{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.14666666666666667,
  "global_step": 4400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 5.000000000000001e-07,
      "loss": 1.0902,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.9914,
      "step": 2
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.5e-06,
      "loss": 1.0351,
      "step": 3
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.9363,
      "step": 4
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.5e-06,
      "loss": 0.7614,
      "step": 5
    },
    {
      "epoch": 0.0,
      "learning_rate": 3e-06,
      "loss": 1.0516,
      "step": 6
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.5000000000000004e-06,
      "loss": 1.0139,
      "step": 7
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.9795,
      "step": 8
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.5e-06,
      "loss": 1.0205,
      "step": 9
    },
    {
      "epoch": 0.0,
      "learning_rate": 5e-06,
      "loss": 0.8007,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.9832,
      "step": 11
    },
    {
      "epoch": 0.0,
      "learning_rate": 6e-06,
      "loss": 0.9382,
      "step": 12
    },
    {
      "epoch": 0.0,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.9567,
      "step": 13
    },
    {
      "epoch": 0.0,
      "learning_rate": 7.000000000000001e-06,
      "loss": 0.9929,
      "step": 14
    },
    {
      "epoch": 0.0,
      "learning_rate": 7.5e-06,
      "loss": 0.7402,
      "step": 15
    },
    {
      "epoch": 0.0,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.9893,
      "step": 16
    },
    {
      "epoch": 0.0,
      "learning_rate": 8.500000000000002e-06,
      "loss": 0.9357,
      "step": 17
    },
    {
      "epoch": 0.0,
      "learning_rate": 9e-06,
      "loss": 0.8532,
      "step": 18
    },
    {
      "epoch": 0.0,
      "learning_rate": 9.5e-06,
      "loss": 0.9383,
      "step": 19
    },
    {
      "epoch": 0.0,
      "learning_rate": 1e-05,
      "loss": 1.0307,
      "step": 20
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.05e-05,
      "loss": 0.9328,
      "step": 21
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.5551,
      "step": 22
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.1500000000000002e-05,
      "loss": 0.4643,
      "step": 23
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.2e-05,
      "loss": 0.8399,
      "step": 24
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.25e-05,
      "loss": 0.8979,
      "step": 25
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.9994,
      "step": 26
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 0.7753,
      "step": 27
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 0.7346,
      "step": 28
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.45e-05,
      "loss": 0.9878,
      "step": 29
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.5e-05,
      "loss": 0.9227,
      "step": 30
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.55e-05,
      "loss": 1.0723,
      "step": 31
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.9868,
      "step": 32
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.65e-05,
      "loss": 0.8494,
      "step": 33
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 0.9761,
      "step": 34
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.75e-05,
      "loss": 0.9686,
      "step": 35
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.8e-05,
      "loss": 0.9145,
      "step": 36
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.85e-05,
      "loss": 1.0263,
      "step": 37
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9e-05,
      "loss": 0.8986,
      "step": 38
    },
    {
      "epoch": 0.0,
      "learning_rate": 1.9500000000000003e-05,
      "loss": 0.9069,
      "step": 39
    },
    {
      "epoch": 0.0,
      "learning_rate": 2e-05,
      "loss": 0.9984,
      "step": 40
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.05e-05,
      "loss": 0.5577,
      "step": 41
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.1e-05,
      "loss": 0.2135,
      "step": 42
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.15e-05,
      "loss": 0.282,
      "step": 43
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 0.3749,
      "step": 44
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.25e-05,
      "loss": 1.0921,
      "step": 45
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.3000000000000003e-05,
      "loss": 1.0201,
      "step": 46
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.35e-05,
      "loss": 1.1992,
      "step": 47
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.4e-05,
      "loss": 0.8502,
      "step": 48
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.45e-05,
      "loss": 0.9322,
      "step": 49
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.5e-05,
      "loss": 1.0279,
      "step": 50
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.5500000000000003e-05,
      "loss": 0.8806,
      "step": 51
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 1.1087,
      "step": 52
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.6500000000000004e-05,
      "loss": 0.9143,
      "step": 53
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7000000000000002e-05,
      "loss": 0.9367,
      "step": 54
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 1.1204,
      "step": 55
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 1.1446,
      "step": 56
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.8499999999999998e-05,
