583737390723e-05,
      "loss": 1.2618,
      "step": 2282
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.350799745893938e-05,
      "loss": 1.326,
      "step": 2283
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.348015302571192e-05,
      "loss": 1.2092,
      "step": 2284
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.345230408353848e-05,
      "loss": 1.0698,
      "step": 2285
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.342445064173417e-05,
      "loss": 1.1601,
      "step": 2286
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.339659270961566e-05,
      "loss": 1.3527,
      "step": 2287
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.336873029650104e-05,
      "loss": 1.5812,
      "step": 2288
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.334086341170997e-05,
      "loss": 1.3498,
      "step": 2289
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.331299206456358e-05,
      "loss": 1.1636,
      "step": 2290
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.32851162643845e-05,
      "loss": 1.2309,
      "step": 2291
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.325723602049684e-05,
      "loss": 1.0352,
      "step": 2292
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.32293513422262e-05,
      "loss": 1.2648,
      "step": 2293
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.320146223889966e-05,
      "loss": 1.2695,
      "step": 2294
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.317356871984578e-05,
      "loss": 1.2654,
      "step": 2295
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.314567079439459e-05,
      "loss": 1.2927,
      "step": 2296
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.311776847187766e-05,
      "loss": 1.1778,
      "step": 2297
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.30898617616279e-05,
      "loss": 1.1834,
      "step": 2298
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.30619506729798e-05,
      "loss": 1.2781,
      "step": 2299
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.303403521526928e-05,
      "loss": 1.2445,
      "step": 2300
    },
    {
      "epoch": 1.26,
      "eval_loss": 1.346007227897644,
      "eval_runtime": 36.7255,
      "eval_samples_per_second": 3.839,
      "eval_steps_per_second": 1.933,
      "step": 2300
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.300611539783372e-05,
      "loss": 1.3997,
      "step": 2301
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.297819123001193e-05,
      "loss": 1.2135,
      "step": 2302
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.295026272114424e-05,
      "loss": 1.2101,
      "step": 2303
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.292232988057235e-05,
      "loss": 1.2473,
      "step": 2304
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.289439271763949e-05,
      "loss": 1.4277,
      "step": 2305
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.28664512416903e-05,
      "loss": 1.5225,
      "step": 2306
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.283850546207082e-05,
      "loss": 1.1468,
      "step": 2307
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.281055538812861e-05,
      "loss": 1.3388,
      "step": 2308
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.278260102921263e-05,
      "loss": 1.2795,
      "step": 2309
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.275464239467325e-05,
      "loss": 1.2947,
      "step": 2310
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.272667949386228e-05,
      "loss": 1.3237,
      "step": 2311
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.269871233613301e-05,
      "loss": 1.3527,
      "step": 2312
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.267074093084006e-05,
      "loss": 1.1984,
      "step": 2313
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.264276528733955e-05,
      "loss": 1.3793,
      "step": 2314
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.2614785414989e-05,
      "loss": 1.3436,
      "step": 2315
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.258680132314731e-05,
      "loss": 1.216,
      "step": 2316
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.255881302117484e-05,
      "loss": 1.3015,
      "step": 2317
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.253082051843332e-05,
      "loss": 1.2884,
      "step": 2318
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.25028238242859e-05,
      "loss": 1.019,
      "step": 2319
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.247482294809712e-05,
      "loss": 1.4475,
      "step": 2320
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.244681789923297e-05,
      "loss": 1.119,
      "step": 2321
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.241880868706075e-05,
      "loss": 1.1403,
      "step": 2322
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.239079532094921e-05,
      "loss": 1.3866,
      "step": 2323
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.236277781026849e-05,
      "loss": 1.3507,
      "step": 2324
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.233475616439011e-05,
      "loss": 1.1978,
      "step": 2325
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.230673039268693e-05,
      "loss": 1.3746,
      "step": 2326
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.227870050453327e-05,
      "loss": 1.3577,
      "step": 2327
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.225066650930476e-05,
      "loss": 1.2129,
      "step": 2328
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.222262841637844e-05,
      "loss": 1.3692,
      "step": 2329
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.219458623513268e-05,
      "loss": 1.2179,
      "step": 2330
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.21665399749473e-05,
      "loss": 1.4359,
      "step": 2331
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.213848964520338e-05,
      "loss": 1.1324,
      "step": 2332
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.211043525528342e-05,
      "loss": 1.3258,
      "step": 2333
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.208237681457132e-05,
      "loss": 1.3863,
      "step": 2334
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.205431433245221e-05,
      "loss": 1.2698,
      "step": 2335
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.202624781831268e-05,
      "loss": 1.1209,
      "step": 2336
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.199817728154064e-05,
      "loss": 1.1825,
      "step": 2337
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.197010273152534e-05,
      "loss": 1.2299,
      "step": 2338
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.194202417765739e-05,
      "loss": 1.5486,
      "step": 2339
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.191394162932867e-05,
      "loss": 1.3655,
      "step": 2340
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.18858550959325e-05,
      "loss": 1.2299,
      "step": 2341
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.185776458686344e-05,
      "loss": 1.4759,
      "step": 2342
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.182967011151746e-05,
      "loss": 1.3196,
      "step": 2343
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.18015716792918e-05,
      "loss": 1.2238,
      "step": 2344
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.177346929958503e-05,
      "loss": 1.0862,
      "step": 2345
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.174536298179707e-05,
      "loss": 1.3453,
      "step": 2346
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.171725273532913e-05,
      "loss": 1.3418,
      "step": 2347
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.168913856958374e-05,
      "loss": 1.3407,
      "step": 2348
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.166102049396475e-05,
      "loss": 1.4058,
      "step": 2349
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.163289851787731e-05,
      "loss": 1.2055,
      "step": 2350
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.160477265072788e-05,
      "loss": 1.1035,
      "step": 2351
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.157664290192421e-05,
      "loss": 1.2769,
      "step": 2352
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.154850928087537e-05,
      "loss": 1.2586,
      "step": 2353
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.152037179699168e-05,
      "loss": 1.1008,
      "step": 2354
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.149223045968482e-05,
      "loss": 1.2319,
      "step": 2355
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.146408527836771e-05,
      "loss": 1.1445,
      "step": 2356
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.143593626245457e-05,
      "loss": 1.0251,
      "step": 2357
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.140778342136088e-05,
      "loss": 1.362,
      "step": 2358
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.137962676450347e-05,
      "loss": 1.3388,
      "step": 2359
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.135146630130034e-05,
      "loss": 1.235,
      "step": 2360
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.132330204117082e-05,
      "loss": 1.0641,
      "step": 2361
    },
    {
      "epoch": 1.29,
      "learning_rate": 6.129513399353558e-05,
      "loss": 1.2061,
      "step": 2362
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.126696216781641e-05,
      "loss": 1.404,
      "step": 2363
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.123878657343648e-05,
      "loss": 1.2322,
      "step": 2364
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.121060721982017e-05,
      "loss": 1.4122,
      "step": 2365
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.118242411639314e-05,
      "loss": 1.278,
      "step": 2366
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.115423727258226e-05,
      "loss": 1.2518,
      "step": 2367
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.112604669781572e-05,
      "loss": 1.2196,
      "step": 2368
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.109785240152292e-05,
      "loss": 1.3598,
      "step": 2369
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.106965439313448e-05,
      "loss": 1.2719,
      "step": 2370
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.104145268208232e-05,
      "loss": 1.2048,
      "step": 2371
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.1013247277799535e-05,
      "loss": 1.2937,
      "step": 2372
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.098503818972051e-05,
      "loss": 1.2406,
      "step": 2373
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.095682542728083e-05,
      "loss": 1.2555,
      "step": 2374
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.0928608999917305e-05,
      "loss": 0.9713,
      "step": 2375
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.090038891706801e-05,
      "loss": 1.4311,
      "step": 2376
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.087216518817221e-05,
      "loss": 1.2476,
      "step": 2377
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.084393782267039e-05,
      "loss": 1.3772,
      "step": 2378
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.081570683000426e-05,
      "loss": 1.3207,
      "step": 2379
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.078747221961675e-05,
      "loss": 1.3181,
      "step": 2380
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.075923400095198e-05,
      "loss": 1.3934,
      "step": 2381
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.0730992183455285e-05,
      "loss": 1.295,
      "step": 2382
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.0702746776573206e-05,
      "loss": 1.1971,
      "step": 2383
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.067449778975349e-05,
      "loss": 1.347,
      "step": 2384
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.06462452324451e-05,
      "loss": 1.2803,
      "step": 2385
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.061798911409813e-05,
      "loss": 1.3043,
      "step": 2386
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.0589729444163925e-05,
      "loss": 1.2165,
      "step": 2387
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.0561466232094996e-05,
      "loss": 1.3802,
      "step": 2388
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.053319948734506e-05,
      "loss": 1.2268,
      "step": 2389
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.050492921936897e-05,
      "loss": 1.3934,
      "step": 2390
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.0476655437622784e-05,
      "loss": 1.2074,
      "step": 2391
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.044837815156377e-05,
      "loss": 1.246,
      "step": 2392
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.042009737065029e-05,
      "loss": 1.193,
      "step": 2393
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.039181310434193e-05,
      "loss": 1.2053,
      "step": 2394
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.036352536209945e-05,
      "loss": 1.3383,
      "step": 2395
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.0335234153384735e-05,
      "loss": 1.1286,
      "step": 2396
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.030693948766085e-05,
      "loss": 1.4138,
      "step": 2397
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.027864137439202e-05,
      "loss": 1.4471,
      "step": 2398
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.025033982304362e-05,
      "loss": 1.3015,
      "step": 2399
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.022203484308216e-05,
      "loss": 1.3229,
      "step": 2400
    },
    {
      "epoch": 1.32,
      "eval_loss": 1.347601294517517,
      "eval_runtime": 36.7838,
      "eval_samples_per_second": 3.833,
      "eval_steps_per_second": 1.93,
      "step": 2400
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.019372644397532e-05,
      "loss": 1.191,
      "step": 2401
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.016541463519191e-05,
      "loss": 1.5396,
      "step": 2402
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.013709942620187e-05,
      "loss": 1.1458,
      "step": 2403
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.0108780826476306e-05,
      "loss": 1.1696,
      "step": 2404
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.008045884548745e-05,
      "loss": 1.0956,
      "step": 2405
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.005213349270864e-05,
      "loss": 1.3663,
      "step": 2406
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.0023804777614366e-05,
      "loss": 1.3415,
      "step": 2407
    },
    {
      "epoch": 1.32,
      "learning_rate": 5.9995472709680234e-05,
      "loss": 1.3746,
      "step": 2408
    },
    {
      "epoch": 1.32,
      "learning_rate": 5.9967137298382967e-05,
      "loss": 1.1571,
      "step": 2409
    },
    {
      "epoch": 1.32,
      "learning_rate": 5.993879855320043e-05,
      "loss": 1.3619,
      "step": 2410
    },
    {
      "epoch": 1.32,
      "learning_rate": 5.9910456483611565e-05,
      "loss": 1.1096,
      "step": 2411
    },
    {
      "epoch": 1.32,
      "learning_rate": 5.988211109909647e-05,
      "loss": 1.4591,
      "step": 2412
    },
    {
      "epoch": 1.32,
      "learning_rate": 5.98537624091363e-05,
      "loss": 1.5643,
      "step": 2413
    },
    {
      "epoch": 1.32,
      "learning_rate": 5.982541042321338e-05,
      "loss": 1.3093,
      "step": 2414
    },
    {
      "epoch": 1.32,
      "learning_rate": 5.979705515081108e-05,
      "loss": 1.3133,
      "step": 2415
    },
    {
      "epoch": 1.32,
      "learning_rate": 5.976869660141389e-05,
      "loss": 1.3085,
      "step": 2416
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.9740334784507415e-05,
      "loss": 1.3729,
      "step": 2417
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.9711969709578306e-05,
      "loss": 1.2205,
      "step": 2418
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.968360138611435e-05,
      "loss": 1.2682,
      "step": 2419
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.9655229823604406e-05,
      "loss": 1.3944,
      "step": 2420
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.962685503153841e-05,
      "loss": 1.1566,
      "step": 2421
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.9598477019407374e-05,
      "loss": 1.2013,
      "step": 2422
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.9570095796703415e-05,
      "loss": 1.209,
      "step": 2423
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.954171137291968e-05,
      "loss": 1.4751,
      "step": 2424
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.951332375755044e-05,
      "loss": 1.1359,
      "step": 2425
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.948493296009098e-05,
      "loss": 1.2623,
      "step": 2426
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.945653899003768e-05,
      "loss": 1.4466,
      "step": 2427
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.9428141856887986e-05,
      "loss": 1.3985,
      "step": 2428
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.939974157014039e-05,
      "loss": 1.1684,
      "step": 2429
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.937133813929442e-05,
      "loss": 1.2923,
      "step": 2430
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.934293157385068e-05,
      "loss": 0.9827,
      "step": 2431
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.931452188331083e-05,
      "loss": 1.3583,
      "step": 2432
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.928610907717756e-05,
      "loss": 1.5671,
      "step": 2433
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.9257693164954606e-05,
      "loss": 1.2953,
      "step": 2434
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.9229274156146744e-05,
      "loss": 1.2089,
      "step": 2435
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.920085206025978e-05,
      "loss": 1.1592,
      "step": 2436
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.917242688680055e-05,
      "loss": 1.1595,
      "step": 2437
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.9143998645276946e-05,
      "loss": 1.1681,
      "step": 2438
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.911556734519784e-05,
      "loss": 1.2578,
      "step": 2439
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.908713299607318e-05,
      "loss": 1.3612,
      "step": 2440
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.905869560741388e-05,
      "loss": 1.2209,
      "step": 2441
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.9030255188731944e-05,
      "loss": 1.1426,
      "step": 2442
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.9001811749540293e-05,
      "loss": 1.3913,
      "step": 2443
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.897336529935292e-05,
      "loss": 1.0379,
      "step": 2444
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.894491584768484e-05,
      "loss": 1.2731,
      "step": 2445
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.891646340405201e-05,
      "loss": 1.3031,
      "step": 2446
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.888800797797145e-05,
      "loss": 1.3575,
      "step": 2447
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.885954957896115e-05,
      "loss": 1.37,
      "step": 2448
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.883108821654008e-05,
      "loss": 1.1803,
      "step": 2449
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.8802623900228224e-05,
      "loss": 1.3081,
      "step": 2450
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.877415663954655e-05,
      "loss": 1.1782,
      "step": 2451
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.874568644401701e-05,
      "loss": 1.3878,
      "step": 2452
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.8717213323162534e-05,
      "loss": 1.3016,
      "step": 2453
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.8688737286507045e-05,
      "loss": 1.0875,
      "step": 2454
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.8660258343575405e-05,
      "loss": 1.2339,
      "step": 2455
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.863177650389347e-05,
      "loss": 1.5544,
      "step": 2456
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.8603291776988077e-05,
      "loss": 1.4501,
      "step": 2457
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.857480417238702e-05,
      "loss": 1.4059,
      "step": 2458
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.854631369961904e-05,
      "loss": 1.2919,
      "step": 2459
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.8517820368213874e-05,
      "loss": 1.1701,
      "step": 2460
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.848932418770218e-05,
      "loss": 1.5769,
      "step": 2461
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.8460825167615585e-05,
      "loss": 1.1497,
      "step": 2462
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.843232331748666e-05,
      "loss": 1.5006,
      "step": 2463
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.8403818646848915e-05,
      "loss": 1.3513,
      "step": 2464
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.837531116523682e-05,
      "loss": 1.3386,
      "step": 2465
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.834680088218579e-05,
      "loss": 1.3936,
      "step": 2466
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.831828780723214e-05,
      "loss": 1.3202,
      "step": 2467
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.8289771949913176e-05,
      "loss": 1.2353,
      "step": 2468
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.8261253319767076e-05,
      "loss": 1.3293,
      "step": 2469
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.823273192633297e-05,
      "loss": 1.2983,
      "step": 2470
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.820420777915094e-05,
      "loss": 1.1926,
      "step": 2471
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.8175680887761955e-05,
      "loss": 1.0125,
      "step": 2472
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.814715126170789e-05,
      "loss": 1.3148,
      "step": 2473
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.811861891053158e-05,
      "loss": 1.4149,
      "step": 2474
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.809008384377673e-05,
      "loss": 1.4511,
      "step": 2475
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.8061546070987994e-05,
      "loss": 1.2651,
      "step": 2476
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.8033005601710886e-05,
      "loss": 1.3499,
      "step": 2477
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.800446244549185e-05,
      "loss": 1.6136,
      "step": 2478
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.79759166118782e-05,
      "loss": 1.2291,
      "step": 2479
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.794736811041821e-05,
      "loss": 1.2305,
      "step": 2480
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.791881695066097e-05,
      "loss": 1.2236,
      "step": 2481
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.789026314215651e-05,
      "loss": 1.3133,
      "step": 2482
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.786170669445572e-05,
      "loss": 1.1843,
      "step": 2483
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.783314761711038e-05,
      "loss": 1.0994,
      "step": 2484
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.780458591967317e-05,
      "loss": 1.4175,
      "step": 2485
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.7776021611697604e-05,
      "loss": 1.4491,
      "step": 2486
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.774745470273811e-05,
      "loss": 1.266,
      "step": 2487
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.771888520234997e-05,
      "loss": 1.2967,
      "step": 2488
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.769031312008931e-05,
      "loss": 1.3036,
      "step": 2489
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.766173846551316e-05,
      "loss": 1.4791,
      "step": 2490
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.763316124817939e-05,
      "loss": 1.5116,
      "step": 2491
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.7604581477646726e-05,
      "loss": 1.2209,
      "step": 2492
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.7575999163474746e-05,
      "loss": 1.4627,
      "step": 2493
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.754741431522389e-05,
      "loss": 1.2821,
      "step": 2494
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.751882694245545e-05,
      "loss": 1.369,
      "step": 2495
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.749023705473153e-05,
      "loss": 1.1912,
      "step": 2496
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.746164466161511e-05,
      "loss": 1.515,
      "step": 2497
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.743304977266999e-05,
      "loss": 1.0519,
      "step": 2498
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.7404452397460816e-05,
      "loss": 1.5066,
      "step": 2499
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.737585254555307e-05,
      "loss": 1.3464,
      "step": 2500
    },
    {
      "epoch": 1.37,
      "eval_loss": 1.3439135551452637,
      "eval_runtime": 36.7954,
      "eval_samples_per_second": 3.832,
      "eval_steps_per_second": 1.93,
      "step": 2500
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.734725022651302e-05,
      "loss": 1.2148,
      "step": 2501
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.731864544990782e-05,
      "loss": 1.3703,
      "step": 2502
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.729003822530539e-05,
      "loss": 1.2933,
      "step": 2503
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.726142856227452e-05,
      "loss": 1.3838,
      "step": 2504
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.723281647038479e-05,
      "loss": 1.2092,
      "step": 2505
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.7204201959206584e-05,
      "loss": 1.0982,
      "step": 2506
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.717558503831111e-05,
      "loss": 1.0454,
      "step": 2507
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.714696571727037e-05,
      "loss": 1.2342,
      "step": 2508
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.7118344005657196e-05,
      "loss": 1.2704,
      "step": 2509
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.708971991304518e-05,
      "loss": 1.3637,
      "step": 2510
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.706109344900874e-05,
      "loss": 1.2651,
      "step": 2511
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.703246462312307e-05,
      "loss": 1.5256,
      "step": 2512
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.700383344496418e-05,
      "loss": 1.2187,
      "step": 2513
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.6975199924108826e-05,
      "loss": 1.2587,
      "step": 2514
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.694656407013458e-05,
      "loss": 1.3894,
      "step": 2515
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.6917925892619774e-05,
      "loss": 1.1207,
      "step": 2516
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.6889285401143545e-05,
      "loss": 1.3782,
      "step": 2517
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.686064260528578e-05,
      "loss": 1.4345,
      "step": 2518
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.6831997514627124e-05,
      "loss": 1.0475,
      "step": 2519
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.6803350138749034e-05,
      "loss": 1.2833,
      "step": 2520
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.677470048723369e-05,
      "loss": 1.3997,
      "step": 2521
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.674604856966407e-05,
      "loss": 1.1472,
      "step": 2522
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.6717394395623865e-05,
      "loss": 1.1976,
      "step": 2523
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.6688737974697556e-05,
      "loss": 1.2886,
      "step": 2524
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.666007931647038e-05,
      "loss": 1.2274,
      "step": 2525
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.6631418430528284e-05,
      "loss": 1.0501,
      "step": 2526
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.660275532645799e-05,
      "loss": 1.4633,
      "step": 2527
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.6574090013846946e-05,
      "loss": 1.0743,
      "step": 2528
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.6545422502283364e-05,
      "loss": 1.364,
      "step": 2529
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.6516752801356177e-05,
      "loss": 1.2229,
      "step": 2530
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.648808092065503e-05,
      "loss": 1.0916,
      "step": 2531
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.645940686977033e-05,
      "loss": 1.2368,
      "step": 2532
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.643073065829319e-05,
      "loss": 1.5034,
      "step": 2533
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.640205229581546e-05,
      "loss": 1.0638,
      "step": 2534
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.637337179192967e-05,
      "loss": 1.0919,
      "step": 2535
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.634468915622915e-05,
      "loss": 1.2221,
      "step": 2536
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.631600439830785e-05,
      "loss": 1.2065,
      "step": 2537
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.6287317527760485e-05,
      "loss": 1.3151,
      "step": 2538
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.6258628554182446e-05,
      "loss": 1.1939,
      "step": 2539
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.622993748716987e-05,
      "loss": 1.417,
      "step": 2540
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.6201244336319534e-05,
      "loss": 1.3365,
      "step": 2541
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.617254911122898e-05,
      "loss": 1.1924,
      "step": 2542
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.614385182149638e-05,
      "loss": 1.1453,
      "step": 2543
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.6115152476720635e-05,
      "loss": 1.2942,
      "step": 2544
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.6086451086501334e-05,
      "loss": 1.2565,
      "step": 2545
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.6057747660438744e-05,
      "loss": 0.9227,
      "step": 2546
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.60290422081338e-05,
      "loss": 1.3896,
      "step": 2547
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.600033473918811e-05,
      "loss": 1.5363,
      "step": 2548
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.597162526320399e-05,
      "loss": 1.1671,
      "step": 2549
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.594291378978439e-05,
      "loss": 1.3293,
      "step": 2550
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.591420032853297e-05,
      "loss": 1.1162,
      "step": 2551
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.5885484889054016e-05,
      "loss": 1.1233,
      "step": 2552
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.585676748095249e-05,
      "loss": 1.2658,
      "step": 2553
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.582804811383402e-05,
      "loss": 1.1808,
      "step": 2554
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.5799326797304873e-05,
      "loss": 1.3706,
      "step": 2555
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.577060354097199e-05,
      "loss": 1.4421,
      "step": 2556
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.574187835444293e-05,
      "loss": 1.434,
      "step": 2557
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.5713151247325936e-05,
      "loss": 1.4268,
      "step": 2558
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.568442222922987e-05,
      "loss": 1.2337,
      "step": 2559
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.565569130976422e-05,
      "loss": 1.309,
      "step": 2560
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.562695849853914e-05,
      "loss": 1.1532,
      "step": 2561
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.559822380516539e-05,
      "loss": 1.2853,
      "step": 2562
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.5569487239254405e-05,
      "loss": 1.4219,
      "step": 2563
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.554074881041817e-05,
      "loss": 1.0067,
      "step": 2564
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.551200852826938e-05,
      "loss": 1.4146,
      "step": 2565
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.548326640242127e-05,
      "loss": 0.8813,
      "step": 2566
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.5454522442487744e-05,
      "loss": 1.3512,
      "step": 2567
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.542577665808332e-05,
      "loss": 1.164,
      "step": 2568
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.539702905882309e-05,
      "loss": 1.3892,
      "step": 2569
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.536827965432277e-05,
      "loss": 1.5047,
      "step": 2570
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.533952845419872e-05,
      "loss": 1.3964,
      "step": 2571
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.531077546806782e-05,
      "loss": 1.4559,
      "step": 2572
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.528202070554761e-05,
      "loss": 1.4975,
      "step": 2573
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.52532641762562e-05,
      "loss": 1.4121,
      "step": 2574
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.522450588981232e-05,
      "loss": 1.213,
      "step": 2575
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.5195745855835226e-05,
      "loss": 1.2789,
      "step": 2576
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.516698408394482e-05,
      "loss": 1.296,
      "step": 2577
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.5138220583761556e-05,
      "loss": 1.4433,
      "step": 2578
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.5109455364906485e-05,
      "loss": 1.4325,
      "step": 2579
    },
    {
      "epoch": 1.41,
      "learning_rate": 5.508068843700121e-05,
      "loss": 1.1154,
      "step": 2580
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.5051919809667904e-05,
      "loss": 1.2191,
      "step": 2581
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.502314949252935e-05,
      "loss": 1.3887,
      "step": 2582
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.4994377495208834e-05,
      "loss": 1.141,
      "step": 2583
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.496560382733028e-05,
      "loss": 1.3455,
      "step": 2584
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.493682849851808e-05,
      "loss": 1.0953,
      "step": 2585
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.490805151839724e-05,
      "loss": 1.4187,
      "step": 2586
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.487927289659332e-05,
      "loss": 1.3979,
      "step": 2587
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.4850492642732406e-05,
      "loss": 1.3404,
      "step": 2588
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.482171076644116e-05,
      "loss": 1.3115,
      "step": 2589
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.479292727734674e-05,
      "loss": 0.9362,
      "step": 2590
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.476414218507688e-05,
      "loss": 1.1086,
      "step": 2591
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.4735355499259855e-05,
      "loss": 1.2548,
      "step": 2592
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.470656722952444e-05,
      "loss": 1.2894,
      "step": 2593
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.467777738549997e-05,
      "loss": 1.3149,
      "step": 2594
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.4648985976816295e-05,
      "loss": 1.1731,
      "step": 2595
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.462019301310378e-05,
      "loss": 1.2873,
      "step": 2596
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.459139850399332e-05,
      "loss": 1.2781,
      "step": 2597
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.456260245911633e-05,
      "loss": 1.215,
      "step": 2598
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.453380488810472e-05,
      "loss": 1.384,
      "step": 2599
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.4505005800590945e-05,
      "loss": 1.2651,
      "step": 2600
    },
    {
      "epoch": 1.43,
      "eval_loss": 1.3438740968704224,
      "eval_runtime": 36.8492,
      "eval_samples_per_second": 3.826,
      "eval_steps_per_second": 1.927,
      "step": 2600
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.447620520620794e-05,
      "loss": 1.051,
      "step": 2601
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.444740311458915e-05,
      "loss": 1.4696,
      "step": 2602
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.441859953536852e-05,
      "loss": 1.2407,
      "step": 2603
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.4389794478180486e-05,
      "loss": 1.3052,
      "step": 2604
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.4360987952660003e-05,
      "loss": 1.3792,
      "step": 2605
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.43321799684425e-05,
      "loss": 1.2866,
      "step": 2606
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.430337053516388e-05,
      "loss": 1.4711,
      "step": 2607
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.427455966246057e-05,
      "loss": 1.191,
      "step": 2608
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.424574735996941e-05,
      "loss": 1.5,
      "step": 2609
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.4216933637327806e-05,
      "loss": 1.2996,
      "step": 2610
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.418811850417358e-05,
      "loss": 1.2099,
      "step": 2611
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.415930197014503e-05,
      "loss": 1.4516,
      "step": 2612
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.4130484044880944e-05,
      "loss": 1.5036,
      "step": 2613
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.410166473802057e-05,
      "loss": 1.1837,
      "step": 2614
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.407284405920361e-05,
      "loss": 1.1809,
      "step": 2615
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.4044022018070214e-05,
      "loss": 1.106,
      "step": 2616
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.401519862426102e-05,
      "loss": 1.4588,
      "step": 2617
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.398637388741709e-05,
      "loss": 1.4255,
      "step": 2618
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.3957547817179943e-05,
      "loss": 1.4519,
      "step": 2619
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.392872042319155e-05,
      "loss": 1.1175,
      "step": 2620
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.389989171509434e-05,