      "loss": 1.0805,
      "step": 57
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9e-05,
      "loss": 1.0292,
      "step": 58
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.95e-05,
      "loss": 0.9298,
      "step": 59
    },
    {
      "epoch": 0.0,
      "learning_rate": 3e-05,
      "loss": 0.9341,
      "step": 60
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.05e-05,
      "loss": 0.8873,
      "step": 61
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.1e-05,
      "loss": 0.7121,
      "step": 62
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.15e-05,
      "loss": 0.889,
      "step": 63
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.9066,
      "step": 64
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 0.9526,
      "step": 65
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.3e-05,
      "loss": 0.8031,
      "step": 66
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.35e-05,
      "loss": 0.8619,
      "step": 67
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 0.9519,
      "step": 68
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.45e-05,
      "loss": 1.0572,
      "step": 69
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.5e-05,
      "loss": 1.0136,
      "step": 70
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.55e-05,
      "loss": 0.7655,
      "step": 71
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.6e-05,
      "loss": 0.865,
      "step": 72
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.65e-05,
      "loss": 1.0115,
      "step": 73
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7e-05,
      "loss": 0.9935,
      "step": 74
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.9198,
      "step": 75
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.8e-05,
      "loss": 0.9909,
      "step": 76
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.85e-05,
      "loss": 0.8318,
      "step": 77
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9000000000000006e-05,
      "loss": 1.119,
      "step": 78
    },
    {
      "epoch": 0.0,
      "learning_rate": 3.9500000000000005e-05,
      "loss": 0.9681,
      "step": 79
    },
    {
      "epoch": 0.0,
      "learning_rate": 4e-05,
      "loss": 1.0352,
      "step": 80
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.05e-05,
      "loss": 1.4678,
      "step": 81
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.1e-05,
      "loss": 1.2257,
      "step": 82
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.15e-05,
      "loss": 0.8349,
      "step": 83
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.2e-05,
      "loss": 0.9431,
      "step": 84
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.25e-05,
      "loss": 1.0355,
      "step": 85
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.3e-05,
      "loss": 1.2265,
      "step": 86
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.35e-05,
      "loss": 1.2948,
      "step": 87
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.9883,
      "step": 88
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.4500000000000004e-05,
      "loss": 1.1066,
      "step": 89
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.5e-05,
      "loss": 1.0885,
      "step": 90
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.55e-05,
      "loss": 1.1,
      "step": 91
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.600000000000001e-05,
      "loss": 0.6561,
      "step": 92
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.6500000000000005e-05,
      "loss": 0.3505,
      "step": 93
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.7e-05,
      "loss": 0.7383,
      "step": 94
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.75e-05,
      "loss": 1.0832,
      "step": 95
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.8e-05,
      "loss": 0.9404,
      "step": 96
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.85e-05,
      "loss": 0.9826,
      "step": 97
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9e-05,
      "loss": 0.9864,
      "step": 98
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9500000000000004e-05,
      "loss": 1.0108,
      "step": 99
    },
    {
      "epoch": 0.0,
      "learning_rate": 5e-05,
      "loss": 2.226,
      "step": 100
    },
    {
      "epoch": 0.0,
      "eval_loss": 0.9811180830001831,
      "eval_runtime": 286.4815,
      "eval_samples_per_second": 7.508,
      "eval_steps_per_second": 1.878,
      "step": 100
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999999986200372e-05,
      "loss": 1.7375,
      "step": 101
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999999448014875e-05,
      "loss": 0.9567,
      "step": 102
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999998758033485e-05,
      "loss": 0.2692,
      "step": 103
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999999779205954e-05,
      "loss": 0.9918,
      "step": 104
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999996550093064e-05,
      "loss": 1.1179,
      "step": 105
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999999503213406e-05,
      "loss": 1.0808,
      "step": 106
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999993238182544e-05,
      "loss": 1.0265,
      "step": 107
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999999116823855e-05,
      "loss": 0.7817,
      "step": 108
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999998882230209e-05,
      "loss": 1.2075,
      "step": 109