      "loss": 1.1208,
      "step": 2621
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.387106170253113e-05,
      "loss": 1.3134,
      "step": 2622
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.384223039514522e-05,
      "loss": 1.4374,
      "step": 2623
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.3813397802580334e-05,
      "loss": 1.3374,
      "step": 2624
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.37845639344806e-05,
      "loss": 1.3374,
      "step": 2625
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.375572880049062e-05,
      "loss": 1.2401,
      "step": 2626
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.3726892410255394e-05,
      "loss": 1.2411,
      "step": 2627
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.369805477342032e-05,
      "loss": 1.1554,
      "step": 2628
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.366921589963123e-05,
      "loss": 1.4064,
      "step": 2629
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.36403757985344e-05,
      "loss": 1.0422,
      "step": 2630
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.361153447977648e-05,
      "loss": 1.1262,
      "step": 2631
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.358269195300454e-05,
      "loss": 1.4904,
      "step": 2632
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.3553848227866046e-05,
      "loss": 1.2287,
      "step": 2633
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.352500331400886e-05,
      "loss": 1.2329,
      "step": 2634
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.349615722108128e-05,
      "loss": 1.3101,
      "step": 2635
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.3467309958731934e-05,
      "loss": 1.0861,
      "step": 2636
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.343846153660992e-05,
      "loss": 1.4553,
      "step": 2637
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.340961196436465e-05,
      "loss": 1.302,
      "step": 2638
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.338076125164596e-05,
      "loss": 1.1738,
      "step": 2639
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.335190940810407e-05,
      "loss": 1.1325,
      "step": 2640
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.332305644338955e-05,
      "loss": 1.3023,
      "step": 2641
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.329420236715339e-05,
      "loss": 1.2482,
      "step": 2642
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.32653471890469e-05,
      "loss": 0.9675,
      "step": 2643
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.3236490918721794e-05,
      "loss": 1.2949,
      "step": 2644
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.320763356583014e-05,
      "loss": 1.3013,
      "step": 2645
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.317877514002436e-05,
      "loss": 1.4556,
      "step": 2646
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.3149915650957246e-05,
      "loss": 1.2158,
      "step": 2647
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.312105510828196e-05,
      "loss": 1.3626,
      "step": 2648
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.309219352165199e-05,
      "loss": 1.0869,
      "step": 2649
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.3063330900721175e-05,
      "loss": 1.1843,
      "step": 2650
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.3034467255143717e-05,
      "loss": 1.239,
      "step": 2651
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.3005602594574145e-05,
      "loss": 1.3657,
      "step": 2652
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.297673692866735e-05,
      "loss": 1.1595,
      "step": 2653
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.2947870267078514e-05,
      "loss": 1.4118,
      "step": 2654
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.291900261946323e-05,
      "loss": 1.3417,
      "step": 2655
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.289013399547732e-05,
      "loss": 1.3795,
      "step": 2656
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.286126440477701e-05,
      "loss": 1.2505,
      "step": 2657
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.2832393857018815e-05,
      "loss": 1.5066,
      "step": 2658
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.2803522361859594e-05,
      "loss": 1.0763,
      "step": 2659
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.27746499289565e-05,
      "loss": 1.3245,
      "step": 2660
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.2745776567967e-05,
      "loss": 1.172,
      "step": 2661
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.27169022885489e-05,
      "loss": 1.3713,
      "step": 2662
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.2688027100360273e-05,
      "loss": 1.2999,
      "step": 2663
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.265915101305952e-05,
      "loss": 1.0413,
      "step": 2664
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.2630274036305335e-05,
      "loss": 1.2728,
      "step": 2665
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.260139617975673e-05,
      "loss": 1.1971,
      "step": 2666
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.257251745307299e-05,
      "loss": 1.3922,
      "step": 2667
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.2543637865913675e-05,
      "loss": 1.1812,
      "step": 2668
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.251475742793868e-05,
      "loss": 1.3421,
      "step": 2669
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.248587614880812e-05,
      "loss": 1.22,
      "step": 2670
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.245699403818247e-05,
      "loss": 1.3591,
      "step": 2671
    },
    {
      "epoch": 1.46,
      "learning_rate": 5.242811110572242e-05,
      "loss": 1.0404,
      "step": 2672
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.239922736108895e-05,
      "loss": 1.2999,
      "step": 2673
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.237034281394335e-05,
      "loss": 1.0345,
      "step": 2674
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.234145747394712e-05,
      "loss": 1.3118,
      "step": 2675
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.2312571350762054e-05,
      "loss": 1.2883,
      "step": 2676
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.228368445405022e-05,
      "loss": 1.0249,
      "step": 2677
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.225479679347392e-05,
      "loss": 1.2112,
      "step": 2678
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.222590837869572e-05,
      "loss": 1.1174,
      "step": 2679
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.219701921937845e-05,
      "loss": 1.5658,
      "step": 2680
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.2168129325185177e-05,
      "loss": 0.9608,
      "step": 2681
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.213923870577921e-05,
      "loss": 1.3886,
      "step": 2682
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.211034737082412e-05,
      "loss": 1.2329,
      "step": 2683
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.208145532998368e-05,
      "loss": 1.2834,
      "step": 2684
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.205256259292195e-05,
      "loss": 1.5192,
      "step": 2685
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.2023669169303194e-05,
      "loss": 1.1421,
      "step": 2686
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.199477506879189e-05,
      "loss": 0.9834,
      "step": 2687
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.1965880301052784e-05,
      "loss": 1.2115,
      "step": 2688
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.19369848757508e-05,
      "loss": 1.1803,
      "step": 2689
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.19080888025511e-05,
      "loss": 1.4576,
      "step": 2690
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.18791920911191e-05,
      "loss": 1.2914,
      "step": 2691
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.1850294751120375e-05,
      "loss": 1.3468,
      "step": 2692
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.182139679222072e-05,
      "loss": 1.2066,
      "step": 2693
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.1792498224086175e-05,
      "loss": 1.4122,
      "step": 2694
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.1763599056382915e-05,
      "loss": 1.2173,
      "step": 2695
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.17346992987774e-05,
      "loss": 1.3119,
      "step": 2696
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.170579896093622e-05,
      "loss": 1.0566,
      "step": 2697
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.16768980525262e-05,
      "loss": 1.2987,
      "step": 2698
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.164799658321433e-05,
      "loss": 1.5232,
      "step": 2699
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.1619094562667804e-05,
      "loss": 1.516,
      "step": 2700
    },
    {
      "epoch": 1.48,
      "eval_loss": 1.3423593044281006,
      "eval_runtime": 36.8371,
      "eval_samples_per_second": 3.828,
      "eval_steps_per_second": 1.927,
      "step": 2700
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.1590192000553984e-05,
      "loss": 1.3623,
      "step": 2701
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.156128890654042e-05,
      "loss": 1.1513,
      "step": 2702
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.153238529029486e-05,
      "loss": 1.4342,
      "step": 2703
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.15034811614852e-05,
      "loss": 1.2797,
      "step": 2704
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.147457652977952e-05,
      "loss": 1.0642,
      "step": 2705
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.144567140484605e-05,
      "loss": 1.1304,
      "step": 2706
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.1416765796353214e-05,
      "loss": 1.2579,
      "step": 2707
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.138785971396959e-05,
      "loss": 1.48,
      "step": 2708
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.135895316736389e-05,
      "loss": 1.3297,
      "step": 2709
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.133004616620501e-05,
      "loss": 1.2712,
      "step": 2710
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.130113872016199e-05,
      "loss": 1.3334,
      "step": 2711
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.1272230838904015e-05,
      "loss": 0.9698,
      "step": 2712
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.124332253210041e-05,
      "loss": 1.4565,
      "step": 2713
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.121441380942066e-05,
      "loss": 1.5314,
      "step": 2714
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.118550468053437e-05,
      "loss": 1.6117,
      "step": 2715
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.11565951551113e-05,
      "loss": 1.1302,
      "step": 2716
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.112768524282132e-05,
      "loss": 1.1427,
      "step": 2717
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.109877495333444e-05,
      "loss": 1.2181,
      "step": 2718
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.1069864296320814e-05,
      "loss": 1.5925,
      "step": 2719
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.1040953281450684e-05,
      "loss": 1.3297,
      "step": 2720
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.101204191839444e-05,
      "loss": 1.3511,
      "step": 2721
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.098313021682257e-05,
      "loss": 0.9995,
      "step": 2722
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.09542181864057e-05,
      "loss": 1.3339,
      "step": 2723
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.0925305836814544e-05,
      "loss": 1.2102,
      "step": 2724
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.0896393177719917e-05,
      "loss": 1.2093,
      "step": 2725
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.086748021879278e-05,
      "loss": 1.386,
      "step": 2726
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.083856696970411e-05,
      "loss": 1.2247,
      "step": 2727
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.080965344012508e-05,
      "loss": 1.3136,
      "step": 2728
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.078073963972689e-05,
      "loss": 1.0948,
      "step": 2729
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.075182557818087e-05,
      "loss": 1.3305,
      "step": 2730
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.072291126515839e-05,
      "loss": 1.2784,
      "step": 2731
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.069399671033096e-05,
      "loss": 1.1724,
      "step": 2732
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.066508192337014e-05,
      "loss": 1.3459,
      "step": 2733
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.0636166913947556e-05,
      "loss": 1.1022,
      "step": 2734
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.060725169173495e-05,
      "loss": 1.032,
      "step": 2735
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.057833626640408e-05,
      "loss": 1.3943,
      "step": 2736
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.0549420647626825e-05,
      "loss": 1.2555,
      "step": 2737
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.052050484507511e-05,
      "loss": 0.8879,
      "step": 2738
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.049158886842088e-05,
      "loss": 1.1408,
      "step": 2739
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.046267272733621e-05,
      "loss": 1.2809,
      "step": 2740
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.043375643149317e-05,
      "loss": 1.3241,
      "step": 2741
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.0404839990563936e-05,
      "loss": 1.4291,
      "step": 2742
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.037592341422068e-05,
      "loss": 1.3781,
      "step": 2743
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.0347006712135646e-05,
      "loss": 1.2037,
      "step": 2744
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.031808989398112e-05,
      "loss": 1.4884,
      "step": 2745
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.028917296942942e-05,
      "loss": 1.3335,
      "step": 2746
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.0260255948152914e-05,
      "loss": 1.2955,
      "step": 2747
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.023133883982397e-05,
      "loss": 1.443,
      "step": 2748
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.020242165411503e-05,
      "loss": 1.3214,
      "step": 2749
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.017350440069853e-05,
      "loss": 1.1729,
      "step": 2750
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.014458708924693e-05,
      "loss": 1.3253,
      "step": 2751
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.011566972943272e-05,
      "loss": 1.3066,
      "step": 2752
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.00867523309284e-05,
      "loss": 1.4454,
      "step": 2753
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.00578349034065e-05,
      "loss": 1.2377,
      "step": 2754
    },
    {
      "epoch": 1.51,
      "learning_rate": 5.0028917456539515e-05,
      "loss": 1.3093,
      "step": 2755
    },
    {
      "epoch": 1.51,
      "learning_rate": 5e-05,
      "loss": 1.1989,
      "step": 2756
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.99710825434605e-05,
      "loss": 1.2678,
      "step": 2757
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.994216509659352e-05,
      "loss": 1.2493,
      "step": 2758
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.9913247669071607e-05,
      "loss": 1.4155,
      "step": 2759
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.988433027056729e-05,
      "loss": 1.1575,
      "step": 2760
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.985541291075308e-05,
      "loss": 0.8821,
      "step": 2761
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.982649559930147e-05,
      "loss": 1.1993,
      "step": 2762
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.979757834588498e-05,
      "loss": 1.2232,
      "step": 2763
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.976866116017604e-05,
      "loss": 1.1284,
      "step": 2764
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.97397440518471e-05,
      "loss": 1.4254,
      "step": 2765
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.971082703057059e-05,
      "loss": 1.5937,
      "step": 2766
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.968191010601889e-05,
      "loss": 1.4062,
      "step": 2767
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.9652993287864365e-05,
      "loss": 1.1059,
      "step": 2768
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.9624076585779335e-05,
      "loss": 1.3125,
      "step": 2769
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.9595160009436075e-05,
      "loss": 1.3196,
      "step": 2770
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.9566243568506834e-05,
      "loss": 1.2288,
      "step": 2771
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.95373272726638e-05,
      "loss": 1.4148,
      "step": 2772
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.9508411131579127e-05,
      "loss": 1.4525,
      "step": 2773
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.947949515492491e-05,
      "loss": 1.3018,
      "step": 2774
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.945057935237318e-05,
      "loss": 0.9521,
      "step": 2775
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.942166373359593e-05,
      "loss": 1.5006,
      "step": 2776
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.939274830826506e-05,
      "loss": 1.374,
      "step": 2777
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.936383308605245e-05,
      "loss": 1.3592,
      "step": 2778
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.933491807662987e-05,
      "loss": 1.1174,
      "step": 2779
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.9306003289669046e-05,
      "loss": 1.282,
      "step": 2780
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.927708873484162e-05,
      "loss": 1.2824,
      "step": 2781
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.924817442181915e-05,
      "loss": 1.2948,
      "step": 2782
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.921926036027312e-05,
      "loss": 1.2255,
      "step": 2783
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.919034655987493e-05,
      "loss": 1.5248,
      "step": 2784
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.916143303029589e-05,
      "loss": 1.4503,
      "step": 2785
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.9132519781207235e-05,
      "loss": 1.4703,
      "step": 2786
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.910360682228009e-05,
      "loss": 1.3212,
      "step": 2787
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.907469416318547e-05,
      "loss": 1.3028,
      "step": 2788
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.904578181359431e-05,
      "loss": 1.2548,
      "step": 2789
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.9016869783177434e-05,
      "loss": 1.5237,
      "step": 2790
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.898795808160557e-05,
      "loss": 1.286,
      "step": 2791
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.895904671854933e-05,
      "loss": 1.4157,
      "step": 2792
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.8930135703679204e-05,
      "loss": 1.4934,
      "step": 2793
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.890122504666557e-05,
      "loss": 1.4054,
      "step": 2794
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.88723147571787e-05,
      "loss": 1.3258,
      "step": 2795
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.884340484488871e-05,
      "loss": 1.3781,
      "step": 2796
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.8814495319465636e-05,
      "loss": 1.2936,
      "step": 2797
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.8785586190579355e-05,
      "loss": 1.3183,
      "step": 2798
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.875667746789961e-05,
      "loss": 1.1438,
      "step": 2799
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.872776916109601e-05,
      "loss": 1.4323,
      "step": 2800
    },
    {
      "epoch": 1.54,
      "eval_loss": 1.3413376808166504,
      "eval_runtime": 36.8479,
      "eval_samples_per_second": 3.827,
      "eval_steps_per_second": 1.927,
      "step": 2800
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.869886127983803e-05,
      "loss": 1.4831,
      "step": 2801
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.866995383379501e-05,
      "loss": 1.2536,
      "step": 2802
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.8641046832636135e-05,
      "loss": 1.2794,
      "step": 2803
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.861214028603044e-05,
      "loss": 1.2865,
      "step": 2804
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.8583234203646804e-05,
      "loss": 1.3933,
      "step": 2805
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.855432859515397e-05,
      "loss": 0.9477,
      "step": 2806
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.8525423470220506e-05,
      "loss": 1.4352,
      "step": 2807
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.849651883851481e-05,
      "loss": 1.2996,
      "step": 2808
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.846761470970516e-05,
      "loss": 1.1606,
      "step": 2809
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.843871109345959e-05,
      "loss": 1.224,
      "step": 2810
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.8409807999446034e-05,
      "loss": 1.4012,
      "step": 2811
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.838090543733222e-05,
      "loss": 1.2177,
      "step": 2812
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.835200341678569e-05,
      "loss": 1.2476,
      "step": 2813
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.832310194747382e-05,
      "loss": 1.2231,
      "step": 2814
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.8294201039063796e-05,
      "loss": 1.2561,
      "step": 2815
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.826530070122262e-05,
      "loss": 1.3433,
      "step": 2816
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.8236400943617097e-05,
      "loss": 1.2197,
      "step": 2817
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.820750177591386e-05,
      "loss": 1.2566,
      "step": 2818
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.817860320777929e-05,
      "loss": 1.1177,
      "step": 2819
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.814970524887964e-05,
      "loss": 1.3384,
      "step": 2820
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.812080790888091e-05,
      "loss": 1.2902,
      "step": 2821
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.8091911197448906e-05,
      "loss": 1.1847,
      "step": 2822
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.8063015124249215e-05,
      "loss": 1.3428,
      "step": 2823
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.803411969894724e-05,
      "loss": 1.3379,
      "step": 2824
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.800522493120813e-05,
      "loss": 1.4252,
      "step": 2825
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.797633083069683e-05,
      "loss": 1.2571,
      "step": 2826
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.794743740707807e-05,
      "loss": 1.3132,
      "step": 2827
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.7918544670016335e-05,
      "loss": 1.2601,
      "step": 2828
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.788965262917591e-05,
      "loss": 1.4643,
      "step": 2829
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.786076129422081e-05,
      "loss": 1.212,
      "step": 2830
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.783187067481484e-05,
      "loss": 1.2619,
      "step": 2831
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.780298078062157e-05,
      "loss": 1.2866,
      "step": 2832
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.7774091621304294e-05,
      "loss": 1.1987,
      "step": 2833
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.77452032065261e-05,
      "loss": 1.1509,
      "step": 2834
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.77163155459498e-05,
      "loss": 1.0878,
      "step": 2835
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.768742864923797e-05,
      "loss": 1.2424,
      "step": 2836
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.765854252605291e-05,
      "loss": 1.3425,
      "step": 2837
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.7629657186056673e-05,
      "loss": 1.3718,
      "step": 2838
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.760077263891106e-05,
      "loss": 1.2271,
      "step": 2839
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.7571888894277604e-05,
      "loss": 1.1407,
      "step": 2840
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.7543005961817535e-05,
      "loss": 1.1294,
      "step": 2841
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.7514123851191875e-05,
      "loss": 1.5967,
      "step": 2842
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.748524257206134e-05,
      "loss": 1.3458,
      "step": 2843
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.745636213408633e-05,
      "loss": 1.2071,
      "step": 2844
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.742748254692702e-05,
      "loss": 1.2541,
      "step": 2845
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.739860382024327e-05,
      "loss": 1.2492,
      "step": 2846
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.736972596369466e-05,
      "loss": 1.4519,
      "step": 2847
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.7340848986940487e-05,
      "loss": 1.144,
      "step": 2848
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.731197289963973e-05,
      "loss": 1.2854,
      "step": 2849
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.72830977114511e-05,
      "loss": 0.9813,
      "step": 2850
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.7254223432033e-05,
      "loss": 1.2261,
      "step": 2851
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.72253500710435e-05,
      "loss": 1.1316,
      "step": 2852
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.7196477638140404e-05,
      "loss": 1.3374,
      "step": 2853
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.716760614298118e-05,
      "loss": 1.2643,
      "step": 2854
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.7138735595222994e-05,
      "loss": 1.1238,
      "step": 2855
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.710986600452269e-05,
      "loss": 1.4179,
      "step": 2856
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.7080997380536786e-05,
      "loss": 1.2392,
      "step": 2857
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.705212973292148e-05,
      "loss": 1.0099,
      "step": 2858
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.7023263071332656e-05,
      "loss": 1.4023,
      "step": 2859
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.699439740542585e-05,
      "loss": 1.3839,
      "step": 2860
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.696553274485628e-05,
      "loss": 1.0829,
      "step": 2861
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.6936669099278823e-05,
      "loss": 1.158,
      "step": 2862
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.6907806478348015e-05,
      "loss": 1.3369,
      "step": 2863
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.687894489171804e-05,
      "loss": 1.2068,
      "step": 2864
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.685008434904275e-05,
      "loss": 1.135,
      "step": 2865
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.682122485997564e-05,
      "loss": 1.0026,
      "step": 2866
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.679236643416987e-05,
      "loss": 1.4711,
      "step": 2867
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.676350908127822e-05,
      "loss": 1.1839,
      "step": 2868
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.6734652810953105e-05,
      "loss": 1.3618,
      "step": 2869
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.6705797632846614e-05,
      "loss": 1.2264,
      "step": 2870
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.667694355661045e-05,
      "loss": 1.296,
      "step": 2871
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.6648090591895935e-05,
      "loss": 1.3319,
      "step": 2872
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.6619238748354036e-05,
      "loss": 1.4495,
      "step": 2873
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.6590388035635355e-05,
      "loss": 1.0679,
      "step": 2874
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.6561538463390084e-05,
      "loss": 1.2388,
      "step": 2875
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.653269004126806e-05,
      "loss": 1.3591,
      "step": 2876
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.650384277891873e-05,
      "loss": 1.3267,
      "step": 2877
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.6474996685991134e-05,
      "loss": 1.2967,
      "step": 2878
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.644615177213395e-05,
      "loss": 1.0614,
      "step": 2879
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.641730804699547e-05,
      "loss": 1.1052,
      "step": 2880
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.638846552022352e-05,
      "loss": 1.2071,
      "step": 2881
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.63596242014656e-05,
      "loss": 1.2605,
      "step": 2882
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.633078410036877e-05,
      "loss": 1.0725,
      "step": 2883
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.6301945226579695e-05,
      "loss": 1.1197,
      "step": 2884
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.627310758974462e-05,
      "loss": 1.4101,
      "step": 2885
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.624427119950938e-05,
      "loss": 1.2834,
      "step": 2886
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.6215436065519404e-05,
      "loss": 1.1328,
      "step": 2887
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.6186602197419685e-05,
      "loss": 1.3299,
      "step": 2888
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.615776960485479e-05,
      "loss": 1.2967,
      "step": 2889
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.612893829746888e-05,
      "loss": 1.0761,
      "step": 2890
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.610010828490567e-05,
      "loss": 1.3464,
      "step": 2891
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.6071279576808454e-05,
      "loss": 1.0113,
      "step": 2892
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.604245218282007e-05,
      "loss": 1.2516,
      "step": 2893
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.6013626112582924e-05,
      "loss": 1.4454,
      "step": 2894
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.5984801375738994e-05,
      "loss": 1.2998,
      "step": 2895
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.59559779819298e-05,
      "loss": 1.2646,
      "step": 2896
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.5927155940796404e-05,
      "loss": 1.2431,
      "step": 2897
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.589833526197944e-05,
      "loss": 1.2379,
      "step": 2898
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.586951595511906e-05,
      "loss": 1.3293,
      "step": 2899
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.5840698029854975e-05,
      "loss": 1.08,
      "step": 2900
    },
    {
      "epoch": 1.59,
      "eval_loss": 1.3435685634613037,
      "eval_runtime": 36.8555,
      "eval_samples_per_second": 3.826,
      "eval_steps_per_second": 1.926,
      "step": 2900
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.581188149582643e-05,
      "loss": 1.2266,
      "step": 2901
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.57830663626722e-05,
      "loss": 1.2217,
      "step": 2902
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.5754252640030596e-05,
      "loss": 1.6397,
      "step": 2903
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.572544033753945e-05,
      "loss": 1.3487,
      "step": 2904
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.5696629464836135e-05,
      "loss": 1.3942,
      "step": 2905
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.566782003155751e-05,
      "loss": 1.3283,
      "step": 2906
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.563901204734e-05,
      "loss": 1.4684,
      "step": 2907
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.561020552181952e-05,
      "loss": 1.5199,
      "step": 2908
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.55814004646315e-05,
      "loss": 1.2982,
      "step": 2909
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.555259688541086e-05,
      "loss": 1.327,
      "step": 2910
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.552379479379207e-05,
      "loss": 0.9579,
      "step": 2911
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.5494994199409067e-05,
      "loss": 1.4348,
      "step": 2912
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.546619511189529e-05,
      "loss": 1.2513,
      "step": 2913
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.543739754088369e-05,
      "loss": 0.9854,
      "step": 2914
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.540860149600669e-05,
      "loss": 0.9938,
      "step": 2915
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.537980698689623e-05,
      "loss": 1.0775,
      "step": 2916
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.5351014023183724e-05,
      "loss": 0.9734,
      "step": 2917
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.532222261450004e-05,
      "loss": 1.1553,
      "step": 2918
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.529343277047557e-05,
      "loss": 1.3816,
      "step": 2919
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.5264644500740156e-05,
      "loss": 1.2522,
      "step": 2920
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.523585781492313e-05,
      "loss": 1.1587,
      "step": 2921
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.5207072722653274e-05,
      "loss": 1.4227,
      "step": 2922
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.5178289233558856e-05,
      "loss": 1.3549,
      "step": 2923
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.51495073572676e-05,
      "loss": 1.1062,
      "step": 2924
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.512072710340669e-05,
      "loss": 1.2059,
      "step": 2925
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.5091948481602773e-05,
      "loss": 1.3321,
      "step": 2926
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.506317150148194e-05,
      "loss": 1.0899,
      "step": 2927
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.503439617266973e-05,
      "loss": 1.4009,
      "step": 2928
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.500562250479117e-05,
      "loss": 1.0929,
      "step": 2929
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.4976850507470655e-05,
      "loss": 1.4673,
      "step": 2930
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.49480801903321e-05,
      "loss": 1.3063,
      "step": 2931
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.49193115629988e-05,
      "loss": 1.1575,
      "step": 2932
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.489054463509352e-05,
      "loss": 1.5072,
      "step": 2933
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.486177941623845e-05,
      "loss": 1.3054,
      "step": 2934
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.4833015916055186e-05,
      "loss": 1.0834,
      "step": 2935
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.4804254144164785e-05,
      "loss": 1.277,
      "step": 2936
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.47754941101877e-05,
      "loss": 0.9579,
      "step": 2937
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.474673582374381e-05,
      "loss": 1.0977,
      "step": 2938
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.47179792944524e-05,
      "loss": 1.4476,
      "step": 2939
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.468922453193219e-05,
      "loss": 1.0711,
      "step": 2940
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.466047154580129e-05,
      "loss": 1.2276,
      "step": 2941
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.463172034567724e-05,
      "loss": 1.3576,
      "step": 2942
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.460297094117692e-05,
      "loss": 1.2506,
      "step": 2943
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.4574223341916695e-05,
      "loss": 1.2181,
      "step": 2944
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.454547755751226e-05,
      "loss": 1.2753,
      "step": 2945
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.451673359757874e-05,
      "loss": 1.293,
      "step": 2946
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.448799147173064e-05,
      "loss": 1.1645,
      "step": 2947
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.445925118958183e-05,
      "loss": 1.2353,
      "step": 2948
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.443051276074561e-05,
      "loss": 1.5015,
      "step": 2949
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.4401776194834613e-05,
      "loss": 1.3871,
      "step": 2950
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.4373041501460875e-05,
      "loss": 1.1062,
      "step": 2951
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.434430869023579e-05,
      "loss": 1.2631,
      "step": 2952
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.4315577770770136e-05,
      "loss": 1.2751,
      "step": 2953
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.4286848752674075e-05,
      "loss": 1.3014,
      "step": 2954
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.425812164555708e-05,
      "loss": 1.2497,
      "step": 2955
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.422939645902803e-05,
      "loss": 1.1118,
      "step": 2956
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.420067320269514e-05,
      "loss": 1.2434,
      "step": 2957
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.4171951886165994e-05,
      "loss": 1.2818,
      "step": 2958
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.414323251904752e-05,
      "loss": 1.1035,
      "step": 2959
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.4114515110945995e-05,
      "loss": 1.3296,
      "step": 2960
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.408579967146704e-05,
      "loss": 1.14,
      "step": 2961
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.405708621021562e-05,
      "loss": 1.0371,
      "step": 2962
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.402837473679602e-05,
      "loss": 1.3307,
      "step": 2963
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.39996652608119e-05,
      "loss": 1.0596,
      "step": 2964
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.397095779186621e-05,
      "loss": 0.9493,
      "step": 2965
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.3942252339561275e-05,
      "loss": 1.2828,
      "step": 2966
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.391354891349868e-05,
      "loss": 1.3389,
      "step": 2967
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.3884847523279376e-05,
      "loss": 1.4592,
      "step": 2968
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.3856148178503635e-05,
      "loss": 1.5247,
      "step": 2969
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.382745088877104e-05,
      "loss": 1.1921,
      "step": 2970
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.379875566368048e-05,
      "loss": 1.1011,
      "step": 2971
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.3770062512830154e-05,
      "loss": 1.1648,
      "step": 2972
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.374137144581757e-05,
      "loss": 0.9964,
      "step": 2973
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.371268247223954e-05,
      "loss": 1.3843,
      "step": 2974
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.368399560169217e-05,
      "loss": 1.4165,
      "step": 2975
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.365531084377087e-05,
      "loss": 1.1144,
      "step": 2976
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.362662820807033e-05,
      "loss": 1.1501,
      "step": 2977
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.3597947704184574e-05,
      "loss": 1.3716,
      "step": 2978
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.356926934170684e-05,
      "loss": 1.343,
      "step": 2979
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.35405931302297e-05,
      "loss": 1.4086,
      "step": 2980
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.3511919079344996e-05,
      "loss": 1.393,
      "step": 2981
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.348324719864385e-05,
      "loss": 1.3639,
      "step": 2982
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.3454577497716654e-05,
      "loss": 1.2571,
      "step": 2983
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.342590998615307e-05,
      "loss": 1.2262,
      "step": 2984
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.339724467354204e-05,
      "loss": 1.3011,
      "step": 2985
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.336858156947174e-05,
      "loss": 1.3449,
      "step": 2986
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.333992068352964e-05,
      "loss": 1.3534,
      "step": 2987
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.331126202530245e-05,
      "loss": 1.2829,
      "step": 2988
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.3282605604376146e-05,
      "loss": 1.253,
      "step": 2989
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.3253951430335945e-05,
      "loss": 1.1948,
      "step": 2990
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.322529951276633e-05,
      "loss": 1.497,
      "step": 2991
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.319664986125099e-05,
      "loss": 1.3572,
      "step": 2992
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.31680024853729e-05,
      "loss": 1.4953,
      "step": 2993
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.3139357394714255e-05,
      "loss": 1.2845,
      "step": 2994
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.311071459885648e-05,
      "loss": 1.2834,
      "step": 2995
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.3082074107380245e-05,
      "loss": 0.964,
      "step": 2996
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.3053435929865444e-05,
      "loss": 1.3418,
      "step": 2997
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.302480007589119e-05,
      "loss": 1.3161,
      "step": 2998
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.2996166555035835e-05,
      "loss": 1.2651,
      "step": 2999
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.2967535376876936e-05,
      "loss": 1.289,
      "step": 3000
    },
    {
      "epoch": 1.64,
      "eval_loss": 1.3379470109939575,
      "eval_runtime": 36.773,
      "eval_samples_per_second": 3.834,
      "eval_steps_per_second": 1.931,
      "step": 3000
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.293890655099127e-05,
      "loss": 1.1154,
      "step": 3001
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.291028008695483e-05,
      "loss": 1.4561,
      "step": 3002
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.288165599434282e-05,
      "loss": 1.2912,
      "step": 3003
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.285303428272964e-05,
      "loss": 1.2749,
      "step": 3004
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.2824414961688914e-05,
      "loss": 1.3127,
      "step": 3005
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.279579804079344e-05,
      "loss": 1.1503,
      "step": 3006
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.276718352961524e-05,
      "loss": 1.2853,
      "step": 3007
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.27385714377255e-05,
      "loss": 1.2052,
      "step": 3008
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.270996177469463e-05,
      "loss": 1.2155,
      "step": 3009
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.2681354550092214e-05,
      "loss": 1.547,
      "step": 3010
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.2652749773486994e-05,
      "loss": 1.2073,
      "step": 3011
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.2624147454446945e-05,
      "loss": 1.3892,
      "step": 3012
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.259554760253919e-05,
      "loss": 1.1076,
      "step": 3013
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.256695022733001e-05,
      "loss": 1.3369,
      "step": 3014
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.2538355338384895e-05,
      "loss": 1.1672,
      "step": 3015
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.2509762945268474e-05,
      "loss": 1.3744,
      "step": 3016
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.2481173057544554e-05,
      "loss": 1.1749,
      "step": 3017
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.24525856847761e-05,
      "loss": 1.0201,
      "step": 3018
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.242400083652525e-05,
      "loss": 1.4579,
      "step": 3019
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.239541852235327e-05,
      "loss": 1.1821,
      "step": 3020
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.236683875182061e-05,
      "loss": 1.1674,
      "step": 3021
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.2338261534486844e-05,
      "loss": 1.3235,
      "step": 3022
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.23096868799107e-05,
      "loss": 1.1849,
      "step": 3023
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.228111479765004e-05,
      "loss": 1.3186,
      "step": 3024
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.2252545297261895e-05,
      "loss": 1.4048,
      "step": 3025
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.2223978388302394e-05,
      "loss": 1.3686,
      "step": 3026
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.2195414080326836e-05,
      "loss": 1.2654,
      "step": 3027
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.2166852382889616e-05,
      "loss": 1.1378,
      "step": 3028
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.213829330554428e-05,
      "loss": 1.3252,
      "step": 3029
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.210973685784349e-05,
      "loss": 1.1346,
      "step": 3030
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.208118304933903e-05,
      "loss": 0.9182,
      "step": 3031
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.205263188958179e-05,
      "loss": 1.4123,
      "step": 3032
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.2024083388121796e-05,
      "loss": 1.366,
      "step": 3033
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.1995537554508156e-05,
      "loss": 1.2217,
      "step": 3034
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.1966994398289125e-05,
      "loss": 1.1588,
      "step": 3035
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.193845392901201e-05,
      "loss": 1.3112,
      "step": 3036
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.190991615622327e-05,
      "loss": 1.511,
      "step": 3037
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.188138108946842e-05,
      "loss": 1.4448,
      "step": 3038
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.185284873829211e-05,
      "loss": 1.1956,
      "step": 3039
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.182431911223805e-05,
      "loss": 1.3575,
      "step": 3040
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.1795792220849064e-05,
      "loss": 1.4071,
      "step": 3041
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.1767268073667026e-05,
      "loss": 1.1457,
      "step": 3042
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.173874668023293e-05,
      "loss": 1.1225,
      "step": 3043
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.171022805008683e-05,
      "loss": 1.0781,
      "step": 3044
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.168171219276785e-05,
      "loss": 1.4665,
      "step": 3045
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.165319911781421e-05,
      "loss": 1.1166,
      "step": 3046
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.162468883476319e-05,
      "loss": 1.2163,
      "step": 3047
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.159618135315109e-05,
      "loss": 1.3239,
      "step": 3048
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.156767668251335e-05,
      "loss": 1.3174,
      "step": 3049
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.153917483238442e-05,
      "loss": 1.3959,
      "step": 3050
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.1510675812297814e-05,
      "loss": 1.4606,
      "step": 3051
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.1482179631786124e-05,
      "loss": 1.1246,
      "step": 3052
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.145368630038096e-05,
      "loss": 1.0761,
      "step": 3053
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.1425195827612986e-05,
      "loss": 1.3334,
      "step": 3054
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.1396708223011935e-05,
      "loss": 1.2337,
      "step": 3055
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.136822349610654e-05,
      "loss": 1.3299,
      "step": 3056
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.1339741656424614e-05,
      "loss": 1.3253,
      "step": 3057
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.1311262713492966e-05,
      "loss": 1.2597,
      "step": 3058
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.128278667683747e-05,
      "loss": 1.1147,
      "step": 3059
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.125431355598299e-05,
      "loss": 1.4207,
      "step": 3060
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.122584336045345e-05,
      "loss": 1.2589,
      "step": 3061
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.119737609977178e-05,
      "loss": 1.203,
      "step": 3062
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.1168911783459935e-05,
      "loss": 1.0854,
      "step": 3063
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.114045042103887e-05,
      "loss": 1.4859,
      "step": 3064
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.111199202202856e-05,
      "loss": 1.4838,
      "step": 3065
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.1083536595948e-05,
      "loss": 1.1221,
      "step": 3066
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.1055084152315174e-05,
      "loss": 1.0407,
      "step": 3067
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.1026634700647084e-05,
      "loss": 1.3232,
      "step": 3068
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.0998188250459725e-05,
      "loss": 1.3656,
      "step": 3069
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.096974481126807e-05,
      "loss": 1.1808,
      "step": 3070
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.094130439258612e-05,
      "loss": 0.9852,
      "step": 3071
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.0912867003926834e-05,
      "loss": 1.4241,
      "step": 3072
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.0884432654802166e-05,
      "loss": 1.2204,
      "step": 3073
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.085600135472307e-05,
      "loss": 1.2366,
      "step": 3074
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.0827573113199456e-05,
      "loss": 1.3921,
      "step": 3075
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.079914793974023e-05,
      "loss": 1.1177,
      "step": 3076
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.077072584385327e-05,
      "loss": 1.0256,
      "step": 3077
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.07423068350454e-05,
      "loss": 1.2731,
      "step": 3078
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.071389092282244e-05,
      "loss": 1.3731,
      "step": 3079
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.068547811668918e-05,
      "loss": 1.2026,
      "step": 3080
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.065706842614933e-05,
      "loss": 1.0625,
      "step": 3081
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.06286618607056e-05,
      "loss": 1.2703,
      "step": 3082
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.060025842985963e-05,
      "loss": 1.3124,
      "step": 3083
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.057185814311203e-05,
      "loss": 1.2272,
      "step": 3084
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.054346100996232e-05,
      "loss": 1.2168,
      "step": 3085
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.0515067039909025e-05,
      "loss": 1.2835,
      "step": 3086
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.0486676242449575e-05,
      "loss": 1.1761,
      "step": 3087
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.045828862708032e-05,
      "loss": 1.175,
      "step": 3088
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.0429904203296596e-05,
      "loss": 0.9039,
      "step": 3089
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.040152298059263e-05,
      "loss": 1.3113,
      "step": 3090
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.037314496846161e-05,
      "loss": 1.2777,
      "step": 3091
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.0344770176395606e-05,
      "loss": 1.3173,
      "step": 3092
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.0316398613885665e-05,
      "loss": 1.0833,
      "step": 3093
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.0288030290421705e-05,
      "loss": 1.2917,
      "step": 3094
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.02596652154926e-05,
      "loss": 1.3052,
      "step": 3095
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.023130339858612e-05,
      "loss": 1.3482,
      "step": 3096
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.020294484918893e-05,
      "loss": 1.5457,
      "step": 3097
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.017458957678663e-05,
      "loss": 1.0574,
      "step": 3098
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.01462375908637e-05,
      "loss": 1.3477,
      "step": 3099
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.011788890090354e-05,
      "loss": 1.1221,
      "step": 3100
    },
    {
      "epoch": 1.7,
      "eval_loss": 1.3383774757385254,
      "eval_runtime": 36.7617,
      "eval_samples_per_second": 3.836,
      "eval_steps_per_second": 1.931,
      "step": 3100
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.0089543516388446e-05,
      "loss": 1.2564,
      "step": 3101
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.0061201446799584e-05,
      "loss": 1.2391,
      "step": 3102
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.003286270161704e-05,
      "loss": 1.1896,
      "step": 3103
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.0004527290319784e-05,
      "loss": 1.236,
      "step": 3104
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.9976195222385646e-05,
      "loss": 1.2754,
      "step": 3105
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.9947866507291364e-05,
      "loss": 1.2341,
      "step": 3106
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.9919541154512554e-05,
      "loss": 1.3315,
      "step": 3107
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.98912191735237e-05,
      "loss": 1.4239,
      "step": 3108
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.986290057379814e-05,
      "loss": 1.4841,
      "step": 3109
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.9834585364808104e-05,
      "loss": 0.9869,
      "step": 3110
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.980627355602469e-05,
      "loss": 1.3603,
      "step": 3111
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.977796515691785e-05,
      "loss": 1.4627,
      "step": 3112
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.974966017695639e-05,
      "loss": 1.3346,
      "step": 3113
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.9721358625607985e-05,
      "loss": 1.2303,
      "step": 3114
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.969306051233916e-05,
      "loss": 1.4493,
      "step": 3115
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.966476584661528e-05,
      "loss": 1.246,
      "step": 3116
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.963647463790057e-05,
      "loss": 1.0168,
      "step": 3117
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.960818689565808e-05,
      "loss": 1.201,
      "step": 3118
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.957990262934973e-05,
      "loss": 1.148,
      "step": 3119
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.955162184843625e-05,
      "loss": 1.2462,
      "step": 3120
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.952334456237723e-05,
      "loss": 0.9312,
      "step": 3121
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.949507078063105e-05,
      "loss": 1.1767,
      "step": 3122
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.9466800512654954e-05,
      "loss": 1.1158,
      "step": 3123
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.943853376790501e-05,
      "loss": 1.2296,
      "step": 3124
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.941027055583608e-05,
      "loss": 1.0388,
      "step": 3125
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.9382010885901884e-05,
      "loss": 1.1804,
      "step": 3126
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.935375476755492e-05,
      "loss": 1.2937,
      "step": 3127
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.9325502210246514e-05,
      "loss": 1.1973,
      "step": 3128
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.9297253223426806e-05,
      "loss": 1.3899,
      "step": 3129
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.9269007816544734e-05,
      "loss": 1.2472,
      "step": 3130
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.9240765999048036e-05,
      "loss": 1.4554,
      "step": 3131
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.921252778038326e-05,
      "loss": 1.1013,
      "step": 3132
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.918429316999575e-05,
      "loss": 1.1368,
      "step": 3133
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.915606217732962e-05,
      "loss": 1.2772,
      "step": 3134
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.91278348118278e-05,
      "loss": 1.6129,
      "step": 3135
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.9099611082932e-05,
      "loss": 1.3867,
      "step": 3136
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.90713910000827e-05,
      "loss": 1.3776,
      "step": 3137
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.9043174572719194e-05,
      "loss": 0.8934,
      "step": 3138
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.901496181027951e-05,
      "loss": 1.1734,
      "step": 3139
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.8986752722200484e-05,
      "loss": 1.2233,
      "step": 3140
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.89585473179177e-05,
      "loss": 1.3494,
      "step": 3141
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.893034560686553e-05,
      "loss": 1.2584,
      "step": 3142
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.8902147598477095e-05,
      "loss": 1.4586,
      "step": 3143
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.887395330218429e-05,
      "loss": 1.0643,
      "step": 3144
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.8845762727417756e-05,
      "loss": 1.315,
      "step": 3145
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.881757588360689e-05,
      "loss": 1.0234,
      "step": 3146
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.878939278017986e-05,
      "loss": 1.3921,
      "step": 3147
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.876121342656355e-05,
      "loss": 1.4616,
      "step": 3148
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.873303783218362e-05,
      "loss": 1.1834,
      "step": 3149
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.870486600646445e-05,
      "loss": 1.2496,
      "step": 3150
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.867669795882919e-05,
      "loss": 1.4284,
      "step": 3151
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.8648533698699695e-05,
      "loss": 1.1631,
      "step": 3152
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.8620373235496565e-05,
      "loss": 1.1284,
      "step": 3153
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.859221657863913e-05,
      "loss": 1.0465,
      "step": 3154
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.856406373754545e-05,
      "loss": 1.0465,
      "step": 3155
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.8535914721632304e-05,
      "loss": 1.4037,
      "step": 3156
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.8507769540315195e-05,
      "loss": 1.0699,
      "step": 3157
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.847962820300834e-05,
      "loss": 1.0739,
      "step": 3158
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.845149071912466e-05,
      "loss": 1.3957,
      "step": 3159
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.8423357098075815e-05,
      "loss": 1.2948,
      "step": 3160
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.839522734927215e-05,
      "loss": 1.1541,
      "step": 3161
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.836710148212271e-05,
      "loss": 1.1617,
      "step": 3162
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.8338979506035266e-05,
      "loss": 1.2586,
      "step": 3163
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.831086143041628e-05,
      "loss": 1.1928,
      "step": 3164
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.8282747264670895e-05,
      "loss": 1.0721,
      "step": 3165
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.825463701820295e-05,
      "loss": 1.3625,
      "step": 3166
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.8226530700414986e-05,
      "loss": 1.1318,
      "step": 3167
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.8198428320708216e-05,
      "loss": 1.3134,
      "step": 3168
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.8170329888482554e-05,
      "loss": 1.4809,
      "step": 3169
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.814223541313657e-05,
      "loss": 1.1603,
      "step": 3170
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.8114144904067526e-05,
      "loss": 1.1996,
      "step": 3171
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.8086058370671355e-05,
      "loss": 1.2022,
      "step": 3172
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.805797582234264e-05,
      "loss": 1.1998,
      "step": 3173
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.802989726847467e-05,
      "loss": 1.3318,
      "step": 3174
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.800182271845937e-05,
      "loss": 1.2553,
      "step": 3175
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.7973752181687335e-05,
      "loss": 0.9224,
      "step": 3176
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.794568566754781e-05,
      "loss": 1.4917,
      "step": 3177
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.7917623185428706e-05,
      "loss": 1.1969,
      "step": 3178
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.7889564744716584e-05,
      "loss": 1.3726,
      "step": 3179
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.7861510354796634e-05,
      "loss": 1.4589,
      "step": 3180
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.783346002505271e-05,
      "loss": 1.4572,
      "step": 3181
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.780541376486731e-05,
      "loss": 1.2989,
      "step": 3182
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.7777371583621566e-05,
      "loss": 1.0356,
      "step": 3183
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.774933349069524e-05,
      "loss": 1.1585,
      "step": 3184
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.772129949546673e-05,
      "loss": 1.1216,
      "step": 3185
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.7693269607313064e-05,
      "loss": 1.0825,
      "step": 3186
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.76652438356099e-05,
      "loss": 1.5082,
      "step": 3187
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.7637222189731504e-05,
      "loss": 1.4168,
      "step": 3188
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.760920467905079e-05,
      "loss": 1.1151,
      "step": 3189
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.758119131293926e-05,
      "loss": 1.3464,
      "step": 3190
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.755318210076704e-05,
      "loss": 1.3071,
      "step": 3191
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.7525177051902874e-05,
      "loss": 1.2375,
      "step": 3192
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.7497176175714097e-05,
      "loss": 1.3197,
      "step": 3193
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.746917948156668e-05,
      "loss": 1.5814,
      "step": 3194
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.744118697882516e-05,
      "loss": 1.3478,
      "step": 3195
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.741319867685268e-05,
      "loss": 1.6568,
      "step": 3196
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.7385214585011e-05,
      "loss": 1.3391,
      "step": 3197
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.735723471266044e-05,
      "loss": 1.3026,
      "step": 3198
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.732925906915994e-05,
      "loss": 1.3283,
      "step": 3199
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.7301287663867005e-05,
      "loss": 1.1895,
      "step": 3200
    },
    {
      "epoch": 1.75,
      "eval_loss": 1.3376129865646362,
      "eval_runtime": 36.7958,
      "eval_samples_per_second": 3.832,
      "eval_steps_per_second": 1.93,
      "step": 3200
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.727332050613772e-05,
      "loss": 1.2586,
      "step": 3201
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.7245357605326766e-05,
      "loss": 1.1842,
      "step": 3202
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.7217398970787376e-05,
      "loss": 1.1799,
      "step": 3203
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.718944461187138e-05,
      "loss": 1.3418,
      "step": 3204
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.7161494537929175e-05,
      "loss": 1.1487,
      "step": 3205
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.713354875830971e-05,
      "loss": 1.2657,
      "step": 3206
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.7105607282360505e-05,
      "loss": 1.2037,
      "step": 3207
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.7077670119427645e-05,
      "loss": 1.0562,
      "step": 3208
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.704973727885577e-05,
      "loss": 1.1634,
      "step": 3209
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.7021808769988064e-05,
      "loss": 1.6461,
      "step": 3210
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.699388460216629e-05,
      "loss": 1.1577,
      "step": 3211
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.696596478473072e-05,
      "loss": 1.0646,
      "step": 3212
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.6938049327020195e-05,
      "loss": 1.3098,
      "step": 3213
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.691013823837211e-05,
      "loss": 1.3013,
      "step": 3214
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.6882231528122355e-05,
      "loss": 1.2911,
      "step": 3215
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.68543292056054e-05,
      "loss": 1.2013,
      "step": 3216
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.6826431280154225e-05,
      "loss": 1.1832,
      "step": 3217
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.679853776110035e-05,
      "loss": 1.4301,
      "step": 3218
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.6770648657773804e-05,
      "loss": 1.1625,
      "step": 3219
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.6742763979503156e-05,
      "loss": 1.2811,
      "step": 3220
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.6714883735615504e-05,
      "loss": 1.405,
      "step": 3221
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.6687007935436416e-05,
      "loss": 1.2789,
      "step": 3222
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.665913658829003e-05,
      "loss": 1.4223,
      "step": 3223
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.663126970349897e-05,
      "loss": 1.15,
      "step": 3224
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.6603407290384357e-05,
      "loss": 1.5526,
      "step": 3225
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.6575549358265836e-05,
      "loss": 1.4332,
      "step": 3226
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.654769591646153e-05,
      "loss": 1.2603,
      "step": 3227
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.6519846974288095e-05,
      "loss": 1.4697,
      "step": 3228
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.649200254106064e-05,
      "loss": 1.1929,
      "step": 3229
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.646416262609279e-05,
      "loss": 1.3525,
      "step": 3230
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.6436327238696665e-05,
      "loss": 1.2545,
      "step": 3231
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.640849638818286e-05,
      "loss": 1.1714,
      "step": 3232
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.638067008386045e-05,
      "loss": 1.2685,
      "step": 3233
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.635284833503699e-05,
      "loss": 1.2941,
      "step": 3234
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.6325031151018514e-05,
      "loss": 1.122,
      "step": 3235
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.6297218541109534e-05,
      "loss": 0.8738,
      "step": 3236
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.626941051461302e-05,
      "loss": 0.8591,
      "step": 3237
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.6241607080830445e-05,
      "loss": 1.2897,
      "step": 3238
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.6213808249061674e-05,
      "loss": 1.4269,
      "step": 3239
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.6186014028605096e-05,
      "loss": 1.2029,
      "step": 3240
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.6158224428757535e-05,
      "loss": 1.2939,
      "step": 3241
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.613043945881428e-05,
      "loss": 1.0381,
      "step": 3242
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.610265912806904e-05,
      "loss": 1.2588,
      "step": 3243
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.607488344581402e-05,
      "loss": 1.3094,
      "step": 3244
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.604711242133984e-05,
      "loss": 1.0613,
      "step": 3245
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.601934606393555e-05,
      "loss": 1.1073,
      "step": 3246
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.599158438288868e-05,
      "loss": 1.1744,
      "step": 3247
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.596382738748516e-05,
      "loss": 1.2426,
      "step": 3248
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.593607508700937e-05,
      "loss": 1.1868,
      "step": 3249
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.5908327490744114e-05,
      "loss": 1.3265,
      "step": 3250
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.5880584607970616e-05,
      "loss": 1.1529,
      "step": 3251
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.585284644796852e-05,
      "loss": 0.9403,
      "step": 3252
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.582511302001593e-05,
      "loss": 1.309,
      "step": 3253
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.5797384333389315e-05,
      "loss": 1.207,
      "step": 3254
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.5769660397363594e-05,
      "loss": 1.2341,
      "step": 3255
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.574194122121207e-05,
      "loss": 1.435,
      "step": 3256
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.5714226814206486e-05,
      "loss": 1.2623,
      "step": 3257
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.5686517185616955e-05,
      "loss": 1.3469,
      "step": 3258
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.565881234471201e-05,
      "loss": 1.148,
      "step": 3259
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.5631112300758595e-05,
      "loss": 1.2537,
      "step": 3260
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.5603417063022014e-05,
      "loss": 1.2531,
      "step": 3261
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.557572664076601e-05,
      "loss": 1.0678,
      "step": 3262
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.554804104325268e-05,
      "loss": 1.202,
      "step": 3263
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.55203602797425e-05,
      "loss": 1.3441,
      "step": 3264
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.549268435949436e-05,
      "loss": 1.2958,
      "step": 3265
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.5465013291765515e-05,
      "loss": 1.3138,
      "step": 3266
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.54373470858116e-05,
      "loss": 1.5203,
      "step": 3267
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.5409685750886623e-05,
      "loss": 0.9429,
      "step": 3268
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.538202929624296e-05,
      "loss": 1.238,
      "step": 3269
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.535437773113135e-05,
      "loss": 1.3681,
      "step": 3270
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.532673106480091e-05,
      "loss": 1.221,
      "step": 3271
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.52990893064991e-05,
      "loss": 1.0855,
      "step": 3272
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.527145246547177e-05,
      "loss": 1.2824,
      "step": 3273
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.524382055096309e-05,
      "loss": 1.3791,
      "step": 3274
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.5216193572215605e-05,
      "loss": 1.3207,
      "step": 3275
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.5188571538470186e-05,
      "loss": 1.0373,
      "step": 3276
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.5160954458966075e-05,
      "loss": 1.3342,
      "step": 3277
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.513334234294085e-05,
      "loss": 1.1789,
      "step": 3278
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.510573519963042e-05,
      "loss": 1.3081,
      "step": 3279
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.507813303826903e-05,
      "loss": 1.3497,
      "step": 3280
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.505053586808929e-05,
      "loss": 0.891,
      "step": 3281
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.502294369832208e-05,
      "loss": 1.1738,
      "step": 3282
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.4995356538196664e-05,
      "loss": 1.3494,
      "step": 3283
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.496777439694061e-05,
      "loss": 1.1181,
      "step": 3284
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.4940197283779795e-05,
      "loss": 1.3094,
      "step": 3285
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.4912625207938433e-05,
      "loss": 1.5083,
      "step": 3286
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.488505817863905e-05,
      "loss": 1.488,
      "step": 3287
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.4857496205102474e-05,
      "loss": 1.1563,
      "step": 3288
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.4829939296547844e-05,
      "loss": 1.2159,
      "step": 3289
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.48023874621926e-05,
      "loss": 1.2329,
      "step": 3290
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.477484071125252e-05,
      "loss": 1.3283,
      "step": 3291
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.474729905294163e-05,
      "loss": 1.3352,
      "step": 3292
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4719762496472284e-05,
      "loss": 1.4482,
      "step": 3293
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4692231051055125e-05,
      "loss": 1.1056,
      "step": 3294
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.466470472589909e-05,
      "loss": 1.4978,
      "step": 3295
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.463718353021138e-05,
      "loss": 1.2936,
      "step": 3296
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.460966747319752e-05,
      "loss": 1.3208,
      "step": 3297
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.458215656406128e-05,
      "loss": 1.2561,
      "step": 3298
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.455465081200472e-05,
      "loss": 1.1821,
      "step": 3299
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.45271502262282e-05,
      "loss": 1.3138,
      "step": 3300
    },
    {
      "epoch": 1.81,
      "eval_loss": 1.3358386754989624,
      "eval_runtime": 36.8202,
      "eval_samples_per_second": 3.829,
      "eval_steps_per_second": 1.928,
      "step": 3300
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4499654815930296e-05,