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999998620037319e-05,
      "loss": 1.1747,
      "step": 110
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999998330245188e-05,
      "loss": 1.0525,
      "step": 111
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999998012853819e-05,
      "loss": 1.1455,
      "step": 112
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999997667863217e-05,
      "loss": 1.0754,
      "step": 113
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999997295273384e-05,
      "loss": 1.0804,
      "step": 114
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999996895084325e-05,
      "loss": 1.0223,
      "step": 115
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999996467296043e-05,
      "loss": 0.9495,
      "step": 116
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999960119085446e-05,
      "loss": 1.1192,
      "step": 117
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999995528921835e-05,
      "loss": 1.0372,
      "step": 118
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999995018335917e-05,
      "loss": 1.0937,
      "step": 119
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999994480150799e-05,
      "loss": 1.0274,
      "step": 120
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999939143664854e-05,
      "loss": 1.8794,
      "step": 121
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999993320982983e-05,
      "loss": 0.8559,
      "step": 122
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999992700000298e-05,
      "loss": 1.0649,
      "step": 123
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999920514184375e-05,
      "loss": 1.0707,
      "step": 124
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999913752374086e-05,
      "loss": 1.1028,
      "step": 125
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999906714572185e-05,
      "loss": 1.0743,
      "step": 126
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999899400778754e-05,
      "loss": 1.1398,
      "step": 127
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999989181099387e-05,
      "loss": 1.0909,
      "step": 128
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999988394521763e-05,
      "loss": 0.9028,
      "step": 129
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.99998758034501e-05,
      "loss": 0.8138,
      "step": 130
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999986738569138e-05,
      "loss": 1.0199,
      "step": 131
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999985869194157e-05,
      "loss": 0.9057,
      "step": 132
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999984972220075e-05,
      "loss": 1.1228,
      "step": 133
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999984047646903e-05,
      "loss": 1.0472,
      "step": 134
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999830954746506e-05,
      "loss": 1.056,
      "step": 135
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999821157033294e-05,
      "loss": 1.034,
      "step": 136
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999811083329494e-05,
      "loss": 0.9589,
      "step": 137
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999800733635214e-05,
      "loss": 0.9328,
      "step": 138
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999790107950576e-05,
      "loss": 1.0168,
      "step": 139
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999779206275695e-05,
      "loss": 1.1359,
      "step": 140
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999976802861069e-05,
      "loss": 1.1179,
      "step": 141
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999975657495568e-05,
      "loss": 0.9603,
      "step": 142
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999744845310796e-05,
      "loss": 1.1063,
      "step": 143
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999973283967617e-05,
      "loss": 1.0692,
      "step": 144
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999972055805193e-05,
      "loss": 1.0666,
      "step": 145
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999970800043822e-05,
      "loss": 1.0826,
      "step": 146
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999969516683517e-05,
      "loss": 1.0553,
      "step": 147
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.999968205724292e-05,
      "loss": 1.0362,
      "step": 148
    },
    {
      "epoch": 0.0,
      "learning_rate": 4.9999668671661614e-05,
      "loss": 1.0572,
      "step": 149
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999965501009142e-05,
      "loss": 0.8805,
      "step": 150
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999641072532455e-05,
      "loss": 0.8617,
      "step": 151
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999962685898491e-05,
      "loss": 1.0921,
      "step": 152
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999612369448905e-05,
      "loss": 1.4899,
      "step": 153
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999959760392463e-05,
      "loss": 2.3447,
      "step": 154
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999958256241223e-05,
      "loss": 0.896,
      "step": 155
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999956724491188e-05,
      "loss": 1.9462,
      "step": 156
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999955165142374e-05,
      "loss": 1.1192,
      "step": 157
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999953578194799e-05,
      "loss": 1.1071,
      "step": 158