      "loss": 1.0988,
      "step": 3301
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.447216459030789e-05,
      "loss": 1.5471,
      "step": 3302
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4444679558556146e-05,
      "loss": 1.0088,
      "step": 3303
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.441719972986846e-05,
      "loss": 1.3355,
      "step": 3304
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.43897251134365e-05,
      "loss": 1.3677,
      "step": 3305
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.436225571845017e-05,
      "loss": 1.4789,
      "step": 3306
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4334791554097686e-05,
      "loss": 1.2445,
      "step": 3307
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4307332629565434e-05,
      "loss": 1.2823,
      "step": 3308
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4279878954038115e-05,
      "loss": 1.298,
      "step": 3309
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.4252430536698635e-05,
      "loss": 1.3202,
      "step": 3310
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.4224987386728155e-05,
      "loss": 0.9671,
      "step": 3311
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.419754951330608e-05,
      "loss": 1.5205,
      "step": 3312
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.4170116925610016e-05,
      "loss": 1.38,
      "step": 3313
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.414268963281586e-05,
      "loss": 1.4597,
      "step": 3314
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.411526764409769e-05,
      "loss": 1.1895,
      "step": 3315
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.408785096862782e-05,
      "loss": 1.3836,
      "step": 3316
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.40604396155768e-05,
      "loss": 1.4234,
      "step": 3317
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.4033033594113396e-05,
      "loss": 1.3485,
      "step": 3318
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.400563291340457e-05,
      "loss": 1.3661,
      "step": 3319
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.397823758261553e-05,
      "loss": 1.1371,
      "step": 3320
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.3950847610909674e-05,
      "loss": 1.1476,
      "step": 3321
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.392346300744861e-05,
      "loss": 1.5735,
      "step": 3322
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.389608378139215e-05,
      "loss": 1.108,
      "step": 3323
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.386870994189833e-05,
      "loss": 1.2148,
      "step": 3324
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.384134149812333e-05,
      "loss": 1.251,
      "step": 3325
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.3813978459221574e-05,
      "loss": 1.1957,
      "step": 3326
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.3786620834345675e-05,
      "loss": 1.1347,
      "step": 3327
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.37592686326464e-05,
      "loss": 1.2395,
      "step": 3328
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.373192186327274e-05,
      "loss": 1.3434,
      "step": 3329
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.370458053537186e-05,
      "loss": 1.2347,
      "step": 3330
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.367724465808909e-05,
      "loss": 1.339,
      "step": 3331
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.3649914240567944e-05,
      "loss": 1.5207,
      "step": 3332
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.3622589291950114e-05,
      "loss": 1.0989,
      "step": 3333
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.359526982137546e-05,
      "loss": 1.2257,
      "step": 3334
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.356795583798202e-05,
      "loss": 1.3859,
      "step": 3335
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.354064735090599e-05,
      "loss": 1.2777,
      "step": 3336
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.351334436928169e-05,
      "loss": 1.4594,
      "step": 3337
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.3486046902241664e-05,
      "loss": 1.5032,
      "step": 3338
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.345875495891657e-05,
      "loss": 1.3801,
      "step": 3339
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.343146854843523e-05,
      "loss": 1.2504,
      "step": 3340
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.340418767992461e-05,
      "loss": 1.0136,
      "step": 3341
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.3376912362509835e-05,
      "loss": 1.3501,
      "step": 3342
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.334964260531416e-05,
      "loss": 1.2774,
      "step": 3343
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.332237841745898e-05,
      "loss": 1.1263,
      "step": 3344
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.329511980806384e-05,
      "loss": 1.0858,
      "step": 3345
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.32678667862464e-05,
      "loss": 1.1712,
      "step": 3346
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.324061936112248e-05,
      "loss": 1.4998,
      "step": 3347
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.3213377541805993e-05,
      "loss": 1.2209,
      "step": 3348
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.318614133740901e-05,
      "loss": 1.0514,
      "step": 3349
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.3158910757041686e-05,
      "loss": 1.1654,
      "step": 3350
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.3131685809812306e-05,
      "loss": 1.3845,
      "step": 3351
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.310446650482732e-05,
      "loss": 1.0781,
      "step": 3352
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.307725285119124e-05,
      "loss": 1.1616,
      "step": 3353
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.30500448580067e-05,
      "loss": 1.2329,
      "step": 3354
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.302284253437442e-05,
      "loss": 1.5758,
      "step": 3355
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.299564588939328e-05,
      "loss": 1.2508,
      "step": 3356
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.296845493216019e-05,
      "loss": 1.35,
      "step": 3357
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.294126967177019e-05,
      "loss": 1.271,
      "step": 3358
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.2914090117316445e-05,
      "loss": 1.2428,
      "step": 3359
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.288691627789017e-05,
      "loss": 1.357,
      "step": 3360
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.2859748162580674e-05,
      "loss": 1.1828,
      "step": 3361
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.283258578047537e-05,
      "loss": 1.4365,
      "step": 3362
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.280542914065973e-05,
      "loss": 1.2081,
      "step": 3363
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.277827825221733e-05,
      "loss": 0.9236,
      "step": 3364
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.2751133124229805e-05,
      "loss": 1.2177,
      "step": 3365
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.272399376577686e-05,
      "loss": 1.3477,
      "step": 3366
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.269686018593629e-05,
      "loss": 0.9967,
      "step": 3367
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.266973239378394e-05,
      "loss": 1.3536,
      "step": 3368
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.264261039839372e-05,
      "loss": 1.2722,
      "step": 3369
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.26154942088376e-05,
      "loss": 1.2836,
      "step": 3370
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.2588383834185634e-05,
      "loss": 1.2201,
      "step": 3371
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.2561279283505883e-05,
      "loss": 1.4871,
      "step": 3372
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.253418056586451e-05,
      "loss": 1.3418,
      "step": 3373
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.2507087690325696e-05,
      "loss": 1.1945,
      "step": 3374
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.248000066595167e-05,
      "loss": 1.1309,
      "step": 3375
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.2452919501802715e-05,
      "loss": 1.3089,
      "step": 3376
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.242584420693715e-05,
      "loss": 1.4167,
      "step": 3377
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.239877479041133e-05,
      "loss": 1.1974,
      "step": 3378
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.2371711261279636e-05,
      "loss": 0.9771,
      "step": 3379
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.234465362859451e-05,
      "loss": 1.1856,
      "step": 3380
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.231760190140636e-05,
      "loss": 1.3796,
      "step": 3381
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.229055608876368e-05,
      "loss": 1.2761,
      "step": 3382
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.2263516199712965e-05,
      "loss": 1.2916,
      "step": 3383
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.2236482243298714e-05,
      "loss": 1.4006,
      "step": 3384
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.220945422856346e-05,
      "loss": 1.1628,
      "step": 3385
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.218243216454775e-05,
      "loss": 1.2588,
      "step": 3386
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.2155416060290136e-05,
      "loss": 1.1483,
      "step": 3387
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.212840592482715e-05,
      "loss": 1.3044,
      "step": 3388
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.210140176719337e-05,
      "loss": 1.311,
      "step": 3389
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.207440359642136e-05,
      "loss": 1.292,
      "step": 3390
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.2047411421541665e-05,
      "loss": 1.3173,
      "step": 3391
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.2020425251582844e-05,
      "loss": 1.3461,
      "step": 3392
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.1993445095571447e-05,
      "loss": 1.2597,
      "step": 3393
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.196647096253198e-05,
      "loss": 1.2306,
      "step": 3394
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.193950286148698e-05,
      "loss": 1.1164,
      "step": 3395
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.191254080145695e-05,
      "loss": 1.2277,
      "step": 3396
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.188558479146036e-05,
      "loss": 1.6038,
      "step": 3397
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.185863484051365e-05,
      "loss": 0.9354,
      "step": 3398
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.1831690957631274e-05,
      "loss": 1.1607,
      "step": 3399
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.180475315182563e-05,
      "loss": 1.3907,
      "step": 3400
    },
    {
      "epoch": 1.86,
      "eval_loss": 1.3342907428741455,
      "eval_runtime": 36.8011,
      "eval_samples_per_second": 3.831,
      "eval_steps_per_second": 1.929,
      "step": 3400
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.1777821432107065e-05,
      "loss": 1.0752,
      "step": 3401
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1750895807483916e-05,
      "loss": 1.1202,
      "step": 3402
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.172397628696249e-05,
      "loss": 1.4083,
      "step": 3403
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.169706287954701e-05,
      "loss": 1.0404,
      "step": 3404
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1670155594239716e-05,
      "loss": 1.3783,
      "step": 3405
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1643254440040734e-05,
      "loss": 1.2018,
      "step": 3406
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1616359425948175e-05,
      "loss": 1.4487,
      "step": 3407
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1589470560958104e-05,
      "loss": 1.1263,
      "step": 3408
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1562587854064506e-05,
      "loss": 1.0039,
      "step": 3409
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.153571131425932e-05,
      "loss": 1.1391,
      "step": 3410
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1508840950532427e-05,
      "loss": 1.1577,
      "step": 3411
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.148197677187162e-05,
      "loss": 1.2178,
      "step": 3412
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.145511878726265e-05,
      "loss": 0.899,
      "step": 3413
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.142826700568918e-05,
      "loss": 1.3948,
      "step": 3414
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.14014214361328e-05,
      "loss": 1.4254,
      "step": 3415
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.137458208757302e-05,
      "loss": 1.3558,
      "step": 3416
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1347748968987304e-05,
      "loss": 1.3415,
      "step": 3417
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.1320922089350945e-05,
      "loss": 1.3341,
      "step": 3418
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.129410145763725e-05,
      "loss": 1.2479,
      "step": 3419
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.1267287082817375e-05,
      "loss": 1.2799,
      "step": 3420
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.1240478973860397e-05,
      "loss": 1.4928,
      "step": 3421
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.121367713973331e-05,
      "loss": 1.3754,
      "step": 3422
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.1186881589401e-05,
      "loss": 1.138,
      "step": 3423
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.116009233182623e-05,
      "loss": 1.1803,
      "step": 3424
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.11333093759697e-05,
      "loss": 1.2758,
      "step": 3425
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.1106532730789986e-05,
      "loss": 1.061,
      "step": 3426
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.1079762405243516e-05,
      "loss": 1.0138,
      "step": 3427
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.105299840828466e-05,
      "loss": 1.1842,
      "step": 3428
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.1026240748865665e-05,
      "loss": 1.3752,
      "step": 3429
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.099948943593659e-05,
      "loss": 1.4015,
      "step": 3430
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.097274447844546e-05,
      "loss": 1.1013,
      "step": 3431
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.0946005885338113e-05,
      "loss": 1.4623,
      "step": 3432
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.09192736655583e-05,
      "loss": 1.2016,
      "step": 3433
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.0892547828047616e-05,
      "loss": 1.1697,
      "step": 3434
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.086582838174551e-05,
      "loss": 1.0282,
      "step": 3435
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.083911533558933e-05,
      "loss": 1.2864,
      "step": 3436
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.081240869851426e-05,
      "loss": 1.1319,
      "step": 3437
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.078570847945334e-05,
      "loss": 1.3829,
      "step": 3438
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.075901468733745e-05,
      "loss": 1.1995,
      "step": 3439
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.073232733109536e-05,
      "loss": 1.2414,
      "step": 3440
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.070564641965364e-05,
      "loss": 1.253,
      "step": 3441
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.0678971961936764e-05,
      "loss": 1.3171,
      "step": 3442
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.0652303966866974e-05,
      "loss": 1.2901,
      "step": 3443
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.06256424433644e-05,
      "loss": 1.2264,
      "step": 3444
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.0598987400347e-05,
      "loss": 1.3412,
      "step": 3445
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.0572338846730566e-05,
      "loss": 1.4925,
      "step": 3446
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.0545696791428694e-05,
      "loss": 1.3166,
      "step": 3447
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.0519061243352834e-05,
      "loss": 1.2658,
      "step": 3448
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.0492432211412248e-05,
      "loss": 1.2787,
      "step": 3449
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.046580970451402e-05,
      "loss": 1.269,
      "step": 3450
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.043919373156306e-05,
      "loss": 1.2951,
      "step": 3451
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.0412584301462076e-05,
      "loss": 1.4223,
      "step": 3452
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.038598142311161e-05,
      "loss": 1.3549,
      "step": 3453
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.0359385105409993e-05,
      "loss": 1.3273,
      "step": 3454
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.0332795357253352e-05,
      "loss": 1.2686,
      "step": 3455
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.0306212187535653e-05,
      "loss": 1.1258,
      "step": 3456
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.027963560514862e-05,
      "loss": 1.4369,
      "step": 3457
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.0253065618981812e-05,
      "loss": 1.322,
      "step": 3458
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.022650223792255e-05,
      "loss": 1.034,
      "step": 3459
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.019994547085597e-05,
      "loss": 1.1433,
      "step": 3460
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.017339532666499e-05,
      "loss": 1.3058,
      "step": 3461
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.0146851814230285e-05,
      "loss": 1.2634,
      "step": 3462
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.012031494243036e-05,
      "loss": 1.6033,
      "step": 3463
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.0093784720141455e-05,
      "loss": 1.2333,
      "step": 3464
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.0067261156237615e-05,
      "loss": 1.1011,
      "step": 3465
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.0040744259590654e-05,
      "loss": 1.1601,
      "step": 3466
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.0014234039070128e-05,
      "loss": 1.0856,
      "step": 3467
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.9987730503543387e-05,
      "loss": 1.3227,
      "step": 3468
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.9961233661875547e-05,
      "loss": 1.3744,
      "step": 3469
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.9934743522929476e-05,
      "loss": 1.3004,
      "step": 3470
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.9908260095565798e-05,
      "loss": 1.3918,
      "step": 3471
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.9881783388642893e-05,
      "loss": 1.3044,
      "step": 3472
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.9855313411016894e-05,
      "loss": 1.2323,
      "step": 3473
    },
    {
      "epoch": 1.9,
      "learning_rate": 2.982885017154169e-05,
      "loss": 1.1998,
      "step": 3474
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9802393679068908e-05,
      "loss": 1.2658,
      "step": 3475
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9775943942447914e-05,
      "loss": 1.2864,
      "step": 3476
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9749500970525835e-05,
      "loss": 1.1781,
      "step": 3477
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.972306477214751e-05,
      "loss": 1.4527,
      "step": 3478
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9696635356155533e-05,
      "loss": 1.2501,
      "step": 3479
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.96702127313902e-05,
      "loss": 0.9683,
      "step": 3480
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.964379690668957e-05,
      "loss": 1.1842,
      "step": 3481
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9617387890889413e-05,
      "loss": 1.3792,
      "step": 3482
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9590985692823224e-05,
      "loss": 1.3393,
      "step": 3483
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9564590321322207e-05,
      "loss": 1.0773,
      "step": 3484
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.953820178521528e-05,
      "loss": 1.235,
      "step": 3485
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.951182009332911e-05,
      "loss": 1.0678,
      "step": 3486
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9485445254488037e-05,
      "loss": 1.1547,
      "step": 3487
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.945907727751412e-05,
      "loss": 1.3451,
      "step": 3488
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9432716171227125e-05,
      "loss": 1.0448,
      "step": 3489
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9406361944444515e-05,
      "loss": 1.2397,
      "step": 3490
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9380014605981477e-05,
      "loss": 1.4267,
      "step": 3491
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.9353674164650847e-05,
      "loss": 1.258,
      "step": 3492
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.9327340629263188e-05,
      "loss": 1.268,
      "step": 3493
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.9301014008626747e-05,
      "loss": 1.1578,
      "step": 3494
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.9274694311547456e-05,
      "loss": 1.246,
      "step": 3495
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.924838154682893e-05,
      "loss": 1.1732,
      "step": 3496
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.9222075723272456e-05,
      "loss": 1.2822,
      "step": 3497
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.9195776849677035e-05,
      "loss": 1.2922,
      "step": 3498
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.916948493483931e-05,
      "loss": 1.1366,
      "step": 3499
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.914319998755357e-05,
      "loss": 1.4544,
      "step": 3500
    },
    {
      "epoch": 1.92,
      "eval_loss": 1.3351472616195679,
      "eval_runtime": 36.7663,
      "eval_samples_per_second": 3.835,
      "eval_steps_per_second": 1.931,
      "step": 3500
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.9116922016611864e-05,
      "loss": 1.0575,
      "step": 3501
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.9090651030803795e-05,
      "loss": 1.2248,
      "step": 3502
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.906438703891674e-05,
      "loss": 1.2227,
      "step": 3503
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.9038130049735634e-05,
      "loss": 1.2088,
      "step": 3504
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.901188007204316e-05,
      "loss": 1.0942,
      "step": 3505
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.8985637114619567e-05,
      "loss": 1.1032,
      "step": 3506
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.895940118624286e-05,
      "loss": 1.362,
      "step": 3507
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.8933172295688576e-05,
      "loss": 1.3046,
      "step": 3508
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.890695045173001e-05,
      "loss": 1.2255,
      "step": 3509
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.8880735663138024e-05,
      "loss": 1.3233,
      "step": 3510
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.885452793868112e-05,
      "loss": 1.0636,
      "step": 3511
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.882832728712551e-05,
      "loss": 1.2503,
      "step": 3512
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.880213371723494e-05,
      "loss": 1.0948,
      "step": 3513
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8775947237770885e-05,
      "loss": 1.2092,
      "step": 3514
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.874976785749236e-05,
      "loss": 1.2329,
      "step": 3515
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8723595585156083e-05,
      "loss": 1.3678,
      "step": 3516
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8697430429516326e-05,
      "loss": 1.2832,
      "step": 3517
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8671272399325044e-05,
      "loss": 1.3599,
      "step": 3518
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8645121503331747e-05,
      "loss": 1.3721,
      "step": 3519
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8618977750283603e-05,
      "loss": 1.2958,
      "step": 3520
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8592841148925398e-05,
      "loss": 1.15,
      "step": 3521
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.856671170799946e-05,
      "loss": 1.1917,
      "step": 3522
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.854058943624582e-05,
      "loss": 1.3024,
      "step": 3523
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8514474342402005e-05,
      "loss": 1.1975,
      "step": 3524
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8488366435203238e-05,
      "loss": 1.1866,
      "step": 3525
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8462265723382252e-05,
      "loss": 1.081,
      "step": 3526
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8436172215669466e-05,
      "loss": 1.4095,
      "step": 3527
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8410085920792807e-05,
      "loss": 1.0715,
      "step": 3528
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.8384006847477807e-05,
      "loss": 1.3792,
      "step": 3529
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.835793500444762e-05,
      "loss": 1.2184,
      "step": 3530
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8331870400422944e-05,
      "loss": 1.3277,
      "step": 3531
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8305813044122097e-05,
      "loss": 1.3865,
      "step": 3532
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8279762944260907e-05,
      "loss": 1.3476,
      "step": 3533
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.825372010955285e-05,
      "loss": 1.3082,
      "step": 3534
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8227684548708892e-05,
      "loss": 1.1613,
      "step": 3535
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8201656270437658e-05,
      "loss": 0.8688,
      "step": 3536
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8175635283445245e-05,
      "loss": 1.4176,
      "step": 3537
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8149621596435393e-05,
      "loss": 1.0274,
      "step": 3538
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8123615218109324e-05,
      "loss": 1.0572,
      "step": 3539
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8097616157165883e-05,
      "loss": 1.2262,
      "step": 3540
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8071624422301428e-05,
      "loss": 1.2336,
      "step": 3541
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.804564002220986e-05,
      "loss": 1.0348,
      "step": 3542
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.8019662965582667e-05,
      "loss": 1.0877,
      "step": 3543
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.7993693261108823e-05,
      "loss": 1.1521,
      "step": 3544
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.7967730917474923e-05,
      "loss": 1.2461,
      "step": 3545
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.7941775943365007e-05,
      "loss": 1.3968,
      "step": 3546
    },
    {
      "epoch": 1.94,
      "learning_rate": 2.7915828347460733e-05,
      "loss": 1.3608,
      "step": 3547
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7889888138441212e-05,
      "loss": 0.9795,
      "step": 3548
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.786395532498318e-05,
      "loss": 1.3424,
      "step": 3549
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7838029915760787e-05,
      "loss": 1.3088,
      "step": 3550
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7812111919445822e-05,
      "loss": 1.1041,
      "step": 3551
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7786201344707486e-05,
      "loss": 1.4716,
      "step": 3552
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7760298200212585e-05,
      "loss": 1.1012,
      "step": 3553
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7734402494625393e-05,
      "loss": 1.0106,
      "step": 3554
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7708514236607685e-05,
      "loss": 1.3126,
      "step": 3555
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7682633434818805e-05,
      "loss": 1.2449,
      "step": 3556
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7656760097915524e-05,
      "loss": 1.2593,
      "step": 3557
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7630894234552185e-05,
      "loss": 1.3971,
      "step": 3558
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7605035853380585e-05,
      "loss": 1.2196,
      "step": 3559
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7579184963050052e-05,
      "loss": 1.3856,
      "step": 3560
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.755334157220737e-05,
      "loss": 1.1626,
      "step": 3561
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7527505689496868e-05,
      "loss": 1.2455,
      "step": 3562
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7501677323560292e-05,
      "loss": 1.2488,
      "step": 3563
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.7475856483036966e-05,
      "loss": 1.4111,
      "step": 3564
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.74500431765636e-05,
      "loss": 1.2835,
      "step": 3565
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.7424237412774434e-05,
      "loss": 1.3342,
      "step": 3566
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.73984392003012e-05,
      "loss": 1.311,
      "step": 3567
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.737264854777306e-05,
      "loss": 0.9594,
      "step": 3568
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.7346865463816697e-05,
      "loss": 1.3031,
      "step": 3569
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.7321089957056206e-05,
      "loss": 1.4357,
      "step": 3570
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.7295322036113212e-05,
      "loss": 1.3335,
      "step": 3571
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.7269561709606738e-05,
      "loss": 1.3412,
      "step": 3572
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.7243808986153335e-05,
      "loss": 1.4035,
      "step": 3573
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.721806387436693e-05,
      "loss": 1.3137,
      "step": 3574
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.719232638285898e-05,
      "loss": 1.243,
      "step": 3575
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.716659652023833e-05,
      "loss": 1.2851,
      "step": 3576
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.714087429511134e-05,
      "loss": 1.1296,
      "step": 3577
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.7115159716081767e-05,
      "loss": 1.0793,
      "step": 3578
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.708945279175079e-05,
      "loss": 1.1566,
      "step": 3579
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.7063753530717117e-05,
      "loss": 1.0968,
      "step": 3580
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.7038061941576777e-05,
      "loss": 0.962,
      "step": 3581
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.7012378032923346e-05,
      "loss": 1.2111,
      "step": 3582
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.6986701813347724e-05,
      "loss": 1.5727,
      "step": 3583
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.6961033291438343e-05,
      "loss": 1.3252,
      "step": 3584
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.693537247578096e-05,
      "loss": 1.2058,
      "step": 3585
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.6909719374958853e-05,
      "loss": 1.3585,
      "step": 3586
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.6884073997552628e-05,
      "loss": 1.2694,
      "step": 3587
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.685843635214038e-05,
      "loss": 1.383,
      "step": 3588
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.683280644729756e-05,
      "loss": 1.2733,
      "step": 3589
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.680718429159709e-05,
      "loss": 1.4156,
      "step": 3590
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.678156989360924e-05,
      "loss": 1.4785,
      "step": 3591
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.6755963261901708e-05,
      "loss": 1.1833,
      "step": 3592
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.673036440503962e-05,
      "loss": 1.1356,
      "step": 3593
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.6704773331585465e-05,
      "loss": 1.1703,
      "step": 3594
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.6679190050099156e-05,
      "loss": 1.2622,
      "step": 3595
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.6653614569137968e-05,
      "loss": 1.2109,
      "step": 3596
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.662804689725662e-05,
      "loss": 1.3678,
      "step": 3597
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.6602487043007135e-05,
      "loss": 1.2213,
      "step": 3598
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.657693501493902e-05,
      "loss": 1.0936,
      "step": 3599
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.6551390821599076e-05,
      "loss": 1.25,
      "step": 3600
    },
    {
      "epoch": 1.97,
      "eval_loss": 1.3334026336669922,
      "eval_runtime": 36.734,
      "eval_samples_per_second": 3.838,
      "eval_steps_per_second": 1.933,
      "step": 3600
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.6525854471531563e-05,
      "loss": 1.1681,
      "step": 3601
    },
    {
      "epoch": 1.97,
      "learning_rate": 2.650032597327805e-05,
      "loss": 1.298,
      "step": 3602
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.64748053353775e-05,
      "loss": 1.2304,
      "step": 3603
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.644929256636628e-05,
      "loss": 1.3095,
      "step": 3604
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.6423787674778068e-05,
      "loss": 1.2755,
      "step": 3605
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.6398290669143967e-05,
      "loss": 1.4486,
      "step": 3606
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.6372801557992382e-05,
      "loss": 1.0648,
      "step": 3607
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.6347320349849146e-05,
      "loss": 1.2202,
      "step": 3608
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.632184705323737e-05,
      "loss": 1.4993,
      "step": 3609
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.6296381676677606e-05,
      "loss": 1.5005,
      "step": 3610
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.627092422868766e-05,
      "loss": 1.343,
      "step": 3611
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.624547471778278e-05,
      "loss": 1.2247,
      "step": 3612
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.6220033152475475e-05,
      "loss": 1.3463,
      "step": 3613
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.6194599541275682e-05,
      "loss": 0.9256,
      "step": 3614
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.616917389269061e-05,
      "loss": 1.5186,
      "step": 3615
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.6143756215224802e-05,
      "loss": 1.3062,
      "step": 3616
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.6118346517380206e-05,
      "loss": 1.2361,
      "step": 3617
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.6092944807656004e-05,
      "loss": 1.3202,
      "step": 3618
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.6067551094548813e-05,
      "loss": 1.2939,
      "step": 3619
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.604216538655247e-05,
      "loss": 1.1569,
      "step": 3620
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.6016787692158217e-05,
      "loss": 1.4219,
      "step": 3621
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.5991418019854553e-05,
      "loss": 1.3058,
      "step": 3622
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.5966056378127358e-05,
      "loss": 1.0302,
      "step": 3623
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.5940702775459747e-05,
      "loss": 1.2877,
      "step": 3624
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.591535722033223e-05,
      "loss": 1.2427,
      "step": 3625
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.5890019721222554e-05,
      "loss": 1.5104,
      "step": 3626
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.586469028660583e-05,
      "loss": 1.1281,
      "step": 3627
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.5839368924954433e-05,
      "loss": 1.2291,
      "step": 3628
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.581405564473801e-05,
      "loss": 1.2432,
      "step": 3629
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.578875045442361e-05,
      "loss": 1.5861,
      "step": 3630
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.5763453362475454e-05,
      "loss": 1.4304,
      "step": 3631
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.5738164377355145e-05,
      "loss": 1.3417,
      "step": 3632
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.571288350752151e-05,
      "loss": 1.1612,
      "step": 3633
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.5687610761430725e-05,
      "loss": 0.8863,
      "step": 3634
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.5662346147536164e-05,
      "loss": 1.3451,
      "step": 3635
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.563708967428859e-05,
      "loss": 1.14,
      "step": 3636
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.561184135013593e-05,
      "loss": 1.1884,
      "step": 3637
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.5586601183523484e-05,
      "loss": 1.2977,
      "step": 3638
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5561369182893757e-05,
      "loss": 0.9943,
      "step": 3639
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5536145356686524e-05,
      "loss": 0.9998,
      "step": 3640
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5510929713338882e-05,
      "loss": 0.8941,
      "step": 3641
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5485722261285126e-05,
      "loss": 1.2052,
      "step": 3642
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5460523008956866e-05,
      "loss": 1.1588,
      "step": 3643
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5435331964782916e-05,
      "loss": 1.4062,
      "step": 3644
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.54101491371894e-05,
      "loss": 1.4462,
      "step": 3645
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.538497453459963e-05,
      "loss": 1.3722,
      "step": 3646
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5359808165434245e-05,
      "loss": 1.2882,
      "step": 3647
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5334650038111048e-05,
      "loss": 1.3775,
      "step": 3648
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5309500161045153e-05,
      "loss": 1.2857,
      "step": 3649
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5284358542648855e-05,
      "loss": 1.1411,
      "step": 3650
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5259225191331758e-05,
      "loss": 1.1649,
      "step": 3651
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5234100115500643e-05,
      "loss": 0.9806,
      "step": 3652
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.520898332355951e-05,
      "loss": 1.3248,
      "step": 3653
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.518387482390966e-05,
      "loss": 1.3742,
      "step": 3654
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5158774624949544e-05,
      "loss": 1.2953,
      "step": 3655
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5133682735074904e-05,
      "loss": 1.1277,
      "step": 3656
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.5108599162678637e-05,
      "loss": 1.0771,
      "step": 3657
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.508352391615093e-05,
      "loss": 1.33,
      "step": 3658
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.50584570038791e-05,
      "loss": 1.0587,
      "step": 3659
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.503339843424777e-05,
      "loss": 1.1885,
      "step": 3660
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.500834821563867e-05,
      "loss": 1.1128,
      "step": 3661
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4983306356430846e-05,
      "loss": 1.0938,
      "step": 3662
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.495827286500045e-05,
      "loss": 1.2262,
      "step": 3663
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4933247749720912e-05,
      "loss": 1.0932,
      "step": 3664
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.490823101896281e-05,
      "loss": 0.9527,
      "step": 3665
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4883222681093914e-05,
      "loss": 1.037,
      "step": 3666
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.485822274447925e-05,
      "loss": 0.9676,
      "step": 3667
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.483323121748094e-05,
      "loss": 1.0287,
      "step": 3668
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4808248108458397e-05,
      "loss": 1.3605,
      "step": 3669
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4783273425768117e-05,
      "loss": 1.2663,
      "step": 3670
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4758307177763863e-05,
      "loss": 1.3392,
      "step": 3671
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4733349372796507e-05,
      "loss": 1.1145,
      "step": 3672
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4708400019214162e-05,
      "loss": 1.321,
      "step": 3673
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4683459125362045e-05,
      "loss": 1.0509,
      "step": 3674
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.4658526699582613e-05,
      "loss": 1.3576,
      "step": 3675
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4633602750215444e-05,
      "loss": 0.972,
      "step": 3676
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4608687285597275e-05,
      "loss": 1.216,
      "step": 3677
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.458378031406206e-05,
      "loss": 1.1302,
      "step": 3678
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4558881843940834e-05,
      "loss": 1.3471,
      "step": 3679
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4533991883561868e-05,
      "loss": 1.1196,
      "step": 3680
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4509110441250517e-05,
      "loss": 1.2717,
      "step": 3681
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4484237525329352e-05,
      "loss": 1.2977,
      "step": 3682
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.445937314411802e-05,
      "loss": 1.1095,
      "step": 3683
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.443451730593339e-05,
      "loss": 1.2135,
      "step": 3684
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4409670019089404e-05,
      "loss": 1.1464,
      "step": 3685
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4384831291897202e-05,
      "loss": 0.9146,
      "step": 3686
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4360001132665007e-05,
      "loss": 1.3343,
      "step": 3687
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4335179549698233e-05,
      "loss": 0.9778,
      "step": 3688
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.431036655129938e-05,
      "loss": 1.4025,
      "step": 3689
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4285562145768047e-05,
      "loss": 1.0932,
      "step": 3690
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4260766341401092e-05,
      "loss": 1.0993,
      "step": 3691
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.4235979146492338e-05,
      "loss": 1.0783,
      "step": 3692
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.421120056933285e-05,
      "loss": 1.1928,
      "step": 3693
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.4186430618210705e-05,
      "loss": 1.2227,
      "step": 3694
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.4161669301411194e-05,
      "loss": 1.2337,
      "step": 3695
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.4136916627216655e-05,
      "loss": 1.2259,
      "step": 3696
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.4112172603906535e-05,
      "loss": 1.3744,
      "step": 3697
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.408743723975745e-05,
      "loss": 1.056,
      "step": 3698
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.406271054304302e-05,
      "loss": 1.4739,
      "step": 3699
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.403799252203408e-05,
      "loss": 1.2682,
      "step": 3700
    },
    {
      "epoch": 2.03,
      "eval_loss": 1.3452341556549072,
      "eval_runtime": 36.8487,
      "eval_samples_per_second": 3.826,
      "eval_steps_per_second": 1.927,
      "step": 3700
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.4013283184998463e-05,
      "loss": 1.1707,
      "step": 3701
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.3988582540201164e-05,
      "loss": 1.3281,
      "step": 3702
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.3963890595904225e-05,
      "loss": 1.2774,
      "step": 3703
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.3939207360366832e-05,
      "loss": 1.0884,
      "step": 3704
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.391453284184518e-05,
      "loss": 1.1875,
      "step": 3705
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.3889867048592633e-05,
      "loss": 0.9327,
      "step": 3706
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.3865209988859583e-05,
      "loss": 0.8896,
      "step": 3707
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.3840561670893496e-05,
      "loss": 1.0997,
      "step": 3708
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.381592210293896e-05,
      "loss": 1.0686,
      "step": 3709
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.3791291293237584e-05,
      "loss": 1.2293,
      "step": 3710
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.3766669250028096e-05,
      "loss": 0.9704,
      "step": 3711
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.374205598154624e-05,
      "loss": 1.2059,
      "step": 3712
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.371745149602489e-05,
      "loss": 1.1319,
      "step": 3713
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3692855801693904e-05,
      "loss": 1.3359,
      "step": 3714
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3668268906780274e-05,
      "loss": 1.0416,
      "step": 3715
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3643690819507986e-05,
      "loss": 0.8776,
      "step": 3716
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.361912154809814e-05,
      "loss": 1.272,
      "step": 3717
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3594561100768826e-05,
      "loss": 1.0822,
      "step": 3718
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.357000948573524e-05,
      "loss": 1.2387,
      "step": 3719
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3545466711209585e-05,
      "loss": 1.1684,
      "step": 3720
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3520932785401097e-05,
      "loss": 0.9405,
      "step": 3721
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.349640771651611e-05,
      "loss": 1.1061,
      "step": 3722
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3471891512757927e-05,
      "loss": 0.9399,
      "step": 3723
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3447384182326947e-05,
      "loss": 1.1918,
      "step": 3724
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.342288573342054e-05,
      "loss": 0.9397,
      "step": 3725
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3398396174233178e-05,
      "loss": 1.2425,
      "step": 3726
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.337391551295627e-05,
      "loss": 1.3044,
      "step": 3727
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3349443757778343e-05,
      "loss": 1.328,
      "step": 3728
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3324980916884857e-05,
      "loss": 1.0524,
      "step": 3729
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.3300526998458372e-05,
      "loss": 1.2823,
      "step": 3730
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.327608201067839e-05,
      "loss": 1.3094,
      "step": 3731
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.3251645961721492e-05,
      "loss": 1.1731,
      "step": 3732
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.322721885976122e-05,
      "loss": 1.1782,
      "step": 3733
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.320280071296812e-05,
      "loss": 1.5621,
      "step": 3734
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.3178391529509806e-05,
      "loss": 1.2209,
      "step": 3735
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.315399131755081e-05,
      "loss": 1.0861,
      "step": 3736
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.312960008525274e-05,
      "loss": 1.4511,
      "step": 3737
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.3105217840774123e-05,
      "loss": 1.1541,
      "step": 3738
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.3080844592270568e-05,
      "loss": 1.0858,
      "step": 3739
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.3056480347894582e-05,
      "loss": 1.4498,
      "step": 3740
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.3032125115795744e-05,
      "loss": 1.2278,
      "step": 3741
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.3007778904120552e-05,
      "loss": 1.1515,
      "step": 3742
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.2983441721012544e-05,
      "loss": 1.1447,
      "step": 3743
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.29591135746122e-05,
      "loss": 1.0326,
      "step": 3744
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.2934794473056957e-05,
      "loss": 1.319,
      "step": 3745
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.2910484424481298e-05,
      "loss": 1.0955,
      "step": 3746
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.28861834370166e-05,
      "loss": 1.2673,
      "step": 3747
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.2861891518791282e-05,
      "loss": 1.1203,
      "step": 3748
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.283760867793066e-05,
      "loss": 1.1761,
      "step": 3749
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.281333492255708e-05,
      "loss": 1.0385,
      "step": 3750
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.278907026078978e-05,
      "loss": 1.103,
      "step": 3751
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.2764814700745025e-05,
      "loss": 1.1239,
      "step": 3752
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.2740568250535964e-05,
      "loss": 1.3648,
      "step": 3753
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.271633091827279e-05,
      "loss": 1.148,
      "step": 3754
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.2692102712062534e-05,
      "loss": 1.0321,
      "step": 3755
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.266788364000929e-05,
      "loss": 1.2674,
      "step": 3756
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.2643673710214013e-05,
      "loss": 1.1937,
      "step": 3757
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.261947293077461e-05,
      "loss": 1.5161,
      "step": 3758
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.259528130978598e-05,
      "loss": 1.1434,
      "step": 3759
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.25710988553399e-05,
      "loss": 1.0991,
      "step": 3760
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.2546925575525123e-05,
      "loss": 1.2622,
      "step": 3761
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.252276147842729e-05,
      "loss": 1.2817,
      "step": 3762
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.2498606572129026e-05,
      "loss": 1.3067,
      "step": 3763
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.2474460864709824e-05,
      "loss": 1.0118,
      "step": 3764
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.245032436424615e-05,
      "loss": 1.3579,
      "step": 3765
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.242619707881134e-05,
      "loss": 1.0482,
      "step": 3766
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.2402079016475712e-05,
      "loss": 1.1737,
      "step": 3767
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.2377970185306424e-05,
      "loss": 1.2789,
      "step": 3768
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.235387059336762e-05,
      "loss": 1.1538,
      "step": 3769
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.23297802487203e-05,
      "loss": 1.1594,
      "step": 3770
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.230569915942237e-05,
      "loss": 1.2204,
      "step": 3771
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.2281627333528697e-05,
      "loss": 1.4282,
      "step": 3772
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.2257564779090984e-05,
      "loss": 1.0044,
      "step": 3773
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.223351150415788e-05,
      "loss": 1.3545,
      "step": 3774
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.2209467516774884e-05,
      "loss": 1.2071,
      "step": 3775
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.2185432824984453e-05,
      "loss": 1.0879,
      "step": 3776
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.216140743682586e-05,
      "loss": 0.9955,
      "step": 3777
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.213739136033533e-05,
      "loss": 1.1734,
      "step": 3778
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.2113384603545924e-05,
      "loss": 1.0252,
      "step": 3779
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.2089387174487632e-05,
      "loss": 1.0726,
      "step": 3780
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.206539908118729e-05,
      "loss": 0.8958,
      "step": 3781
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.20414203316686e-05,
      "loss": 1.0953,
      "step": 3782
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.2017450933952192e-05,
      "loss": 1.0898,
      "step": 3783
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.1993490896055512e-05,
      "loss": 1.1721,
      "step": 3784
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.1969540225992923e-05,
      "loss": 0.8334,
      "step": 3785
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.1945598931775595e-05,
      "loss": 1.1953,
      "step": 3786
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.1921667021411645e-05,
      "loss": 1.3453,
      "step": 3787
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.1897744502905954e-05,
      "loss": 1.2422,
      "step": 3788
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.1873831384260358e-05,
      "loss": 1.4016,
      "step": 3789
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.184992767347346e-05,
      "loss": 1.0666,
      "step": 3790
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.182603337854079e-05,
      "loss": 1.3017,
      "step": 3791
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.180214850745467e-05,
      "loss": 1.3947,
      "step": 3792
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.177827306820433e-05,
      "loss": 1.3305,
      "step": 3793
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.1754407068775794e-05,
      "loss": 1.2991,
      "step": 3794
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.1730550517151925e-05,
      "loss": 1.1285,
      "step": 3795
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.1706703421312492e-05,
      "loss": 1.0355,
      "step": 3796
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.1682865789234007e-05,
      "loss": 1.3535,
      "step": 3797
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.165903762888991e-05,
      "loss": 1.2593,
      "step": 3798
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.163521894825039e-05,
      "loss": 1.2031,
      "step": 3799
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.161140975528254e-05,
      "loss": 1.3107,
      "step": 3800
    },
    {
      "epoch": 2.08,
      "eval_loss": 1.3471230268478394,
      "eval_runtime": 36.7727,
      "eval_samples_per_second": 3.834,
      "eval_steps_per_second": 1.931,
      "step": 3800
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.1587610057950214e-05,
      "loss": 0.9583,
      "step": 3801
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.1563819864214147e-05,
      "loss": 1.1917,
      "step": 3802
    },
    {
      "epoch": 2.08,
      "learning_rate": 2.154003918203183e-05,
      "loss": 1.266,
      "step": 3803
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1516268019357656e-05,
      "loss": 1.3505,
      "step": 3804
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.149250638414274e-05,
      "loss": 1.3439,
      "step": 3805
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1468754284335097e-05,
      "loss": 1.3365,
      "step": 3806
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.14450117278795e-05,
      "loss": 1.1143,
      "step": 3807
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1421278722717524e-05,
      "loss": 1.2478,
      "step": 3808
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1397555276787602e-05,
      "loss": 1.0857,
      "step": 3809
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1373841398024896e-05,
      "loss": 1.0777,
      "step": 3810
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1350137094361454e-05,
      "loss": 1.2432,
      "step": 3811
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1326442373726028e-05,
      "loss": 1.2638,
      "step": 3812
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1302757244044257e-05,
      "loss": 1.2286,
      "step": 3813
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.127908171323849e-05,
      "loss": 1.2331,
      "step": 3814
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1255415789227927e-05,
      "loss": 1.0644,
      "step": 3815
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.123175947992851e-05,
      "loss": 1.402,
      "step": 3816
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1208112793253003e-05,
      "loss": 1.332,
      "step": 3817
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1184475737110932e-05,
      "loss": 1.0413,
      "step": 3818
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1160848319408573e-05,
      "loss": 1.1846,
      "step": 3819
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1137230548049043e-05,
      "loss": 1.1743,
      "step": 3820
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.1113622430932167e-05,
      "loss": 1.0189,
      "step": 3821
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.109002397595461e-05,
      "loss": 1.0243,
      "step": 3822
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.1066435191009715e-05,
      "loss": 1.2382,
      "step": 3823
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.1042856083987695e-05,
      "loss": 1.2305,
      "step": 3824
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.101928666277543e-05,
      "loss": 1.1277,
      "step": 3825
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.0995726935256634e-05,
      "loss": 1.1992,
      "step": 3826
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.0972176909311712e-05,
      "loss": 1.338,
      "step": 3827
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.09486365928179e-05,
      "loss": 0.9652,
      "step": 3828
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.0925105993649107e-05,
      "loss": 1.1123,
      "step": 3829
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.0901585119676064e-05,
      "loss": 1.2286,
      "step": 3830
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.0878073978766203e-05,
      "loss": 1.1115,
      "step": 3831
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.0854572578783686e-05,
      "loss": 1.1612,
      "step": 3832
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.083108092758949e-05,
      "loss": 1.244,
      "step": 3833
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.0807599033041235e-05,
      "loss": 1.2954,
      "step": 3834
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.078412690299338e-05,
      "loss": 1.0739,
      "step": 3835
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.076066454529701e-05,
      "loss": 1.1561,
      "step": 3836
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.0737211967800047e-05,
      "loss": 1.3484,
      "step": 3837
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.0713769178347043e-05,
      "loss": 0.8982,
      "step": 3838
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.069033618477937e-05,
      "loss": 1.1869,
      "step": 3839
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0666912994935034e-05,
      "loss": 1.1992,
      "step": 3840
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.064349961664885e-05,
      "loss": 1.166,
      "step": 3841
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.062009605775226e-05,
      "loss": 1.201,
      "step": 3842
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0596702326073503e-05,
      "loss": 1.0259,
      "step": 3843
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.057331842943749e-05,
      "loss": 1.2963,
      "step": 3844
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0549944375665815e-05,
      "loss": 1.0619,
      "step": 3845
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.052658017257686e-05,
      "loss": 1.3202,
      "step": 3846
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.050322582798562e-05,
      "loss": 1.2343,
      "step": 3847
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0479881349703883e-05,
      "loss": 1.1146,
      "step": 3848
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0456546745540046e-05,
      "loss": 1.3117,
      "step": 3849
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0433222023299287e-05,
      "loss": 1.3327,
      "step": 3850
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0409907190783406e-05,
      "loss": 1.1523,
      "step": 3851
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.038660225579096e-05,
      "loss": 0.9723,
      "step": 3852
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0363307226117134e-05,
      "loss": 1.3503,
      "step": 3853
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0340022109553864e-05,
      "loss": 1.2821,
      "step": 3854
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0316746913889712e-05,
      "loss": 1.2108,
      "step": 3855
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0293481646909934e-05,
      "loss": 0.9674,
      "step": 3856
    },
    {
      "epoch": 2.11,
      "learning_rate": 2.0270226316396512e-05,
      "loss": 0.9987,
      "step": 3857
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.0246980930128034e-05,
      "loss": 1.1556,
      "step": 3858
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.022374549587983e-05,
      "loss": 1.0495,
      "step": 3859
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.0200520021423813e-05,
      "loss": 1.1535,
      "step": 3860
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.01773045145287e-05,
      "loss": 1.2717,
      "step": 3861
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.0154098982959745e-05,
      "loss": 1.1549,
      "step": 3862
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.013090343447891e-05,
      "loss": 1.4695,
      "step": 3863
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.0107717876844838e-05,
      "loss": 1.2852,
      "step": 3864
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.0084542317812798e-05,
      "loss": 1.137,
      "step": 3865
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.006137676513476e-05,
      "loss": 1.0051,
      "step": 3866
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.003822122655928e-05,
      "loss": 1.0642,
      "step": 3867
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.0015075709831634e-05,
      "loss": 1.1157,
      "step": 3868
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.9991940222693683e-05,
      "loss": 0.9963,
      "step": 3869
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.9968814772883997e-05,
      "loss": 1.0846,
      "step": 3870
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.9945699368137722e-05,
      "loss": 1.0863,
      "step": 3871
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.9922594016186713e-05,
      "loss": 1.2545,
      "step": 3872
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.989949872475939e-05,
      "loss": 1.0995,
      "step": 3873
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.987641350158088e-05,
      "loss": 1.14,
      "step": 3874
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.9853338354372895e-05,
      "loss": 1.3606,
      "step": 3875
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.983027329085377e-05,
      "loss": 1.0092,
      "step": 3876
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9807218318738506e-05,
      "loss": 1.0819,
      "step": 3877
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9784173445738697e-05,
      "loss": 1.162,
      "step": 3878
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.976113867956259e-05,
      "loss": 1.2499,
      "step": 3879
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9738114027915006e-05,
      "loss": 1.0992,
      "step": 3880
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9715099498497437e-05,
      "loss": 1.3919,
      "step": 3881
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9692095099007923e-05,
      "loss": 0.9763,
      "step": 3882
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9669100837141203e-05,
      "loss": 1.222,
      "step": 3883
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9646116720588525e-05,
      "loss": 1.0724,
      "step": 3884
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9623142757037845e-05,
      "loss": 1.1757,
      "step": 3885
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9600178954173635e-05,
      "loss": 1.2661,
      "step": 3886
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9577225319677005e-05,
      "loss": 0.9276,
      "step": 3887
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9554281861225694e-05,
      "loss": 1.0956,
      "step": 3888
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9531348586493976e-05,
      "loss": 1.3022,
      "step": 3889
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9508425503152773e-05,
      "loss": 1.1138,
      "step": 3890
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9485512618869557e-05,
      "loss": 1.4978,
      "step": 3891
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.946260994130843e-05,
      "loss": 1.1553,
      "step": 3892
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9439717478130027e-05,
      "loss": 0.9687,
      "step": 3893
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.9416835236991625e-05,
      "loss": 1.2288,
      "step": 3894
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9393963225547035e-05,
      "loss": 1.2796,
      "step": 3895
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.937110145144668e-05,
      "loss": 1.3476,
      "step": 3896
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.934824992233752e-05,
      "loss": 1.1772,
      "step": 3897
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9325408645863142e-05,
      "loss": 1.1801,
      "step": 3898
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9302577629663653e-05,
      "loss": 1.2684,
      "step": 3899
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9279756881375743e-05,
      "loss": 1.2096,
      "step": 3900
    },
    {
      "epoch": 2.14,
      "eval_loss": 1.34959077835083,
      "eval_runtime": 36.7869,
      "eval_samples_per_second": 3.833,
      "eval_steps_per_second": 1.93,
      "step": 3900
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9256946408632693e-05,
      "loss": 1.4328,
      "step": 3901
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9234146219064298e-05,
      "loss": 1.2425,
      "step": 3902
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9211356320296974e-05,
      "loss": 1.272,
      "step": 3903
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9188576719953633e-05,
      "loss": 1.1198,
      "step": 3904
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9165807425653803e-05,
      "loss": 1.3337,
      "step": 3905
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9143048445013507e-05,
      "loss": 1.4266,
      "step": 3906
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9120299785645364e-05,
      "loss": 1.117,
      "step": 3907
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9097561455158502e-05,
      "loss": 1.2563,
      "step": 3908
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.907483346115864e-05,
      "loss": 0.8884,
      "step": 3909
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9052115811247985e-05,
      "loss": 0.9847,
      "step": 3910
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9029408513025338e-05,
      "loss": 0.8826,
      "step": 3911
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.9006711574086005e-05,
      "loss": 1.203,
      "step": 3912
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.898402500202181e-05,
      "loss": 1.1284,
      "step": 3913
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.8961348804421175e-05,
      "loss": 1.2436,
      "step": 3914
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.8938682988868966e-05,
      "loss": 1.2757,
      "step": 3915
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.891602756294666e-05,
      "loss": 1.3427,
      "step": 3916
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.8893382534232186e-05,
      "loss": 1.2345,
      "step": 3917
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.887074791030006e-05,
      "loss": 1.1832,
      "step": 3918
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.884812369872126e-05,
      "loss": 1.2515,
      "step": 3919
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.8825509907063327e-05,
      "loss": 1.0917,
      "step": 3920
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.8802906542890276e-05,
      "loss": 1.0505,
      "step": 3921
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.8780313613762685e-05,
      "loss": 1.1441,
      "step": 3922
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.8757731127237593e-05,
      "loss": 1.1228,
      "step": 3923
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.8735159090868547e-05,
      "loss": 1.2866,
      "step": 3924
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.871259751220566e-05,
      "loss": 1.029,
      "step": 3925
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.869004639879546e-05,
      "loss": 1.2398,
      "step": 3926
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.8667505758181052e-05,
      "loss": 1.0749,
      "step": 3927
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.8644975597901977e-05,
      "loss": 1.2975,
      "step": 3928
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.862245592549432e-05,
      "loss": 1.0657,
      "step": 3929
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.8599946748490615e-05,
      "loss": 1.0841,
      "step": 3930
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8577448074419933e-05,
      "loss": 1.1605,
      "step": 3931
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8554959910807775e-05,
      "loss": 1.4937,
      "step": 3932
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8532482265176183e-05,
      "loss": 1.1345,
      "step": 3933
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.851001514504362e-05,
      "loss": 1.2752,
      "step": 3934
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8487558557925106e-05,
      "loss": 0.9941,
      "step": 3935
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8465112511332065e-05,
      "loss": 1.4631,
      "step": 3936
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.844267701277242e-05,
      "loss": 1.2188,
      "step": 3937
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8420252069750598e-05,
      "loss": 1.2472,
      "step": 3938
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8397837689767426e-05,
      "loss": 1.0498,
      "step": 3939
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8375433880320292e-05,
      "loss": 1.1593,
      "step": 3940
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.835304064890294e-05,
      "loss": 1.2015,
      "step": 3941
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8330658003005685e-05,
      "loss": 1.2426,
      "step": 3942
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.83082859501152e-05,
      "loss": 1.102,
      "step": 3943
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8285924497714703e-05,
      "loss": 1.1932,
      "step": 3944
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8263573653283788e-05,
      "loss": 1.4091,
      "step": 3945
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.824123342429858e-05,
      "loss": 1.0211,
      "step": 3946
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.8218903818231568e-05,
      "loss": 1.2746,
      "step": 3947
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.819658484255177e-05,
      "loss": 0.9283,
      "step": 3948
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.8174276504724596e-05,
      "loss": 1.1089,
      "step": 3949
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.815197881221189e-05,
      "loss": 0.9855,
      "step": 3950
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.8129691772471997e-05,
      "loss": 1.0758,
      "step": 3951
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.8107415392959614e-05,
      "loss": 1.3456,
      "step": 3952
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.8085149681125964e-05,
      "loss": 1.0392,
      "step": 3953
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.8062894644418604e-05,
      "loss": 1.2081,
      "step": 3954
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.8040650290281625e-05,
      "loss": 1.0933,
      "step": 3955
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.8018416626155443e-05,
      "loss": 1.0971,
      "step": 3956
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.7996193659476983e-05,
      "loss": 1.3542,
      "step": 3957
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.7973981397679525e-05,
      "loss": 1.2651,
      "step": 3958
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.7951779848192822e-05,
      "loss": 1.2927,
      "step": 3959
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.7929589018443016e-05,
      "loss": 1.1177,
      "step": 3960
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.7907408915852645e-05,
      "loss": 1.3053,
      "step": 3961
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.7885239547840716e-05,
      "loss": 1.3014,
      "step": 3962
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.786308092182258e-05,
      "loss": 1.1756,
      "step": 3963
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.7840933045210052e-05,
      "loss": 1.0874,
      "step": 3964
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.7818795925411297e-05,
      "loss": 1.1689,
      "step": 3965
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.7796669569830943e-05,
      "loss": 1.1182,
      "step": 3966
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.777455398586995e-05,
      "loss": 1.1503,
      "step": 3967
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7752449180925747e-05,
      "loss": 1.0798,
      "step": 3968
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.773035516239208e-05,
      "loss": 1.0795,
      "step": 3969
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7708271937659173e-05,
      "loss": 1.1757,
      "step": 3970
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7686199514113543e-05,
      "loss": 1.2341,
      "step": 3971
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7664137899138193e-05,
      "loss": 1.2888,
      "step": 3972
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7642087100112443e-05,
      "loss": 1.1414,
      "step": 3973
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7620047124412e-05,
      "loss": 1.0883,
      "step": 3974
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7598017979408987e-05,
      "loss": 1.2858,
      "step": 3975
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7575999672471867e-05,
      "loss": 0.8241,
      "step": 3976
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7553992210965515e-05,
      "loss": 1.0252,
      "step": 3977
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.753199560225113e-05,
      "loss": 1.1216,
      "step": 3978
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.751000985368633e-05,
      "loss": 1.1897,
      "step": 3979
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7488034972625066e-05,
      "loss": 1.2212,
      "step": 3980
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.746607096641768e-05,
      "loss": 1.1014,
      "step": 3981
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7444117842410845e-05,
      "loss": 1.4068,