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999519636484816e-05,
      "loss": 1.0947,
      "step": 159
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999950321503437e-05,
      "loss": 1.0485,
      "step": 160
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999948651759685e-05,
      "loss": 0.9885,
      "step": 161
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999946954417244e-05,
      "loss": 0.9017,
      "step": 162
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999945229476132e-05,
      "loss": 1.0936,
      "step": 163
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999943476936369e-05,
      "loss": 0.9098,
      "step": 164
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999416967979736e-05,
      "loss": 0.8462,
      "step": 165
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999398890609664e-05,
      "loss": 0.9667,
      "step": 166
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999938053725366e-05,
      "loss": 1.0523,
      "step": 167
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999936190791194e-05,
      "loss": 1.1363,
      "step": 168
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999934300258469e-05,
      "loss": 1.0255,
      "step": 169
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999932382127214e-05,
      "loss": 1.0358,
      "step": 170
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999930436397449e-05,
      "loss": 1.007,
      "step": 171
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999284630691954e-05,
      "loss": 1.0034,
      "step": 172
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999926462142477e-05,
      "loss": 0.8748,
      "step": 173
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999924433617313e-05,
      "loss": 0.94,
      "step": 174
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999223774937274e-05,
      "loss": 0.9111,
      "step": 175
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999202937717426e-05,
      "loss": 0.7797,
      "step": 176
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999918182451382e-05,
      "loss": 0.307,
      "step": 177
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999916043532667e-05,
      "loss": 0.8486,
      "step": 178
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999913877015625e-05,
      "loss": 0.8855,
      "step": 179
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999911682900276e-05,
      "loss": 1.0274,
      "step": 180
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999909461186646e-05,
      "loss": 1.0564,
      "step": 181
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9999072118747606e-05,
      "loss": 1.142,
      "step": 182
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999904934964642e-05,
      "loss": 1.1928,
      "step": 183
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999902630456317e-05,
      "loss": 1.1151,
      "step": 184
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999900298349811e-05,
      "loss": 1.0257,
      "step": 185
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99989793864515e-05,
      "loss": 0.9999,
      "step": 186
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999895551342359e-05,
      "loss": 0.898,
      "step": 187
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999893136441465e-05,
      "loss": 1.0332,
      "step": 188
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999890693942495e-05,
      "loss": 0.9958,
      "step": 189
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998882238454754e-05,
      "loss": 1.0744,
      "step": 190
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998857261504335e-05,
      "loss": 1.0616,
      "step": 191
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999883200857397e-05,
      "loss": 1.0013,
      "step": 192
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999880647966394e-05,
      "loss": 1.0385,
      "step": 193
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998780674774525e-05,
      "loss": 1.0094,
      "step": 194
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999875459390601e-05,
      "loss": 1.0483,
      "step": 195
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999872823705868e-05,
      "loss": 0.9253,
      "step": 196
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998701604232825e-05,
      "loss": 0.8676,
      "step": 197
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999867469542875e-05,
      "loss": 0.8982,
      "step": 198
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998647510646736e-05,
      "loss": 0.9916,
      "step": 199
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99986200498871e-05,
      "loss": 0.5986,
      "step": 200
    },
    {
      "epoch": 0.01,
      "eval_loss": 1.0025880336761475,
      "eval_runtime": 288.262,
      "eval_samples_per_second": 7.462,
      "eval_steps_per_second": 1.866,
      "step": 200
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998592313150124e-05,
      "loss": 0.4505,
      "step": 201
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999856430043614e-05,
      "loss": 0.7945,
      "step": 202
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998536011745444e-05,
      "loss": 0.9991,
      "step": 203
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999850744707835e-05,
      "loss": 0.9203,
      "step": 204
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999847860643516e-05,
      "loss": 0.9455,
      "step": 205
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998449489816215e-05,