      "step": 3982
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7422175607947634e-05,
      "loss": 1.3333,
      "step": 3983
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7400244270367428e-05,
      "loss": 1.217,
      "step": 3984
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.7378323837006017e-05,
      "loss": 1.1804,
      "step": 3985
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.735641431519551e-05,
      "loss": 1.3416,
      "step": 3986
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.733451571226435e-05,
      "loss": 1.2149,
      "step": 3987
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7312628035537387e-05,
      "loss": 1.1771,
      "step": 3988
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.729075129233575e-05,
      "loss": 1.2426,
      "step": 3989
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.726888548997697e-05,
      "loss": 1.11,
      "step": 3990
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7247030635774865e-05,
      "loss": 1.1395,
      "step": 3991
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7225186737039638e-05,
      "loss": 1.2179,
      "step": 3992
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7203353801077774e-05,
      "loss": 1.1921,
      "step": 3993
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7181531835192165e-05,
      "loss": 1.3615,
      "step": 3994
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7159720846681953e-05,
      "loss": 1.2025,
      "step": 3995
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7137920842842674e-05,
      "loss": 1.2992,
      "step": 3996
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7116131830966154e-05,
      "loss": 1.5106,
      "step": 3997
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.709435381834053e-05,
      "loss": 1.3364,
      "step": 3998
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7072586812250307e-05,
      "loss": 1.1915,
      "step": 3999
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7050830819976267e-05,
      "loss": 1.4503,
      "step": 4000
    },
    {
      "epoch": 2.19,
      "eval_loss": 1.3502633571624756,
      "eval_runtime": 36.7962,
      "eval_samples_per_second": 3.832,
      "eval_steps_per_second": 1.93,
      "step": 4000
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7029085848795533e-05,
      "loss": 1.1674,
      "step": 4001
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.7007351905981512e-05,
      "loss": 1.313,
      "step": 4002
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.6985628998803977e-05,
      "loss": 1.2348,
      "step": 4003
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.696391713452893e-05,
      "loss": 1.1665,
      "step": 4004
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6942216320418763e-05,
      "loss": 1.2483,
      "step": 4005
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.69205265637321e-05,
      "loss": 0.9852,
      "step": 4006
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6898847871723934e-05,
      "loss": 1.0014,
      "step": 4007
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6877180251645486e-05,
      "loss": 1.1503,
      "step": 4008
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6855523710744337e-05,
      "loss": 1.2257,
      "step": 4009
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6833878256264324e-05,
      "loss": 1.0522,
      "step": 4010
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6812243895445572e-05,
      "loss": 1.2645,
      "step": 4011
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.679062063552454e-05,
      "loss": 0.7984,
      "step": 4012
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6769008483733906e-05,
      "loss": 1.3196,
      "step": 4013
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.674740744730271e-05,
      "loss": 1.0894,
      "step": 4014
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6725817533456195e-05,
      "loss": 1.2404,
      "step": 4015
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6704238749415957e-05,
      "loss": 1.0299,
      "step": 4016
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6682671102399805e-05,
      "loss": 1.1645,
      "step": 4017
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6661114599621885e-05,
      "loss": 1.3444,
      "step": 4018
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6639569248292548e-05,
      "loss": 1.1354,
      "step": 4019
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6618035055618485e-05,
      "loss": 1.2408,
      "step": 4020
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.6596512028802585e-05,
      "loss": 1.0601,
      "step": 4021
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.657500017504407e-05,
      "loss": 1.1251,
      "step": 4022
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.6553499501538373e-05,
      "loss": 1.2974,
      "step": 4023
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.653201001547719e-05,
      "loss": 1.3047,
      "step": 4024
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.651053172404853e-05,
      "loss": 1.1635,
      "step": 4025
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.648906463443658e-05,
      "loss": 1.1243,
      "step": 4026
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.6467608753821843e-05,
      "loss": 1.0848,
      "step": 4027
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.6446164089381034e-05,
      "loss": 1.0962,
      "step": 4028
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.642473064828715e-05,
      "loss": 1.1423,
      "step": 4029
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.640330843770938e-05,
      "loss": 1.361,
      "step": 4030
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.6381897464813218e-05,
      "loss": 1.0129,
      "step": 4031
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.6360497736760383e-05,
      "loss": 0.956,
      "step": 4032
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.6339109260708785e-05,
      "loss": 1.2607,
      "step": 4033
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.6317732043812646e-05,
      "loss": 1.0519,
      "step": 4034
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.629636609322234e-05,
      "loss": 1.0951,
      "step": 4035
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.6275011416084562e-05,
      "loss": 0.917,
      "step": 4036
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.6253668019542157e-05,
      "loss": 1.0586,
      "step": 4037
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.623233591073425e-05,
      "loss": 1.0374,
      "step": 4038
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.621101509679614e-05,
      "loss": 1.2356,
      "step": 4039
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.618970558485942e-05,
      "loss": 1.0804,
      "step": 4040
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.6168407382051838e-05,
      "loss": 1.2188,
      "step": 4041
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.6147120495497365e-05,
      "loss": 0.9901,
      "step": 4042
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.6125844932316237e-05,
      "loss": 1.2815,
      "step": 4043
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.6104580699624837e-05,
      "loss": 0.9778,
      "step": 4044
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.6083327804535827e-05,
      "loss": 1.1289,
      "step": 4045
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.6062086254157997e-05,
      "loss": 1.2118,
      "step": 4046
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.6040856055596428e-05,
      "loss": 1.201,
      "step": 4047
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.601963721595232e-05,
      "loss": 1.1249,
      "step": 4048
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.5998429742323156e-05,
      "loss": 1.2476,
      "step": 4049
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.5977233641802543e-05,
      "loss": 1.1646,
      "step": 4050
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.595604892148034e-05,
      "loss": 1.2445,
      "step": 4051
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.593487558844255e-05,
      "loss": 1.3236,
      "step": 4052
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.5913713649771427e-05,
      "loss": 1.2304,
      "step": 4053
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.5892563112545366e-05,
      "loss": 1.3378,
      "step": 4054
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.587142398383894e-05,
      "loss": 1.362,
      "step": 4055
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.5850296270722964e-05,
      "loss": 1.0514,
      "step": 4056
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.582917998026437e-05,
      "loss": 1.1942,
      "step": 4057
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.5808075119526324e-05,
      "loss": 1.0021,
      "step": 4058
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.578698169556811e-05,
      "loss": 1.2474,
      "step": 4059
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.576589971544526e-05,
      "loss": 1.2171,
      "step": 4060
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5744829186209404e-05,
      "loss": 1.1501,
      "step": 4061
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5723770114908403e-05,
      "loss": 1.164,
      "step": 4062
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.570272250858623e-05,
      "loss": 1.1443,
      "step": 4063
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5681686374283088e-05,
      "loss": 0.9631,
      "step": 4064
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.566066171903528e-05,
      "loss": 1.3121,
      "step": 4065
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5639648549875287e-05,
      "loss": 1.0409,
      "step": 4066
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.561864687383178e-05,
      "loss": 1.341,
      "step": 4067
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5597656697929542e-05,
      "loss": 1.1175,
      "step": 4068
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.557667802918955e-05,
      "loss": 1.1883,
      "step": 4069
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5555710874628893e-05,
      "loss": 1.1429,
      "step": 4070
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5534755241260844e-05,
      "loss": 1.2015,
      "step": 4071
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5513811136094787e-05,
      "loss": 1.0961,
      "step": 4072
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5492878566136292e-05,
      "loss": 1.2079,
      "step": 4073
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5471957538387015e-05,
      "loss": 1.2899,
      "step": 4074
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.5451048059844825e-05,
      "loss": 1.3243,
      "step": 4075
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.543015013750364e-05,
      "loss": 1.1496,
      "step": 4076
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5409263778353605e-05,
      "loss": 1.0936,
      "step": 4077
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5388388989380924e-05,
      "loss": 0.9974,
      "step": 4078
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5367525777567955e-05,
      "loss": 1.1658,
      "step": 4079
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5346674149893202e-05,
      "loss": 1.274,
      "step": 4080
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5325834113331255e-05,
      "loss": 1.2334,
      "step": 4081
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5305005674852884e-05,
      "loss": 1.3164,
      "step": 4082
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5284188841424917e-05,
      "loss": 1.1057,
      "step": 4083
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5263383620010357e-05,
      "loss": 1.2125,
      "step": 4084
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5242590017568264e-05,
      "loss": 1.3051,
      "step": 4085
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5221808041053875e-05,
      "loss": 1.2588,
      "step": 4086
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.520103769741848e-05,
      "loss": 1.2459,
      "step": 4087
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5180278993609526e-05,
      "loss": 1.2664,
      "step": 4088
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.515953193657052e-05,
      "loss": 1.292,
      "step": 4089
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5138796533241134e-05,
      "loss": 1.1671,
      "step": 4090
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.511807279055708e-05,
      "loss": 1.1115,
      "step": 4091
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5097360715450187e-05,
      "loss": 1.1799,
      "step": 4092
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5076660314848423e-05,
      "loss": 1.3909,
      "step": 4093
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.5055971595675783e-05,
      "loss": 1.1174,
      "step": 4094
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.503529456485242e-05,
      "loss": 1.0242,
      "step": 4095
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.5014629229294524e-05,
      "loss": 1.2444,
      "step": 4096
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4993975595914422e-05,
      "loss": 1.1438,
      "step": 4097
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4973333671620465e-05,
      "loss": 1.3815,
      "step": 4098
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.495270346331717e-05,
      "loss": 1.2914,
      "step": 4099
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4932084977905042e-05,
      "loss": 1.142,
      "step": 4100
    },
    {
      "epoch": 2.25,
      "eval_loss": 1.3484505414962769,
      "eval_runtime": 36.8038,
      "eval_samples_per_second": 3.831,
      "eval_steps_per_second": 1.929,
      "step": 4100
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.491147822228075e-05,
      "loss": 1.1104,
      "step": 4101
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4890883203336981e-05,
      "loss": 1.0898,
      "step": 4102
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4870299927962511e-05,
      "loss": 1.0645,
      "step": 4103
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4849728403042213e-05,
      "loss": 0.9813,
      "step": 4104
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4829168635456986e-05,
      "loss": 1.127,
      "step": 4105
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4808620632083853e-05,
      "loss": 1.2253,
      "step": 4106
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4788084399795832e-05,
      "loss": 1.2059,
      "step": 4107
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4767559945462072e-05,
      "loss": 1.3234,
      "step": 4108
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4747047275947728e-05,
      "loss": 1.1313,
      "step": 4109
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.472654639811406e-05,
      "loss": 1.1896,
      "step": 4110
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4706057318818333e-05,
      "loss": 1.2212,
      "step": 4111
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4685580044913921e-05,
      "loss": 1.297,
      "step": 4112
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4665114583250195e-05,
      "loss": 1.2384,
      "step": 4113
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 1.1747,
      "step": 4114
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4624219124022698e-05,
      "loss": 1.1083,
      "step": 4115
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.460378914013793e-05,
      "loss": 1.2075,
      "step": 4116
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4583370995851931e-05,
      "loss": 1.1368,
      "step": 4117
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4562964697994286e-05,
      "loss": 1.2265,
      "step": 4118
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4542570253390685e-05,
      "loss": 1.2804,
      "step": 4119
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4522187668862796e-05,
      "loss": 1.1823,
      "step": 4120
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.450181695122837e-05,
      "loss": 1.4426,
      "step": 4121
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4481458107301121e-05,
      "loss": 0.9146,
      "step": 4122
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4461111143890882e-05,
      "loss": 1.1519,
      "step": 4123
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4440776067803419e-05,
      "loss": 1.2365,
      "step": 4124
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4420452885840596e-05,
      "loss": 1.176,
      "step": 4125
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4400141604800249e-05,
      "loss": 1.1624,
      "step": 4126
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4379842231476275e-05,
      "loss": 1.1219,
      "step": 4127
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4359554772658552e-05,
      "loss": 1.2473,
      "step": 4128
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4339279235132962e-05,
      "loss": 1.0613,
      "step": 4129
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4319015625681475e-05,
      "loss": 1.2941,
      "step": 4130
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.4298763951081973e-05,
      "loss": 1.0536,
      "step": 4131
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4278524218108424e-05,
      "loss": 1.0487,
      "step": 4132
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4258296433530749e-05,
      "loss": 1.2619,
      "step": 4133
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4238080604114917e-05,
      "loss": 1.1712,
      "step": 4134
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4217876736622849e-05,
      "loss": 1.3036,
      "step": 4135
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.419768483781252e-05,
      "loss": 1.15,
      "step": 4136
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4177504914437839e-05,
      "loss": 1.3309,
      "step": 4137
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.415733697324878e-05,
      "loss": 1.2518,
      "step": 4138
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4137181020991252e-05,
      "loss": 1.0117,
      "step": 4139
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4117037064407162e-05,
      "loss": 1.2923,
      "step": 4140
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4096905110234442e-05,
      "loss": 0.8438,
      "step": 4141
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4076785165206962e-05,
      "loss": 1.1496,
      "step": 4142
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4056677236054622e-05,
      "loss": 1.185,
      "step": 4143
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4036581329503245e-05,
      "loss": 1.1629,
      "step": 4144
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.4016497452274701e-05,
      "loss": 1.1869,
      "step": 4145
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.3996425611086767e-05,
      "loss": 0.9562,
      "step": 4146
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.3976365812653259e-05,
      "loss": 1.2053,
      "step": 4147
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.3956318063683904e-05,
      "loss": 1.1092,
      "step": 4148
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.3936282370884457e-05,
      "loss": 1.1192,
      "step": 4149
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3916258740956583e-05,
      "loss": 1.0836,
      "step": 4150
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3896247180597976e-05,
      "loss": 1.3616,
      "step": 4151
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3876247696502238e-05,
      "loss": 0.9973,
      "step": 4152
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.385626029535893e-05,
      "loss": 1.2017,
      "step": 4153
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3836284983853632e-05,
      "loss": 1.0917,
      "step": 4154
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3816321768667806e-05,
      "loss": 1.1964,
      "step": 4155
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3796370656478935e-05,
      "loss": 1.119,
      "step": 4156
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3776431653960392e-05,
      "loss": 1.1552,
      "step": 4157
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3756504767781559e-05,
      "loss": 1.3205,
      "step": 4158
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3736590004607703e-05,
      "loss": 1.1208,
      "step": 4159
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3716687371100096e-05,
      "loss": 1.2736,
      "step": 4160
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.36967968739159e-05,
      "loss": 1.09,
      "step": 4161
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3676918519708276e-05,
      "loss": 1.0809,
      "step": 4162
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.365705231512625e-05,
      "loss": 0.9621,
      "step": 4163
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.363719826681486e-05,
      "loss": 1.2032,
      "step": 4164
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3617356381415025e-05,
      "loss": 1.041,
      "step": 4165
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3597526665563598e-05,
      "loss": 1.1552,
      "step": 4166
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.3577709125893406e-05,
      "loss": 1.1417,
      "step": 4167
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3557903769033148e-05,
      "loss": 1.2933,
      "step": 4168
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.353811060160749e-05,
      "loss": 1.0238,
      "step": 4169
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3518329630236987e-05,
      "loss": 1.2929,
      "step": 4170
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3498560861538156e-05,
      "loss": 1.232,
      "step": 4171
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3478804302123382e-05,
      "loss": 1.3254,
      "step": 4172
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3459059958601018e-05,
      "loss": 1.155,
      "step": 4173
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3439327837575271e-05,
      "loss": 0.9473,
      "step": 4174
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3419607945646334e-05,
      "loss": 1.0703,
      "step": 4175
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3399900289410245e-05,
      "loss": 1.2622,
      "step": 4176
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3380204875458963e-05,
      "loss": 1.0976,
      "step": 4177
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3360521710380391e-05,
      "loss": 1.2412,
      "step": 4178
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3340850800758275e-05,
      "loss": 1.2597,
      "step": 4179
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3321192153172329e-05,
      "loss": 1.2102,
      "step": 4180
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3301545774198098e-05,
      "loss": 1.0519,
      "step": 4181
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3281911670407076e-05,
      "loss": 1.1362,
      "step": 4182
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3262289848366616e-05,
      "loss": 1.1624,
      "step": 4183
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3242680314639993e-05,
      "loss": 1.3805,
      "step": 4184
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.322308307578633e-05,
      "loss": 1.1551,
      "step": 4185
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.3203498138360693e-05,
      "loss": 1.1333,
      "step": 4186
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.318392550891398e-05,
      "loss": 1.1299,
      "step": 4187
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.3164365193993016e-05,
      "loss": 1.159,
      "step": 4188
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.314481720014048e-05,
      "loss": 1.1642,
      "step": 4189
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.3125281533894912e-05,
      "loss": 1.1907,
      "step": 4190
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.3105758201790791e-05,
      "loss": 1.2737,
      "step": 4191
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.3086247210358398e-05,
      "loss": 0.9054,
      "step": 4192
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.3066748566123955e-05,
      "loss": 1.1547,
      "step": 4193
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.3047262275609483e-05,
      "loss": 1.1962,
      "step": 4194
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.3027788345332947e-05,
      "loss": 1.097,
      "step": 4195
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.30083267818081e-05,
      "loss": 1.299,
      "step": 4196
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.2988877591544624e-05,
      "loss": 0.9234,
      "step": 4197
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.2969440781048014e-05,
      "loss": 1.1286,
      "step": 4198
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.2950016356819656e-05,
      "loss": 1.2676,
      "step": 4199
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.2930604325356794e-05,
      "loss": 0.8439,
      "step": 4200
    },
    {
      "epoch": 2.3,
      "eval_loss": 1.3490395545959473,
      "eval_runtime": 36.9356,
      "eval_samples_per_second": 3.817,
      "eval_steps_per_second": 1.922,
      "step": 4200
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.2911204693152484e-05,
      "loss": 1.1589,
      "step": 4201
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.2891817466695705e-05,
      "loss": 1.1842,
      "step": 4202
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.2872442652471206e-05,
      "loss": 1.0346,
      "step": 4203
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.2853080256959655e-05,
      "loss": 1.2236,
      "step": 4204
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2833730286637514e-05,
      "loss": 1.2574,
      "step": 4205
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2814392747977139e-05,
      "loss": 1.0559,
      "step": 4206
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2795067647446674e-05,
      "loss": 1.234,
      "step": 4207
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.277575499151013e-05,
      "loss": 1.1502,
      "step": 4208
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.275645478662737e-05,
      "loss": 1.3326,
      "step": 4209
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2737167039254055e-05,
      "loss": 1.3096,
      "step": 4210
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2717891755841722e-05,
      "loss": 1.1984,
      "step": 4211
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2698628942837699e-05,
      "loss": 1.2467,
      "step": 4212
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2679378606685177e-05,
      "loss": 1.2435,
      "step": 4213
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2660140753823135e-05,
      "loss": 0.9712,
      "step": 4214
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2640915390686426e-05,
      "loss": 1.1814,
      "step": 4215
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2621702523705676e-05,
      "loss": 1.1514,
      "step": 4216
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2602502159307372e-05,
      "loss": 1.2863,
      "step": 4217
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2583314303913778e-05,
      "loss": 1.3162,
      "step": 4218
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2564138963943028e-05,
      "loss": 1.2525,
      "step": 4219
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2544976145809018e-05,
      "loss": 1.0678,
      "step": 4220
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2525825855921463e-05,
      "loss": 1.1169,
      "step": 4221
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.2506688100685931e-05,
      "loss": 1.2562,
      "step": 4222
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2487562886503734e-05,
      "loss": 1.0627,
      "step": 4223
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2468450219772054e-05,
      "loss": 1.0061,
      "step": 4224
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2449350106883818e-05,
      "loss": 1.1214,
      "step": 4225
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.24302625542278e-05,
      "loss": 1.1884,
      "step": 4226
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2411187568188532e-05,
      "loss": 1.1897,
      "step": 4227
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2392125155146383e-05,
      "loss": 1.1455,
      "step": 4228
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2373075321477479e-05,
      "loss": 1.3092,
      "step": 4229
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2354038073553776e-05,
      "loss": 0.9628,
      "step": 4230
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2335013417742969e-05,
      "loss": 0.9216,
      "step": 4231
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2316001360408614e-05,
      "loss": 1.0602,
      "step": 4232
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2297001907909983e-05,
      "loss": 1.1874,
      "step": 4233
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.227801506660215e-05,
      "loss": 1.0751,
      "step": 4234
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2259040842836018e-05,
      "loss": 1.3944,
      "step": 4235
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2240079242958191e-05,
      "loss": 1.1219,
      "step": 4236
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2221130273311126e-05,
      "loss": 1.1489,
      "step": 4237
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2202193940232993e-05,
      "loss": 1.163,
      "step": 4238
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2183270250057794e-05,
      "loss": 1.2054,
      "step": 4239
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.2164359209115234e-05,
      "loss": 1.152,
      "step": 4240
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.2145460823730864e-05,
      "loss": 1.3554,
      "step": 4241
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.2126575100225929e-05,
      "loss": 1.2804,
      "step": 4242
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.2107702044917496e-05,
      "loss": 1.0844,
      "step": 4243
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.2088841664118367e-05,
      "loss": 1.1371,
      "step": 4244
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.2069993964137082e-05,
      "loss": 1.0034,
      "step": 4245
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.2051158951278008e-05,
      "loss": 1.0522,
      "step": 4246
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.2032336631841185e-05,
      "loss": 1.0691,
      "step": 4247
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.2013527012122478e-05,
      "loss": 1.2541,
      "step": 4248
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1994730098413454e-05,
      "loss": 1.2203,
      "step": 4249
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1975945897001472e-05,
      "loss": 1.1302,
      "step": 4250
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1957174414169592e-05,
      "loss": 1.2403,
      "step": 4251
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1938415656196672e-05,
      "loss": 1.185,
      "step": 4252
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1919669629357256e-05,
      "loss": 1.2487,
      "step": 4253
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1900936339921693e-05,
      "loss": 1.0674,
      "step": 4254
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1882215794156005e-05,
      "loss": 1.0011,
      "step": 4255
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.186350799832202e-05,
      "loss": 1.0881,
      "step": 4256
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1844812958677248e-05,
      "loss": 1.1368,
      "step": 4257
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1826130681474939e-05,
      "loss": 1.1959,
      "step": 4258
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1807461172964108e-05,
      "loss": 1.1316,
      "step": 4259
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1788804439389457e-05,
      "loss": 1.2099,
      "step": 4260
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1770160486991461e-05,
      "loss": 1.1815,
      "step": 4261
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.175152932200626e-05,
      "loss": 1.1913,
      "step": 4262
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1732910950665782e-05,
      "loss": 1.1367,
      "step": 4263
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1714305379197615e-05,
      "loss": 1.1911,
      "step": 4264
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1695712613825127e-05,
      "loss": 1.1285,
      "step": 4265
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1677132660767337e-05,
      "loss": 1.2102,
      "step": 4266
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.165856552623904e-05,
      "loss": 1.2003,
      "step": 4267
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1640011216450691e-05,
      "loss": 1.4088,
      "step": 4268
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1621469737608504e-05,
      "loss": 1.2831,
      "step": 4269
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1602941095914361e-05,
      "loss": 1.1621,
      "step": 4270
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.158442529756586e-05,
      "loss": 1.0212,
      "step": 4271
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1565922348756325e-05,
      "loss": 1.0913,
      "step": 4272
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.154743225567474e-05,
      "loss": 1.1487,
      "step": 4273
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.152895502450585e-05,
      "loss": 1.1764,
      "step": 4274
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.151049066143003e-05,
      "loss": 0.9963,
      "step": 4275
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1492039172623408e-05,
      "loss": 1.2095,
      "step": 4276
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1473600564257759e-05,
      "loss": 1.368,
      "step": 4277
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1455174842500593e-05,
      "loss": 1.2262,
      "step": 4278
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1436762013515068e-05,
      "loss": 1.16,
      "step": 4279
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1418362083460066e-05,
      "loss": 1.3496,
      "step": 4280
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1399975058490137e-05,
      "loss": 1.2277,
      "step": 4281
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1381600944755494e-05,
      "loss": 1.3971,
      "step": 4282
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1363239748402083e-05,
      "loss": 1.2024,
      "step": 4283
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1344891475571474e-05,
      "loss": 0.9508,
      "step": 4284
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1326556132400962e-05,
      "loss": 1.0122,
      "step": 4285
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1308233725023465e-05,
      "loss": 1.2093,
      "step": 4286
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1289924259567642e-05,
      "loss": 1.2656,
      "step": 4287
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1271627742157742e-05,
      "loss": 1.1511,
      "step": 4288
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1253344178913772e-05,
      "loss": 1.279,
      "step": 4289
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1235073575951316e-05,
      "loss": 1.1357,
      "step": 4290
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1216815939381704e-05,
      "loss": 1.1715,
      "step": 4291
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1198571275311858e-05,
      "loss": 1.0413,
      "step": 4292
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1180339589844424e-05,
      "loss": 1.0171,
      "step": 4293
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1162120889077665e-05,
      "loss": 1.0317,
      "step": 4294
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.1143915179105502e-05,
      "loss": 1.0055,
      "step": 4295
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.1125722466017547e-05,
      "loss": 1.1341,
      "step": 4296
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.1107542755899004e-05,
      "loss": 1.011,
      "step": 4297
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.1089376054830803e-05,
      "loss": 1.1138,
      "step": 4298
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.107122236888945e-05,
      "loss": 1.0076,
      "step": 4299
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.1053081704147162e-05,
      "loss": 1.2749,
      "step": 4300
    },
    {
      "epoch": 2.36,
      "eval_loss": 1.3507847785949707,
      "eval_runtime": 36.7665,
      "eval_samples_per_second": 3.835,
      "eval_steps_per_second": 1.931,
      "step": 4300
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.1034954066671733e-05,
      "loss": 1.2795,
      "step": 4301
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.1016839462526668e-05,
      "loss": 0.8109,
      "step": 4302
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0998737897771056e-05,
      "loss": 1.3081,
      "step": 4303
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0980649378459668e-05,
      "loss": 1.2137,
      "step": 4304
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.096257391064287e-05,
      "loss": 1.3515,
      "step": 4305
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0944511500366716e-05,
      "loss": 1.2862,
      "step": 4306
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.092646215367284e-05,
      "loss": 1.2773,
      "step": 4307
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.090842587659851e-05,
      "loss": 1.222,
      "step": 4308
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0890402675176681e-05,
      "loss": 0.9897,
      "step": 4309
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0872392555435856e-05,
      "loss": 1.259,
      "step": 4310
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0854395523400224e-05,
      "loss": 1.1939,
      "step": 4311
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.083641158508955e-05,
      "loss": 1.1453,
      "step": 4312
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.081844074651927e-05,
      "loss": 1.2405,
      "step": 4313
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.080048301370038e-05,
      "loss": 0.9867,
      "step": 4314
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.078253839263954e-05,
      "loss": 1.4107,
      "step": 4315
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0764606889338995e-05,
      "loss": 1.3759,
      "step": 4316
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0746688509796626e-05,
      "loss": 1.1857,
      "step": 4317
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0728783260005904e-05,
      "loss": 1.3521,
      "step": 4318
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0710891145955904e-05,
      "loss": 1.19,
      "step": 4319
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0693012173631344e-05,
      "loss": 1.348,
      "step": 4320
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0675146349012489e-05,
      "loss": 1.2458,
      "step": 4321
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0657293678075275e-05,
      "loss": 0.9657,
      "step": 4322
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0639454166791163e-05,
      "loss": 1.0457,
      "step": 4323
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0621627821127289e-05,
      "loss": 1.1058,
      "step": 4324
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0603814647046317e-05,
      "loss": 1.1289,
      "step": 4325
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.058601465050656e-05,
      "loss": 1.3411,
      "step": 4326
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.056822783746187e-05,
      "loss": 1.1013,
      "step": 4327
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.055045421386175e-05,
      "loss": 1.3095,
      "step": 4328
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0532693785651227e-05,
      "loss": 1.2116,
      "step": 4329
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0514946558770977e-05,
      "loss": 1.2466,
      "step": 4330
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.049721253915722e-05,
      "loss": 1.3715,
      "step": 4331
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0479491732741747e-05,
      "loss": 1.358,
      "step": 4332
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0461784145451986e-05,
      "loss": 0.9568,
      "step": 4333
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.044408978321087e-05,
      "loss": 1.1508,
      "step": 4334
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0426408651936986e-05,
      "loss": 1.2516,
      "step": 4335
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0408740757544416e-05,
      "loss": 1.1602,
      "step": 4336
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0391086105942893e-05,
      "loss": 1.1548,
      "step": 4337
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0373444703037644e-05,
      "loss": 1.0856,
      "step": 4338
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0355816554729536e-05,
      "loss": 1.0068,
      "step": 4339
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0338201666914943e-05,
      "loss": 1.0973,
      "step": 4340
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0320600045485851e-05,
      "loss": 1.1785,
      "step": 4341
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0303011696329756e-05,
      "loss": 1.4103,
      "step": 4342
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0285436625329786e-05,
      "loss": 1.156,
      "step": 4343
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.026787483836456e-05,
      "loss": 1.1208,
      "step": 4344
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0250326341308276e-05,
      "loss": 0.9913,
      "step": 4345
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0232791140030713e-05,
      "loss": 1.0152,
      "step": 4346
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0215269240397151e-05,
      "loss": 1.4133,
      "step": 4347
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0197760648268484e-05,
      "loss": 1.3601,
      "step": 4348
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0180265369501097e-05,
      "loss": 1.1231,
      "step": 4349
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0162783409946964e-05,
      "loss": 1.4296,
      "step": 4350
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0145314775453563e-05,
      "loss": 1.4141,
      "step": 4351
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.012785947186397e-05,
      "loss": 1.3599,
      "step": 4352
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0110417505016734e-05,
      "loss": 1.1695,
      "step": 4353
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0092988880746018e-05,
      "loss": 1.2472,
      "step": 4354
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.007557360488146e-05,
      "loss": 1.1553,
      "step": 4355
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0058171683248246e-05,
      "loss": 1.1048,
      "step": 4356
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0040783121667136e-05,
      "loss": 1.2142,
      "step": 4357
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0023407925954353e-05,
      "loss": 1.2773,
      "step": 4358
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.000604610192173e-05,
      "loss": 1.3741,
      "step": 4359
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.988697655376544e-06,
      "loss": 1.3636,
      "step": 4360
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.971362592121668e-06,
      "loss": 1.3022,
      "step": 4361
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.954040917955449e-06,
      "loss": 1.1372,
      "step": 4362
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.936732638671787e-06,
      "loss": 1.1291,
      "step": 4363
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.919437760060073e-06,
      "loss": 1.2784,
      "step": 4364
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.902156287905262e-06,
      "loss": 1.3431,
      "step": 4365
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.88488822798776e-06,
      "loss": 1.1614,
      "step": 4366
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.867633586083552e-06,
      "loss": 1.1368,
      "step": 4367
    },
    {
      "epoch": 2.39,
      "learning_rate": 9.850392367964085e-06,
      "loss": 1.0494,
      "step": 4368
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.833164579396315e-06,
      "loss": 1.0067,
      "step": 4369
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.815950226142778e-06,
      "loss": 1.2263,
      "step": 4370
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.798749313961424e-06,
      "loss": 1.1567,
      "step": 4371
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.781561848605775e-06,
      "loss": 1.0265,
      "step": 4372
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.764387835824796e-06,
      "loss": 1.2595,
      "step": 4373
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.74722728136302e-06,
      "loss": 1.1666,
      "step": 4374
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.730080190960417e-06,
      "loss": 1.2178,
      "step": 4375
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.712946570352472e-06,
      "loss": 1.14,
      "step": 4376
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.695826425270198e-06,
      "loss": 1.286,
      "step": 4377
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.678719761440047e-06,
      "loss": 0.9206,
      "step": 4378
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.661626584584015e-06,
      "loss": 1.1745,
      "step": 4379
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.644546900419533e-06,
      "loss": 1.1125,
      "step": 4380
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.627480714659571e-06,
      "loss": 1.1943,
      "step": 4381
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.61042803301253e-06,
      "loss": 1.1239,
      "step": 4382
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.593388861182356e-06,
      "loss": 1.1512,
      "step": 4383
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.576363204868416e-06,
      "loss": 1.2561,
      "step": 4384
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.559351069765609e-06,
      "loss": 1.072,
      "step": 4385
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.54235246156427e-06,
      "loss": 1.1561,
      "step": 4386
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.525367385950207e-06,
      "loss": 0.8836,
      "step": 4387
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.508395848604756e-06,
      "loss": 1.2036,
      "step": 4388
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.49143785520466e-06,
      "loss": 1.2631,
      "step": 4389
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.474493411422186e-06,
      "loss": 1.3101,
      "step": 4390
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.457562522925013e-06,
      "loss": 1.1241,
      "step": 4391
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.44064519537634e-06,
      "loss": 1.1018,
      "step": 4392
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.42374143443479e-06,
      "loss": 1.0042,
      "step": 4393
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.406851245754477e-06,
      "loss": 1.1517,
      "step": 4394
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.38997463498495e-06,
      "loss": 1.2589,
      "step": 4395
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.37311160777124e-06,
      "loss": 0.7844,
      "step": 4396
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.35626216975381e-06,
      "loss": 1.1829,
      "step": 4397
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.339426326568606e-06,
      "loss": 1.2388,
      "step": 4398
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.322604083847002e-06,
      "loss": 1.3472,
      "step": 4399
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.305795447215827e-06,
      "loss": 0.9578,
      "step": 4400
    },
    {
      "epoch": 2.41,
      "eval_loss": 1.350223183631897,
      "eval_runtime": 36.9145,
      "eval_samples_per_second": 3.82,
      "eval_steps_per_second": 1.923,
      "step": 4400
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.289000422297373e-06,
      "loss": 1.2599,
      "step": 4401
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.27221901470936e-06,
      "loss": 1.3843,
      "step": 4402
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.255451230064983e-06,
      "loss": 1.1445,
      "step": 4403
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.23869707397283e-06,
      "loss": 1.3498,
      "step": 4404
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.221956552036992e-06,
      "loss": 1.2874,
      "step": 4405
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.205229669856947e-06,
      "loss": 1.3218,
      "step": 4406
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.18851643302765e-06,
      "loss": 1.0363,
      "step": 4407
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.171816847139448e-06,
      "loss": 1.1459,
      "step": 4408
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.155130917778181e-06,
      "loss": 1.2191,
      "step": 4409
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.138458650525055e-06,
      "loss": 1.2886,
      "step": 4410
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.12180005095677e-06,
      "loss": 1.4623,
      "step": 4411
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.105155124645403e-06,
      "loss": 1.2151,
      "step": 4412
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.088523877158473e-06,
      "loss": 1.1601,
      "step": 4413
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.071906314058948e-06,
      "loss": 1.0678,
      "step": 4414
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.055302440905177e-06,
      "loss": 1.0694,
      "step": 4415
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.03871226325098e-06,
      "loss": 1.1103,
      "step": 4416
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.022135786645541e-06,
      "loss": 1.2814,
      "step": 4417
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.005573016633511e-06,
      "loss": 1.3763,
      "step": 4418
    },
    {
      "epoch": 2.42,
      "learning_rate": 8.98902395875491e-06,
      "loss": 1.0413,
      "step": 4419
    },
    {
      "epoch": 2.42,
      "learning_rate": 8.972488618545222e-06,
      "loss": 1.1046,
      "step": 4420
    },
    {
      "epoch": 2.42,
      "learning_rate": 8.955967001535286e-06,
      "loss": 1.3721,
      "step": 4421
    },
    {
      "epoch": 2.42,
      "learning_rate": 8.939459113251408e-06,
      "loss": 1.0899,
      "step": 4422
    },
    {
      "epoch": 2.42,
      "learning_rate": 8.922964959215264e-06,
      "loss": 1.2505,
      "step": 4423
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.906484544943932e-06,
      "loss": 1.4849,
      "step": 4424
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.890017875949935e-06,
      "loss": 1.3465,
      "step": 4425
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.873564957741144e-06,
      "loss": 0.9585,
      "step": 4426
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.857125795820882e-06,
      "loss": 0.9,
      "step": 4427
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.840700395687823e-06,
      "loss": 1.1151,
      "step": 4428
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.824288762836097e-06,
      "loss": 1.1073,
      "step": 4429
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.807890902755156e-06,
      "loss": 1.3417,
      "step": 4430
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.791506820929918e-06,
      "loss": 1.0085,
      "step": 4431
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.775136522840621e-06,
      "loss": 0.9367,
      "step": 4432
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.758780013962969e-06,
      "loss": 1.1751,
      "step": 4433
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.742437299767975e-06,
      "loss": 1.0277,
      "step": 4434
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.726108385722109e-06,
      "loss": 1.1799,
      "step": 4435
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.70979327728718e-06,
      "loss": 1.2462,
      "step": 4436
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.693491979920381e-06,
      "loss": 1.1197,
      "step": 4437
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.677204499074314e-06,
      "loss": 1.2756,
      "step": 4438
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.660930840196923e-06,
      "loss": 1.1224,
      "step": 4439
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.644671008731569e-06,
      "loss": 1.3161,
      "step": 4440
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.628425010116947e-06,
      "loss": 1.3876,
      "step": 4441
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.61219284978716e-06,
      "loss": 1.2725,
      "step": 4442
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.595974533171652e-06,
      "loss": 1.4529,
      "step": 4443
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.579770065695264e-06,
      "loss": 1.1422,
      "step": 4444
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.56357945277818e-06,
      "loss": 0.9594,
      "step": 4445
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.547402699835976e-06,
      "loss": 0.8425,
      "step": 4446
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.531239812279562e-06,
      "loss": 0.9293,
      "step": 4447
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.515090795515247e-06,
      "loss": 1.0622,
      "step": 4448
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.498955654944657e-06,
      "loss": 1.0812,
      "step": 4449
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.4828343959648e-06,
      "loss": 1.326,
      "step": 4450
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.466727023968063e-06,
      "loss": 1.1865,
      "step": 4451
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.450633544342134e-06,
      "loss": 1.1587,
      "step": 4452
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.43455396247011e-06,
      "loss": 1.1181,
      "step": 4453
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.418488283730392e-06,
      "loss": 1.3828,
      "step": 4454
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.402436513496775e-06,
      "loss": 1.1948,
      "step": 4455
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.386398657138356e-06,
      "loss": 1.1884,
      "step": 4456
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.370374720019631e-06,
      "loss": 1.4686,
      "step": 4457
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.354364707500379e-06,
      "loss": 1.1778,
      "step": 4458
    },
    {
      "epoch": 2.44,
      "learning_rate": 8.338368624935778e-06,
      "loss": 1.1994,
      "step": 4459
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.322386477676308e-06,
      "loss": 1.4324,
      "step": 4460
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.306418271067795e-06,
      "loss": 1.0223,
      "step": 4461
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.290464010451432e-06,
      "loss": 1.3804,
      "step": 4462
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.274523701163694e-06,
      "loss": 1.25,
      "step": 4463
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.25859734853645e-06,
      "loss": 1.1065,
      "step": 4464
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.24268495789684e-06,
      "loss": 1.1645,
      "step": 4465
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.226786534567393e-06,
      "loss": 1.395,
      "step": 4466
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.210902083865906e-06,
      "loss": 1.2112,
      "step": 4467
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.195031611105564e-06,
      "loss": 1.3015,
      "step": 4468
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.179175121594818e-06,
      "loss": 1.1276,
      "step": 4469
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.163332620637493e-06,
      "loss": 1.0529,
      "step": 4470
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.147504113532683e-06,
      "loss": 1.1338,
      "step": 4471
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.131689605574866e-06,
      "loss": 0.8845,
      "step": 4472
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.115889102053782e-06,
      "loss": 1.0735,
      "step": 4473
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.10010260825449e-06,
      "loss": 1.029,
      "step": 4474
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.084330129457412e-06,
      "loss": 1.1257,
      "step": 4475
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.068571670938218e-06,
      "loss": 0.8912,
      "step": 4476
    },
    {
      "epoch": 2.45,
      "learning_rate": 8.05282723796794e-06,
      "loss": 1.2875,
      "step": 4477
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.037096835812885e-06,
      "loss": 1.3981,
      "step": 4478
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.021380469734695e-06,
      "loss": 1.2037,
      "step": 4479
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.00567814499028e-06,
      "loss": 1.1612,
      "step": 4480
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.989989866831898e-06,
      "loss": 1.0144,
      "step": 4481
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.974315640507064e-06,
      "loss": 0.9789,
      "step": 4482
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.958655471258641e-06,
      "loss": 1.324,
      "step": 4483
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.943009364324733e-06,
      "loss": 1.1746,
      "step": 4484
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.927377324938801e-06,
      "loss": 1.282,
      "step": 4485
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.911759358329552e-06,
      "loss": 1.1489,
      "step": 4486
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.896155469721e-06,
      "loss": 1.0015,
      "step": 4487
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.880565664332474e-06,
      "loss": 1.2489,
      "step": 4488
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.864989947378548e-06,
      "loss": 1.0651,
      "step": 4489
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.849428324069142e-06,
      "loss": 0.9491,
      "step": 4490
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.833880799609389e-06,
      "loss": 1.037,
      "step": 4491
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.81834737919978e-06,
      "loss": 1.1108,
      "step": 4492
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.802828068036028e-06,
      "loss": 0.9012,
      "step": 4493
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.787322871309172e-06,
      "loss": 1.1575,
      "step": 4494
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.771831794205492e-06,
      "loss": 1.1942,
      "step": 4495
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.756354841906582e-06,
      "loss": 1.0748,
      "step": 4496
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.740892019589286e-06,
      "loss": 1.323,
      "step": 4497
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.725443332425719e-06,
      "loss": 1.2022,
      "step": 4498
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.71000878558329e-06,
      "loss": 1.1125,
      "step": 4499
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.694588384224655e-06,
      "loss": 1.2203,
      "step": 4500
    },
    {
      "epoch": 2.47,
      "eval_loss": 1.3496108055114746,
      "eval_runtime": 36.9376,
      "eval_samples_per_second": 3.817,
      "eval_steps_per_second": 1.922,
      "step": 4500
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.679182133507761e-06,
      "loss": 1.1351,
      "step": 4501
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.663790038585793e-06,
      "loss": 1.3209,
      "step": 4502
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.648412104607245e-06,
      "loss": 0.92,
      "step": 4503
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.633048336715814e-06,
      "loss": 0.9773,
      "step": 4504
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.617698740050527e-06,
      "loss": 1.2033,
      "step": 4505
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.602363319745609e-06,
      "loss": 1.264,
      "step": 4506
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.587042080930584e-06,
      "loss": 1.2509,
      "step": 4507
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.571735028730209e-06,
      "loss": 1.2509,
      "step": 4508
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.5564421682645215e-06,
      "loss": 1.114,
      "step": 4509
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.5411635046487794e-06,
      "loss": 1.2,
      "step": 4510
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.525899042993512e-06,
      "loss": 1.1366,
      "step": 4511
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.510648788404501e-06,
      "loss": 1.3159,
      "step": 4512
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.495412745982761e-06,
      "loss": 1.1203,
      "step": 4513
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.4801909208245706e-06,
      "loss": 1.135,
      "step": 4514
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.4649833180214294e-06,
      "loss": 1.2854,
      "step": 4515
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.449789942660118e-06,
      "loss": 1.2489,
      "step": 4516
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.434610799822611e-06,
      "loss": 1.3695,
      "step": 4517
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.419445894586163e-06,
      "loss": 1.2408,
      "step": 4518
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.404295232023228e-06,
      "loss": 1.2043,
      "step": 4519
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.389158817201542e-06,
      "loss": 1.1676,
      "step": 4520
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.37403665518403e-06,
      "loss": 1.0893,
      "step": 4521
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.358928751028887e-06,
      "loss": 1.2435,
      "step": 4522
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.343835109789515e-06,
      "loss": 1.1005,
      "step": 4523
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.3287557365145365e-06,
      "loss": 0.9936,
      "step": 4524
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.313690636247844e-06,
      "loss": 1.339,
      "step": 4525
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.2986398140285015e-06,
      "loss": 1.2615,
      "step": 4526
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.283603274890849e-06,
      "loss": 1.2153,
      "step": 4527
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.268581023864402e-06,
      "loss": 1.0642,
      "step": 4528
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.253573065973945e-06,
      "loss": 1.0801,
      "step": 4529
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.238579406239432e-06,
      "loss": 1.1703,
      "step": 4530
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.2236000496760825e-06,
      "loss": 1.1795,
      "step": 4531
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.208635001294278e-06,
      "loss": 1.1167,
      "step": 4532
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.193684266099671e-06,
      "loss": 1.3063,
      "step": 4533
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.178747849093093e-06,
      "loss": 1.0052,
      "step": 4534
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.16382575527057e-06,
      "loss": 1.0173,
      "step": 4535
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.1489179896233885e-06,
      "loss": 1.1573,
      "step": 4536
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.134024557137997e-06,
      "loss": 1.2848,
      "step": 4537
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.119145462796079e-06,
      "loss": 1.2626,
      "step": 4538
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.104280711574474e-06,
      "loss": 1.2334,
      "step": 4539
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.089430308445316e-06,
      "loss": 1.0221,
      "step": 4540
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.07459425837585e-06,
      "loss": 1.2087,
      "step": 4541
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.059772566328543e-06,
      "loss": 1.0794,
      "step": 4542
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.044965237261097e-06,
      "loss": 1.3208,
      "step": 4543
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.030172276126351e-06,
      "loss": 1.215,
      "step": 4544
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.015393687872401e-06,
      "loss": 1.0331,
      "step": 4545
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.000629477442478e-06,
      "loss": 1.3234,
      "step": 4546
    },
    {
      "epoch": 2.49,
      "learning_rate": 6.985879649775051e-06,
      "loss": 1.1053,
      "step": 4547
    },
    {
      "epoch": 2.49,
      "learning_rate": 6.9711442098037375e-06,
      "loss": 1.2625,
      "step": 4548
    },
    {
      "epoch": 2.49,
      "learning_rate": 6.956423162457376e-06,
      "loss": 1.1519,
      "step": 4549
    },
    {
      "epoch": 2.49,
      "learning_rate": 6.941716512659957e-06,
      "loss": 1.1266,
      "step": 4550
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.9270242653307005e-06,
      "loss": 1.1469,
      "step": 4551
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.912346425383964e-06,
      "loss": 1.264,
      "step": 4552
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.897682997729316e-06,
      "loss": 1.07,
      "step": 4553
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.883033987271492e-06,
      "loss": 0.9305,
      "step": 4554
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.868399398910391e-06,
      "loss": 1.0439,
      "step": 4555
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.85377923754113e-06,
      "loss": 1.174,
      "step": 4556
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.839173508053947e-06,
      "loss": 1.1148,
      "step": 4557
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.824582215334302e-06,
      "loss": 0.9149,
      "step": 4558
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.810005364262784e-06,
      "loss": 1.0667,
      "step": 4559
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.795442959715192e-06,
      "loss": 0.9927,
      "step": 4560
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.780895006562449e-06,
      "loss": 1.4555,
      "step": 4561
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.766361509670688e-06,
      "loss": 1.1572,
      "step": 4562
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.751842473901166e-06,
      "loss": 1.2981,
      "step": 4563
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.737337904110341e-06,
      "loss": 1.1536,
      "step": 4564
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.722847805149812e-06,
      "loss": 1.2184,
      "step": 4565
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.708372181866318e-06,
      "loss": 1.117,
      "step": 4566
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.693911039101803e-06,
      "loss": 1.122,
      "step": 4567
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.679464381693323e-06,
      "loss": 0.9688,
      "step": 4568
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.665032214473127e-06,
      "loss": 1.099,
      "step": 4569
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.650614542268574e-06,
      "loss": 1.0224,
      "step": 4570
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.636211369902229e-06,
      "loss": 1.0877,
      "step": 4571
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.621822702191743e-06,
      "loss": 1.0677,
      "step": 4572
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.607448543949979e-06,
      "loss": 1.2572,
      "step": 4573
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.593088899984895e-06,
      "loss": 0.9206,
      "step": 4574
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.5787437750996325e-06,
      "loss": 1.1421,
      "step": 4575
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.564413174092443e-06,
      "loss": 1.1782,
      "step": 4576
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.5500971017567584e-06,
      "loss": 1.1725,
      "step": 4577
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.535795562881114e-06,
      "loss": 1.3758,
      "step": 4578
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.521508562249196e-06,
      "loss": 1.0178,
      "step": 4579
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.507236104639841e-06,
      "loss": 1.208,
      "step": 4580
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.492978194827004e-06,
      "loss": 1.4196,
      "step": 4581
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.478734837579792e-06,
      "loss": 1.24,
      "step": 4582
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.464506037662416e-06,
      "loss": 1.2915,
      "step": 4583
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.4502917998342575e-06,
      "loss": 1.2789,
      "step": 4584
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.43609212884978e-06,
      "loss": 1.0477,
      "step": 4585
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.421907029458624e-06,
      "loss": 1.1113,
      "step": 4586
    },
    {
      "epoch": 2.51,
      "learning_rate": 6.407736506405509e-06,
      "loss": 1.2245,
      "step": 4587
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.3935805644303184e-06,
      "loss": 0.9357,
      "step": 4588
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.3794392082680285e-06,
      "loss": 1.1854,
      "step": 4589
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.365312442648769e-06,
      "loss": 1.2002,
      "step": 4590
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.351200272297753e-06,
      "loss": 1.2121,
      "step": 4591
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.337102701935321e-06,
      "loss": 1.0492,
      "step": 4592
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.3230197362769585e-06,
      "loss": 1.2695,
      "step": 4593
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.3089513800332286e-06,
      "loss": 1.309,
      "step": 4594
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.2948976379098416e-06,
      "loss": 1.1784,
      "step": 4595
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.28085851460758e-06,
      "loss": 1.2486,
      "step": 4596
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.266834014822376e-06,
      "loss": 1.187,
      "step": 4597
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.2528241432452394e-06,
      "loss": 1.2707,
      "step": 4598
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.238828904562316e-06,
      "loss": 1.2714,
      "step": 4599
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.2248483034548276e-06,
      "loss": 0.9451,
      "step": 4600
    },
    {
      "epoch": 2.52,
      "eval_loss": 1.3497511148452759,
      "eval_runtime": 36.882,
      "eval_samples_per_second": 3.823,
      "eval_steps_per_second": 1.925,
      "step": 4600
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.210882344599128e-06,
      "loss": 1.1801,
      "step": 4601
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.196931032666648e-06,
      "loss": 1.0118,
      "step": 4602
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.182994372323925e-06,
      "loss": 1.3247,
      "step": 4603
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.16907236823262e-06,
      "loss": 1.3121,
      "step": 4604
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.155165025049453e-06,
      "loss": 1.3266,
      "step": 4605
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.14127234742628e-06,
      "loss": 1.124,
      "step": 4606
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.127394340010006e-06,
      "loss": 1.0274,
      "step": 4607
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.11353100744268e-06,
      "loss": 1.5834,
      "step": 4608
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.099682354361397e-06,
      "loss": 1.2576,
      "step": 4609
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.085848385398385e-06,
      "loss": 1.0809,
      "step": 4610
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.07202910518091e-06,
      "loss": 0.8983,
      "step": 4611
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.058224518331374e-06,
      "loss": 1.1902,
      "step": 4612
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.044434629467216e-06,
      "loss": 1.0484,
      "step": 4613
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.03065944320102e-06,
      "loss": 1.1623,
      "step": 4614
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.016898964140394e-06,
      "loss": 1.0922,
      "step": 4615
    },
    {
      "epoch": 2.53,
      "learning_rate": 6.003153196888045e-06,
      "loss": 1.2446,
      "step": 4616
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.989422146041784e-06,
      "loss": 1.2106,
      "step": 4617
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.97570581619446e-06,
      "loss": 1.2625,
      "step": 4618
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.96200421193403e-06,
      "loss": 1.177,
      "step": 4619
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.948317337843501e-06,
      "loss": 1.0684,
      "step": 4620
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.9346451985009825e-06,
      "loss": 1.3593,
      "step": 4621
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.920987798479616e-06,
      "loss": 1.0403,
      "step": 4622
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.9073451423476535e-06,
      "loss": 1.1246,
      "step": 4623
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.893717234668383e-06,
      "loss": 1.2769,
      "step": 4624
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.8801040800001815e-06,
      "loss": 1.0296,
      "step": 4625
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.866505682896472e-06,
      "loss": 1.0483,
      "step": 4626
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.852922047905767e-06,
      "loss": 1.1013,
      "step": 4627
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.839353179571616e-06,
      "loss": 1.3506,
      "step": 4628
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.8257990824326314e-06,
      "loss": 1.0212,
      "step": 4629
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.812259761022509e-06,
      "loss": 1.2131,
      "step": 4630
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.7987352198699715e-06,
      "loss": 1.0425,
      "step": 4631
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.785225463498828e-06,
      "loss": 1.1607,
      "step": 4632
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.771730496427907e-06,
      "loss": 1.0756,
      "step": 4633
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.758250323171127e-06,
      "loss": 1.3756,
      "step": 4634
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.744784948237425e-06,
      "loss": 1.0209,
      "step": 4635
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.731334376130826e-06,
      "loss": 1.2276,
      "step": 4636
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.717898611350358e-06,
      "loss": 1.3349,
      "step": 4637
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.704477658390139e-06,
      "loss": 1.177,
      "step": 4638
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.6910715217393e-06,
      "loss": 1.0557,
      "step": 4639
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.67768020588203e-06,
      "loss": 1.1914,
      "step": 4640
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.66430371529757e-06,
      "loss": 1.0945,
      "step": 4641
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.650942054460173e-06,
      "loss": 1.0917,
      "step": 4642
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.6375952278391765e-06,
      "loss": 0.9477,
      "step": 4643
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.6242632398989085e-06,
      "loss": 1.1471,
      "step": 4644
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.610946095098773e-06,
      "loss": 1.12,
      "step": 4645
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.597643797893176e-06,
      "loss": 1.3741,
      "step": 4646
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.58435635273159e-06,
      "loss": 1.3397,
      "step": 4647
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.571083764058482e-06,
      "loss": 1.1313,
      "step": 4648
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.557826036313391e-06,
      "loss": 1.2062,
      "step": 4649
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.544583173930839e-06,
      "loss": 1.091,
      "step": 4650
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.531355181340436e-06,
      "loss": 1.1741,
      "step": 4651
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.518142062966758e-06,
      "loss": 0.9172,
      "step": 4652
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.504943823229431e-06,
      "loss": 0.815,
      "step": 4653
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.491760466543117e-06,
      "loss": 1.082,
      "step": 4654
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.478591997317472e-06,
      "loss": 1.146,
      "step": 4655
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.465438419957208e-06,
      "loss": 1.2247,
      "step": 4656
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.452299738862015e-06,
      "loss": 1.2607,
      "step": 4657
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.439175958426634e-06,
      "loss": 1.213,
      "step": 4658
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.426067083040803e-06,
      "loss": 1.1403,
      "step": 4659
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.412973117089287e-06,
      "loss": 1.2881,
      "step": 4660
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.399894064951838e-06,
      "loss": 1.1139,
      "step": 4661
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.386829931003268e-06,
      "loss": 0.9797,
      "step": 4662
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.373780719613342e-06,
      "loss": 1.3781,
      "step": 4663