      "loss": 0.8298,
      "step": 206
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999842009722182e-05,
      "loss": 1.1219,
      "step": 207
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999839042865231e-05,
      "loss": 1.1029,
      "step": 208
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999836048410801e-05,
      "loss": 0.9851,
      "step": 209
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999833026358924e-05,
      "loss": 1.0381,
      "step": 210
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999829976709634e-05,
      "loss": 1.0459,
      "step": 211
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999826899462966e-05,
      "loss": 0.9523,
      "step": 212
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9998237946189524e-05,
      "loss": 0.9829,
      "step": 213
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999820662177628e-05,
      "loss": 0.9213,
      "step": 214
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999817502139027e-05,
      "loss": 0.9902,
      "step": 215
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999814314503185e-05,
      "loss": 0.7602,
      "step": 216
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999811099270136e-05,
      "loss": 1.2037,
      "step": 217
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999807856439917e-05,
      "loss": 1.1092,
      "step": 218
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999804586012563e-05,
      "loss": 1.0892,
      "step": 219
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999801287988109e-05,
      "loss": 1.0275,
      "step": 220
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999797962366595e-05,
      "loss": 1.0424,
      "step": 221
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999794609148053e-05,
      "loss": 1.0709,
      "step": 222
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9997912283325234e-05,
      "loss": 0.9883,
      "step": 223
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999787819920042e-05,
      "loss": 0.9614,
      "step": 224
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9997843839106476e-05,
      "loss": 0.8684,
      "step": 225
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999780920304377e-05,
      "loss": 0.9717,
      "step": 226
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999777429101269e-05,
      "loss": 0.9541,
      "step": 227
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999773910301361e-05,
      "loss": 0.8166,
      "step": 228
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999770363904694e-05,
      "loss": 0.9351,
      "step": 229
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999766789911305e-05,
      "loss": 0.9637,
      "step": 230
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9997631883212356e-05,
      "loss": 0.7872,
      "step": 231
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9997595591345234e-05,
      "loss": 1.0338,
      "step": 232
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9997559023512095e-05,
      "loss": 1.0239,
      "step": 233
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999752217971335e-05,
      "loss": 1.0428,
      "step": 234
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999748505994939e-05,
      "loss": 1.1327,
      "step": 235
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999744766422064e-05,
      "loss": 1.0609,
      "step": 236
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99974099925275e-05,
      "loss": 0.9827,
      "step": 237
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999737204487039e-05,
      "loss": 0.8606,
      "step": 238
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999733382124974e-05,
      "loss": 0.4381,
      "step": 239
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999729532166595e-05,
      "loss": 1.2517,
      "step": 240
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999725654611946e-05,
      "loss": 1.0025,
      "step": 241
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99972174946107e-05,
      "loss": 1.0465,
      "step": 242
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99971781671401e-05,
      "loss": 1.331,
      "step": 243
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999713856370809e-05,
      "loss": 1.1863,
      "step": 244
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99970986843151e-05,
      "loss": 1.3004,
      "step": 245
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999705852896159e-05,
      "loss": 1.2611,
      "step": 246
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999701809764798e-05,
      "loss": 1.3557,
      "step": 247
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999697739037473e-05,
      "loss": 1.3999,
      "step": 248
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999693640714229e-05,
      "loss": 1.3641,
      "step": 249
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999689514795112e-05,
      "loss": 1.3026,
      "step": 250
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9996853612801646e-05,
      "loss": 1.1124,
      "step": 251
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9996811801694356e-05,
      "loss": 0.979,
      "step": 252
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99967697146297e-05,
      "loss": 0.951,
      "step": 253
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999672735160814e-05,
      "loss": 0.9954,
      "step": 254
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999668471263016e-05,
      "loss": 1.1348,
      "step": 255
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9996641797696206e-05,
      "loss": 1.0196,
      "step": 256
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999659860680676e-05,
      "loss": 1.0767,
      "step": 257
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999655513996232e-05,
      "loss": 1.1463,
      "step": 258
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999651139716334e-05,
      "loss": 0.9773,
      "step": 259
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9996467378410314e-05,
      "loss": 1.0357,
      "step": 260
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9996423083703726e-05,
      "loss": 0.8769,
      "step": 261
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999637851304406e-05,
      "loss": 0.8828,
      "step": 262
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999633366643182e-05,
      "loss": 0.825,
      "step": 263
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9996288543867484e-05,
      "loss": 0.8397,
      "step": 264
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9996243145351564e-05,
      "loss": 1.0061,
      "step": 265
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9996197470884554e-05,
      "loss": 0.3803,
      "step": 266
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999615152046697e-05,
      "loss": 0.2163,
      "step": 267
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999610529409931e-05,
      "loss": 0.2657,
      "step": 268
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999605879178207e-05,
      "loss": 0.32,
      "step": 269
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99960120135158e-05,
      "loss": 1.1206,
      "step": 270
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999596495930098e-05,
      "loss": 1.04,
      "step": 271
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999591762913815e-05,
      "loss": 0.7424,
      "step": 272
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999587002302782e-05,
      "loss": 0.8903,
      "step": 273
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9995822140970535e-05,
      "loss": 0.9659,
      "step": 274
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99957739829668e-05,
      "loss": 1.0901,
      "step": 275
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9995725549017166e-05,
      "loss": 0.8668,
      "step": 276
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9995676839122155e-05,
      "loss": 0.9852,
      "step": 277
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999562785328231e-05,
      "loss": 0.9686,
      "step": 278
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999557859149816e-05,
      "loss": 1.0635,
      "step": 279
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9995529053770274e-05,
      "loss": 1.2015,
      "step": 280
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999547924009918e-05,
      "loss": 1.0039,
      "step": 281
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999542915048543e-05,
      "loss": 1.0388,
      "step": 282
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999537878492958e-05,
      "loss": 0.9747,
      "step": 283
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999532814343219e-05,
      "loss": 1.0518,
      "step": 284
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999527722599381e-05,
      "loss": 1.0952,
      "step": 285
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999522603261501e-05,
      "loss": 1.1718,
      "step": 286
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999517456329635e-05,
      "loss": 1.0976,
      "step": 287
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99951228180384e-05,
      "loss": 1.0271,
      "step": 288
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999507079684173e-05,
      "loss": 1.0371,
      "step": 289
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999501849970691e-05,
      "loss": 0.9077,
      "step": 290
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999496592663454e-05,
      "loss": 0.9086,
      "step": 291
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999491307762517e-05,
      "loss": 0.8976,
      "step": 292
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99948599526794e-05,
      "loss": 1.1761,
      "step": 293
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999480655179781e-05,
      "loss": 0.9361,
      "step": 294
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9994752874980996e-05,
      "loss": 1.0782,
      "step": 295
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999469892222955e-05,
      "loss": 0.8044,
      "step": 296
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999464469354406e-05,
      "loss": 1.0306,
      "step": 297
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999459018892513e-05,
      "loss": 1.0473,
      "step": 298
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999453540837337e-05,
      "loss": 0.9747,
      "step": 299
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9994480351889364e-05,
      "loss": 1.0657,
      "step": 300
    },
    {
      "epoch": 0.01,
      "eval_loss": 0.9896363019943237,
      "eval_runtime": 287.5501,
      "eval_samples_per_second": 7.48,
      "eval_steps_per_second": 1.871,
      "step": 300
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999442501947373e-05,
      "loss": 1.0927,
      "step": 301
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99943694111271e-05,
      "loss": 0.9354,