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.360746435146885e-06,
      "loss": 1.0257,
      "step": 4664
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.347727081963694e-06,
      "loss": 1.2763,
      "step": 4665
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.334722664418573e-06,
      "loss": 1.3926,
      "step": 4666
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.3217331868613555e-06,
      "loss": 1.1155,
      "step": 4667
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.3087586536368525e-06,
      "loss": 1.3754,
      "step": 4668
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.2957990690849e-06,
      "loss": 1.5542,
      "step": 4669
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.2828544375403025e-06,
      "loss": 1.1355,
      "step": 4670
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.2699247633329e-06,
      "loss": 1.0423,
      "step": 4671
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.257010050787486e-06,
      "loss": 1.0929,
      "step": 4672
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.244110304223898e-06,
      "loss": 1.2663,
      "step": 4673
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.2312255279569236e-06,
      "loss": 1.3121,
      "step": 4674
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.2183557262963815e-06,
      "loss": 0.9919,
      "step": 4675
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.20550090354705e-06,
      "loss": 1.2039,
      "step": 4676
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.192661064008708e-06,
      "loss": 1.1611,
      "step": 4677
    },
    {
      "epoch": 2.56,
      "learning_rate": 5.179836211976135e-06,
      "loss": 1.098,
      "step": 4678
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.167026351739068e-06,
      "loss": 1.224,
      "step": 4679
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.154231487582273e-06,
      "loss": 1.198,
      "step": 4680
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.141451623785454e-06,
      "loss": 1.2601,
      "step": 4681
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.128686764623336e-06,
      "loss": 1.1564,
      "step": 4682
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.115936914365593e-06,
      "loss": 1.4455,
      "step": 4683
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.103202077276908e-06,
      "loss": 1.2033,
      "step": 4684
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.090482257616913e-06,
      "loss": 1.036,
      "step": 4685
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.077777459640248e-06,
      "loss": 1.3917,
      "step": 4686
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.065087687596492e-06,
      "loss": 1.0445,
      "step": 4687
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.05241294573024e-06,
      "loss": 0.9899,
      "step": 4688
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.039753238281026e-06,
      "loss": 1.1991,
      "step": 4689
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.027108569483363e-06,
      "loss": 1.2371,
      "step": 4690
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.014478943566748e-06,
      "loss": 1.3144,
      "step": 4691
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.0018643647556265e-06,
      "loss": 1.2209,
      "step": 4692
    },
    {
      "epoch": 2.57,
      "learning_rate": 4.98926483726943e-06,
      "loss": 1.0848,
      "step": 4693
    },
    {
      "epoch": 2.57,
      "learning_rate": 4.976680365322533e-06,
      "loss": 1.1364,
      "step": 4694
    },
    {
      "epoch": 2.57,
      "learning_rate": 4.964110953124307e-06,
      "loss": 1.2253,
      "step": 4695
    },
    {
      "epoch": 2.57,
      "learning_rate": 4.951556604879048e-06,
      "loss": 1.1288,
      "step": 4696
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.939017324786044e-06,
      "loss": 0.9869,
      "step": 4697
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.9264931170395155e-06,
      "loss": 1.4412,
      "step": 4698
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.913983985828674e-06,
      "loss": 1.1326,
      "step": 4699
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.901489935337661e-06,
      "loss": 0.9602,
      "step": 4700
    },
    {
      "epoch": 2.58,
      "eval_loss": 1.3490681648254395,
      "eval_runtime": 36.797,
      "eval_samples_per_second": 3.832,
      "eval_steps_per_second": 1.93,
      "step": 4700
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.8890109697455885e-06,
      "loss": 1.318,
      "step": 4701
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.876547093226513e-06,
      "loss": 1.1189,
      "step": 4702
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.864098309949439e-06,
      "loss": 1.1141,
      "step": 4703
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.851664624078356e-06,
      "loss": 1.2005,
      "step": 4704
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.839246039772155e-06,
      "loss": 1.1024,
      "step": 4705
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.826842561184719e-06,
      "loss": 1.3654,
      "step": 4706
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.814454192464846e-06,
      "loss": 1.3303,
      "step": 4707
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.802080937756304e-06,
      "loss": 1.1608,
      "step": 4708
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.789722801197788e-06,
      "loss": 1.2107,
      "step": 4709
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.777379786922937e-06,
      "loss": 1.1484,
      "step": 4710
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.765051899060363e-06,
      "loss": 1.1464,
      "step": 4711
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.752739141733564e-06,
      "loss": 1.0259,
      "step": 4712
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.740441519061029e-06,
      "loss": 1.0934,
      "step": 4713
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.728159035156143e-06,
      "loss": 0.9677,
      "step": 4714
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.7158916941272665e-06,
      "loss": 1.1931,
      "step": 4715
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.703639500077656e-06,
      "loss": 1.2802,
      "step": 4716
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.691402457105537e-06,
      "loss": 1.1381,
      "step": 4717
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.679180569304032e-06,
      "loss": 1.2012,
      "step": 4718
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.66697384076123e-06,
      "loss": 1.1678,
      "step": 4719
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.654782275560127e-06,
      "loss": 1.1598,
      "step": 4720
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.642605877778639e-06,
      "loss": 1.3308,
      "step": 4721
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.63044465148964e-06,
      "loss": 1.1476,
      "step": 4722
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.618298600760895e-06,
      "loss": 0.96,
      "step": 4723
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.606167729655131e-06,
      "loss": 1.2757,
      "step": 4724
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.594052042229952e-06,
      "loss": 1.0686,
      "step": 4725
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.581951542537921e-06,
      "loss": 1.3011,
      "step": 4726
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.569866234626502e-06,
      "loss": 0.8829,
      "step": 4727
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.557796122538089e-06,
      "loss": 1.2527,
      "step": 4728
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.545741210309973e-06,
      "loss": 1.3635,
      "step": 4729
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.533701501974391e-06,
      "loss": 1.1127,
      "step": 4730
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.521677001558461e-06,
      "loss": 1.3424,
      "step": 4731
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.509667713084254e-06,
      "loss": 1.0706,
      "step": 4732
    },
    {
      "epoch": 2.59,
      "learning_rate": 4.497673640568712e-06,
      "loss": 1.2613,
      "step": 4733
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.485694788023698e-06,
      "loss": 1.0438,
      "step": 4734
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.473731159456007e-06,
      "loss": 1.0686,
      "step": 4735
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.4617827588673166e-06,
      "loss": 0.9529,
      "step": 4736
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.449849590254235e-06,
      "loss": 1.2268,
      "step": 4737
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.437931657608246e-06,
      "loss": 1.24,
      "step": 4738
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.426028964915757e-06,
      "loss": 1.2996,
      "step": 4739
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.414141516158072e-06,
      "loss": 1.327,
      "step": 4740
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.402269315311402e-06,
      "loss": 1.2471,
      "step": 4741
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.390412366346841e-06,
      "loss": 0.8294,
      "step": 4742
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.378570673230409e-06,
      "loss": 1.2149,
      "step": 4743
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.366744239922998e-06,
      "loss": 1.2627,
      "step": 4744
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.354933070380396e-06,
      "loss": 1.0574,
      "step": 4745
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.3431371685533125e-06,
      "loss": 1.1286,
      "step": 4746
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.331356538387321e-06,
      "loss": 1.2226,
      "step": 4747
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.319591183822902e-06,
      "loss": 1.208,
      "step": 4748
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.307841108795418e-06,
      "loss": 1.3102,
      "step": 4749
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.296106317235138e-06,
      "loss": 1.2822,
      "step": 4750
    },
    {
      "epoch": 2.6,
      "learning_rate": 4.284386813067182e-06,
      "loss": 0.8627,
      "step": 4751
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.2726826002116085e-06,
      "loss": 1.1247,
      "step": 4752
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.2609936825833085e-06,
      "loss": 1.1664,
      "step": 4753
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.2493200640921e-06,
      "loss": 0.971,
      "step": 4754
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.237661748642651e-06,
      "loss": 0.9688,
      "step": 4755
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.226018740134541e-06,
      "loss": 0.9164,
      "step": 4756
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.2143910424622015e-06,
      "loss": 1.3863,
      "step": 4757
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.202778659514956e-06,
      "loss": 1.0602,
      "step": 4758
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.191181595177007e-06,
      "loss": 1.2393,
      "step": 4759
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.1795998533274265e-06,
      "loss": 1.0945,
      "step": 4760
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.168033437840179e-06,
      "loss": 1.0966,
      "step": 4761
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.156482352584068e-06,
      "loss": 1.0337,
      "step": 4762
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.144946601422806e-06,
      "loss": 0.9999,
      "step": 4763
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.1334261882149505e-06,
      "loss": 1.2783,
      "step": 4764
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.121921116813948e-06,
      "loss": 1.0585,
      "step": 4765
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.110431391068093e-06,
      "loss": 1.0675,
      "step": 4766
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.09895701482057e-06,
      "loss": 1.1667,
      "step": 4767
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.0874979919094e-06,
      "loss": 1.1278,
      "step": 4768
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.076054326167511e-06,
      "loss": 1.3378,
      "step": 4769
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.064626021422652e-06,
      "loss": 1.2749,
      "step": 4770
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.053213081497442e-06,
      "loss": 1.1783,
      "step": 4771
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.041815510209396e-06,
      "loss": 0.958,
      "step": 4772
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.030433311370835e-06,
      "loss": 1.1787,
      "step": 4773
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.0190664887889805e-06,
      "loss": 1.0904,
      "step": 4774
    },
    {
      "epoch": 2.62,
      "learning_rate": 4.0077150462658865e-06,
      "loss": 1.22,
      "step": 4775
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.996378987598487e-06,
      "loss": 1.196,
      "step": 4776
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.985058316578538e-06,
      "loss": 1.1076,
      "step": 4777
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.973753036992678e-06,
      "loss": 1.0215,
      "step": 4778
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.962463152622364e-06,
      "loss": 1.0996,
      "step": 4779
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.951188667243955e-06,
      "loss": 1.1405,
      "step": 4780
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.939929584628604e-06,
      "loss": 1.1333,
      "step": 4781
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.92868590854234e-06,
      "loss": 1.3091,
      "step": 4782
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.917457642746048e-06,
      "loss": 1.1336,
      "step": 4783
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.906244790995422e-06,
      "loss": 1.296,
      "step": 4784
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.895047357041049e-06,
      "loss": 1.2681,
      "step": 4785
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.883865344628307e-06,
      "loss": 1.1098,
      "step": 4786
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.8726987574974724e-06,
      "loss": 1.0797,
      "step": 4787
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.861547599383602e-06,
      "loss": 0.9994,
      "step": 4788
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.850411874016646e-06,
      "loss": 1.1357,
      "step": 4789
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.839291585121352e-06,
      "loss": 1.1357,
      "step": 4790
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.828186736417333e-06,
      "loss": 1.2188,
      "step": 4791
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.8170973316190074e-06,
      "loss": 1.1762,
      "step": 4792
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.8060233744356633e-06,
      "loss": 1.167,
      "step": 4793
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.794964868571399e-06,
      "loss": 1.3536,
      "step": 4794
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.7839218177251336e-06,
      "loss": 1.1587,
      "step": 4795
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.7728942255906564e-06,
      "loss": 1.25,
      "step": 4796
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.761882095856539e-06,
      "loss": 1.2676,
      "step": 4797
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.7508854322062305e-06,
      "loss": 1.1495,
      "step": 4798
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.7399042383179485e-06,
      "loss": 1.1813,
      "step": 4799
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.728938517864794e-06,
      "loss": 0.9501,
      "step": 4800
    },
    {
      "epoch": 2.63,
      "eval_loss": 1.3490906953811646,
      "eval_runtime": 36.8099,
      "eval_samples_per_second": 3.83,
      "eval_steps_per_second": 1.929,
      "step": 4800
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.71798827451465e-06,
      "loss": 1.1624,
      "step": 4801
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.707053511930253e-06,
      "loss": 1.0945,
      "step": 4802
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.6961342337691276e-06,
      "loss": 0.979,
      "step": 4803
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.685230443683657e-06,
      "loss": 1.1015,
      "step": 4804
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.6743421453210136e-06,
      "loss": 1.1043,
      "step": 4805
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.6634693423232114e-06,
      "loss": 1.4737,
      "step": 4806
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.6526120383270634e-06,
      "loss": 1.0816,
      "step": 4807
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.6417702369641925e-06,
      "loss": 0.9885,
      "step": 4808
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.6309439418610703e-06,
      "loss": 1.1051,
      "step": 4809
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.620133156638944e-06,
      "loss": 1.2463,
      "step": 4810
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.609337884913899e-06,
      "loss": 1.1236,
      "step": 4811
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.5985581302968085e-06,
      "loss": 1.0976,
      "step": 4812
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.5877938963933876e-06,
      "loss": 1.1237,
      "step": 4813
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.5770451868041176e-06,
      "loss": 1.2092,
      "step": 4814
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.5663120051243282e-06,
      "loss": 1.0143,
      "step": 4815
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.555594354944125e-06,
      "loss": 1.099,
      "step": 4816
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.5448922398484463e-06,
      "loss": 1.303,
      "step": 4817
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.534205663417006e-06,
      "loss": 1.2443,
      "step": 4818
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.5235346292243286e-06,
      "loss": 1.2842,
      "step": 4819
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.512879140839764e-06,
      "loss": 1.2898,
      "step": 4820
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.5022392018274177e-06,
      "loss": 1.3068,
      "step": 4821
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.491614815746236e-06,
      "loss": 1.2905,
      "step": 4822
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.4810059861499377e-06,
      "loss": 1.0618,
      "step": 4823
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.4704127165870517e-06,
      "loss": 1.149,
      "step": 4824
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.4598350106008924e-06,
      "loss": 1.0922,
      "step": 4825
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.449272871729581e-06,
      "loss": 1.1,
      "step": 4826
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.438726303506007e-06,
      "loss": 1.2128,
      "step": 4827
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.4281953094578877e-06,
      "loss": 1.6075,
      "step": 4828
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.417679893107695e-06,
      "loss": 1.0525,
      "step": 4829
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.407180057972725e-06,
      "loss": 1.1885,
      "step": 4830
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.3966958075650313e-06,
      "loss": 1.1908,
      "step": 4831
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.386227145391463e-06,
      "loss": 1.2706,
      "step": 4832
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.3757740749536694e-06,
      "loss": 1.1808,
      "step": 4833
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.3653365997480644e-06,
      "loss": 1.2151,
      "step": 4834
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.354914723265867e-06,
      "loss": 1.1281,
      "step": 4835
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.344508448993061e-06,
      "loss": 1.3371,
      "step": 4836
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.334117780410417e-06,
      "loss": 1.0976,
      "step": 4837
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.323742720993489e-06,
      "loss": 1.2348,
      "step": 4838
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.3133832742126057e-06,
      "loss": 1.2589,
      "step": 4839
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.303039443532874e-06,
      "loss": 1.1193,
      "step": 4840
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.2927112324141863e-06,
      "loss": 1.4147,
      "step": 4841
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.2823986443111854e-06,
      "loss": 1.1293,
      "step": 4842
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.272101682673323e-06,
      "loss": 1.1073,
      "step": 4843
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.2618203509448053e-06,
      "loss": 1.3974,
      "step": 4844
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.2515546525645925e-06,
      "loss": 1.0935,
      "step": 4845
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.2413045909664596e-06,
      "loss": 1.3187,
      "step": 4846
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.2310701695789093e-06,
      "loss": 1.0131,
      "step": 4847
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.220851391825247e-06,
      "loss": 1.3517,
      "step": 4848
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.2106482611235057e-06,
      "loss": 1.1496,
      "step": 4849
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.2004607808865273e-06,
      "loss": 1.3294,
      "step": 4850
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.190288954521886e-06,
      "loss": 1.1392,
      "step": 4851
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.1801327854319496e-06,
      "loss": 1.0925,
      "step": 4852
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.1699922770138113e-06,
      "loss": 1.1941,
      "step": 4853
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.1598674326593635e-06,
      "loss": 1.1661,
      "step": 4854
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.149758255755242e-06,
      "loss": 1.2713,
      "step": 4855
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.1396647496828247e-06,
      "loss": 1.2161,
      "step": 4856
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.1295869178182833e-06,
      "loss": 1.2085,
      "step": 4857
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.1195247635325166e-06,
      "loss": 1.2304,
      "step": 4858
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.109478290191209e-06,
      "loss": 1.2915,
      "step": 4859
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.0994475011547675e-06,
      "loss": 1.0192,
      "step": 4860
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.089432399778375e-06,
      "loss": 1.1227,
      "step": 4861
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.079432989411957e-06,
      "loss": 1.2107,
      "step": 4862
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.069449273400199e-06,
      "loss": 0.9471,
      "step": 4863
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.059481255082519e-06,
      "loss": 1.3919,
      "step": 4864
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.0495289377931168e-06,
      "loss": 1.45,
      "step": 4865
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.039592324860896e-06,
      "loss": 1.3192,
      "step": 4866
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.029671419609559e-06,
      "loss": 1.2485,
      "step": 4867
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.019766225357512e-06,
      "loss": 1.1482,
      "step": 4868
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.0098767454179167e-06,
      "loss": 1.3209,
      "step": 4869
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.0000029830986933e-06,
      "loss": 1.0241,
      "step": 4870
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.990144941702483e-06,
      "loss": 1.3075,
      "step": 4871
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.980302624526693e-06,
      "loss": 0.9562,
      "step": 4872
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.9704760348634453e-06,
      "loss": 1.0941,
      "step": 4873
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.9606651759996275e-06,
      "loss": 1.1114,
      "step": 4874
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.950870051216831e-06,
      "loss": 1.1495,
      "step": 4875
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.9410906637914348e-06,
      "loss": 1.1306,
      "step": 4876
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.931327016994495e-06,
      "loss": 1.0302,
      "step": 4877
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.9215791140918437e-06,
      "loss": 1.0766,
      "step": 4878
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.9118469583440446e-06,
      "loss": 1.2017,
      "step": 4879
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.9021305530063658e-06,
      "loss": 1.2697,
      "step": 4880
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.8924299013288415e-06,
      "loss": 1.2258,
      "step": 4881
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.8827450065562088e-06,
      "loss": 1.1092,
      "step": 4882
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.8730758719279605e-06,
      "loss": 1.4164,
      "step": 4883
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.8634225006782865e-06,
      "loss": 1.0637,
      "step": 4884
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.8537848960361378e-06,
      "loss": 1.1995,
      "step": 4885
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.8441630612251693e-06,
      "loss": 1.023,
      "step": 4886
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.8345569994637555e-06,
      "loss": 1.0842,
      "step": 4887
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.8249667139650216e-06,
      "loss": 1.2038,
      "step": 4888
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.81539220793679e-06,
      "loss": 1.2495,
      "step": 4889
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.8058334845816213e-06,
      "loss": 1.047,
      "step": 4890
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.7962905470967917e-06,
      "loss": 1.2484,
      "step": 4891
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.786763398674297e-06,
      "loss": 0.9408,
      "step": 4892
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.777252042500844e-06,
      "loss": 1.1074,
      "step": 4893
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.7677564817578772e-06,
      "loss": 1.2989,
      "step": 4894
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.758276719621533e-06,
      "loss": 1.1213,
      "step": 4895
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.7488127592626866e-06,
      "loss": 1.2693,
      "step": 4896
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.7393646038469055e-06,
      "loss": 1.0519,
      "step": 4897
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.729932256534495e-06,
      "loss": 1.2401,
      "step": 4898
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.7205157204804543e-06,
      "loss": 1.3339,
      "step": 4899
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.711114998834485e-06,
      "loss": 1.2062,
      "step": 4900
    },
    {
      "epoch": 2.69,
      "eval_loss": 1.349608063697815,
      "eval_runtime": 36.924,
      "eval_samples_per_second": 3.819,
      "eval_steps_per_second": 1.923,
      "step": 4900
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.7017300947410395e-06,
      "loss": 1.4775,
      "step": 4901
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.692361011339228e-06,
      "loss": 1.1752,
      "step": 4902
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.683007751762917e-06,
      "loss": 1.0168,
      "step": 4903
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.6736703191406366e-06,
      "loss": 1.2087,
      "step": 4904
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.6643487165956604e-06,
      "loss": 1.1499,
      "step": 4905
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.6550429472459394e-06,
      "loss": 1.1553,
      "step": 4906
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.6457530142041444e-06,
      "loss": 1.2449,
      "step": 4907
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.636478920577634e-06,
      "loss": 0.7747,
      "step": 4908
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.6272206694684998e-06,
      "loss": 1.3692,
      "step": 4909
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.6179782639734918e-06,
      "loss": 1.109,
      "step": 4910
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.6087517071840985e-06,
      "loss": 1.2318,
      "step": 4911
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.5995410021864787e-06,
      "loss": 1.3062,
      "step": 4912
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.5903461520615015e-06,
      "loss": 1.2021,
      "step": 4913
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.5811671598847343e-06,
      "loss": 1.1226,
      "step": 4914
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.572004028726438e-06,
      "loss": 1.1684,
      "step": 4915
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.5628567616515663e-06,
      "loss": 0.988,
      "step": 4916
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.5537253617197665e-06,
      "loss": 1.2026,
      "step": 4917
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.5446098319853894e-06,
      "loss": 1.1936,
      "step": 4918
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.5355101754974462e-06,
      "loss": 1.1191,
      "step": 4919
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.5264263952996915e-06,
      "loss": 1.1618,
      "step": 4920
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.517358494430505e-06,
      "loss": 1.2592,
      "step": 4921
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.508306475923017e-06,
      "loss": 1.2079,
      "step": 4922
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.4992703428050046e-06,
      "loss": 1.2482,
      "step": 4923
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.4902500980989395e-06,
      "loss": 1.1773,
      "step": 4924
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.481245744821986e-06,
      "loss": 1.1197,
      "step": 4925
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.4722572859859904e-06,
      "loss": 1.148,
      "step": 4926
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.4632847245974922e-06,
      "loss": 1.2822,
      "step": 4927
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.4543280636576794e-06,
      "loss": 1.1015,
      "step": 4928
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.445387306162472e-06,
      "loss": 0.9587,
      "step": 4929
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.436462455102423e-06,
      "loss": 1.1617,
      "step": 4930
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.4275535134628045e-06,
      "loss": 1.2403,
      "step": 4931
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.4186604842235282e-06,
      "loss": 1.1764,
      "step": 4932
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.4097833703592197e-06,
      "loss": 1.1054,
      "step": 4933
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.400922174839154e-06,
      "loss": 1.1525,
      "step": 4934
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.392076900627299e-06,
      "loss": 1.2795,
      "step": 4935
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.3832475506822937e-06,
      "loss": 1.2864,
      "step": 4936
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.3744341279574313e-06,
      "loss": 1.2011,
      "step": 4937
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.365636635400714e-06,
      "loss": 1.1376,
      "step": 4938
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.356855075954778e-06,
      "loss": 1.1122,
      "step": 4939
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.3480894525569562e-06,
      "loss": 1.2366,
      "step": 4940
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.3393397681392425e-06,
      "loss": 1.246,
      "step": 4941
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.3306060256283014e-06,
      "loss": 1.0631,
      "step": 4942
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.3218882279454513e-06,
      "loss": 1.029,
      "step": 4943
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.3131863780067043e-06,
      "loss": 1.0176,
      "step": 4944
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.3045004787227096e-06,
      "loss": 1.2399,
      "step": 4945
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.2958305329987996e-06,
      "loss": 0.9926,
      "step": 4946
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.2871765437349646e-06,
      "loss": 1.0912,
      "step": 4947
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.278538513825862e-06,
      "loss": 1.2063,
      "step": 4948
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.2699164461608027e-06,
      "loss": 1.083,
      "step": 4949
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.261310343623757e-06,
      "loss": 1.27,
      "step": 4950
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.2527202090933676e-06,
      "loss": 1.2472,
      "step": 4951
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.24414604544293e-06,
      "loss": 1.2061,
      "step": 4952
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.2355878555403942e-06,
      "loss": 1.3032,
      "step": 4953
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.2270456422483655e-06,
      "loss": 1.0624,
      "step": 4954
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.2185194084241243e-06,
      "loss": 1.3563,
      "step": 4955
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.2100091569195735e-06,
      "loss": 0.9627,
      "step": 4956
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.2015148905813022e-06,
      "loss": 1.2401,
      "step": 4957
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.1930366122505263e-06,
      "loss": 0.9484,
      "step": 4958
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.184574324763139e-06,
      "loss": 1.2834,
      "step": 4959
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.1761280309496646e-06,
      "loss": 0.7383,
      "step": 4960
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.1676977336352766e-06,
      "loss": 1.3771,
      "step": 4961
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.159283435639825e-06,
      "loss": 1.1538,
      "step": 4962
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.15088513977777e-06,
      "loss": 0.9596,
      "step": 4963
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.1425028488582523e-06,
      "loss": 1.3117,
      "step": 4964
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.1341365656850354e-06,
      "loss": 1.4369,
      "step": 4965
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.1257862930565477e-06,
      "loss": 1.0508,
      "step": 4966
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.1174520337658487e-06,
      "loss": 1.2029,
      "step": 4967
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.1091337906006482e-06,
      "loss": 1.1649,
      "step": 4968
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.100831566343292e-06,
      "loss": 1.1192,
      "step": 4969
    },
    {
      "epoch": 2.72,
      "learning_rate": 2.092545363770776e-06,
      "loss": 1.132,
      "step": 4970
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.0842751856547325e-06,
      "loss": 0.838,
      "step": 4971
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.0760210347614385e-06,
      "loss": 1.1956,
      "step": 4972
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.0677829138518014e-06,
      "loss": 1.1015,
      "step": 4973
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.059560825681367e-06,
      "loss": 1.0787,
      "step": 4974
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.0513547730003355e-06,
      "loss": 1.0479,
      "step": 4975
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.043164758553523e-06,
      "loss": 1.1956,
      "step": 4976
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.0349907850803927e-06,
      "loss": 1.1824,
      "step": 4977
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.0268328553150307e-06,
      "loss": 1.3335,
      "step": 4978
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.0186909719861757e-06,
      "loss": 1.317,
      "step": 4979
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.010565137817172e-06,
      "loss": 1.1832,
      "step": 4980
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.002455355526034e-06,
      "loss": 1.1055,
      "step": 4981
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.994361627825364e-06,
      "loss": 1.1086,
      "step": 4982
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.9862839574224302e-06,
      "loss": 1.5187,
      "step": 4983
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.9782223470191042e-06,
      "loss": 1.4218,
      "step": 4984
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.9701767993119013e-06,
      "loss": 1.2162,
      "step": 4985
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.9621473169919526e-06,
      "loss": 1.1384,
      "step": 4986
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.9541339027450256e-06,
      "loss": 1.0762,
      "step": 4987
    },
    {
      "epoch": 2.73,
      "learning_rate": 1.94613655925151e-06,
      "loss": 0.9471,
      "step": 4988
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.938155289186416e-06,
      "loss": 1.0712,
      "step": 4989
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.930190095219386e-06,
      "loss": 1.2108,
      "step": 4990
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.9222409800146723e-06,
      "loss": 1.0423,
      "step": 4991
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.914307946231164e-06,
      "loss": 0.9264,
      "step": 4992
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.906390996522356e-06,
      "loss": 1.1113,
      "step": 4993
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.8984901335363792e-06,
      "loss": 1.1717,
      "step": 4994
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.89060535991597e-06,
      "loss": 1.1244,
      "step": 4995
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.8827366782984913e-06,
      "loss": 1.0546,
      "step": 4996
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.8748840913159205e-06,
      "loss": 0.8658,
      "step": 4997
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.8670476015948457e-06,
      "loss": 0.9015,
      "step": 4998
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.8592272117564869e-06,
      "loss": 1.2926,
      "step": 4999
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.8514229244166569e-06,
      "loss": 1.1728,
      "step": 5000
    },
    {
      "epoch": 2.74,
      "eval_loss": 1.349087119102478,
      "eval_runtime": 36.8719,
      "eval_samples_per_second": 3.824,
      "eval_steps_per_second": 1.926,
      "step": 5000
    }
  ],
  "logging_steps": 1,
  "max_steps": 5472,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.91244256935936e+18,
  "trial_name": null,
  "trial_params": null
}