      "step": 302
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9994313526850054e-05,
      "loss": 1.0478,
      "step": 303
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9994257366643225e-05,
      "loss": 1.2211,
      "step": 304
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999420093050723e-05,
      "loss": 1.2361,
      "step": 305
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99941442184427e-05,
      "loss": 0.8499,
      "step": 306
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999408723045025e-05,
      "loss": 1.0831,
      "step": 307
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999402996653051e-05,
      "loss": 2.0121,
      "step": 308
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999397242668412e-05,
      "loss": 2.4411,
      "step": 309
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999391461091172e-05,
      "loss": 1.0633,
      "step": 310
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999385651921392e-05,
      "loss": 1.0095,
      "step": 311
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9993798151591396e-05,
      "loss": 1.3694,
      "step": 312
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999373950804478e-05,
      "loss": 1.9499,
      "step": 313
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99936805885747e-05,
      "loss": 1.0088,
      "step": 314
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999362139318184e-05,
      "loss": 1.055,
      "step": 315
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9993561921866826e-05,
      "loss": 1.0745,
      "step": 316
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999350217463033e-05,
      "loss": 1.0512,
      "step": 317
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9993442151473e-05,
      "loss": 1.0328,
      "step": 318
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9993381852395516e-05,
      "loss": 0.8703,
      "step": 319
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999332127739853e-05,
      "loss": 0.9811,
      "step": 320
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999326042648271e-05,
      "loss": 1.0212,
      "step": 321
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999319929964873e-05,
      "loss": 0.4433,
      "step": 322
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999313789689727e-05,
      "loss": 1.1043,
      "step": 323
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999307621822901e-05,
      "loss": 1.1282,
      "step": 324
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999301426364461e-05,
      "loss": 0.9737,
      "step": 325
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999295203314478e-05,
      "loss": 0.9718,
      "step": 326
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999288952673019e-05,
      "loss": 1.0851,
      "step": 327
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999282674440154e-05,
      "loss": 0.9304,
      "step": 328
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999276368615952e-05,
      "loss": 0.9937,
      "step": 329
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999270035200483e-05,
      "loss": 0.8852,
      "step": 330
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999263674193816e-05,
      "loss": 1.6157,
      "step": 331
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999257285596021e-05,
      "loss": 1.4129,
      "step": 332
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99925086940717e-05,
      "loss": 1.339,
      "step": 333
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999244425627333e-05,
      "loss": 1.6537,
      "step": 334
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9992379542565804e-05,
      "loss": 0.9072,
      "step": 335
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9992314552949854e-05,
      "loss": 5.0355,
      "step": 336
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999224928742618e-05,
      "loss": 2.8767,
      "step": 337
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9992183745995516e-05,
      "loss": 1.1188,
      "step": 338
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999211792865858e-05,
      "loss": 1.1393,
      "step": 339
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99920518354161e-05,
      "loss": 1.3189,
      "step": 340
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99919854662688e-05,
      "loss": 1.0183,
      "step": 341
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999191882121741e-05,
      "loss": 0.525,
      "step": 342
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9991851900262684e-05,
      "loss": 0.1827,
      "step": 343
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999178470340534e-05,
      "loss": 0.1538,
      "step": 344
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9991717230646134e-05,
      "loss": 0.0958,
      "step": 345
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99916494819858e-05,
      "loss": 0.6668,
      "step": 346
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99915814574251e-05,
      "loss": 1.037,
      "step": 347
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999151315696477e-05,
      "loss": 0.7468,
      "step": 348
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999144458060558e-05,
      "loss": 0.9648,
      "step": 349
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999137572834828e-05,
      "loss": 1.088,
      "step": 350
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9991306600193614e-05,
      "loss": 1.0317,
      "step": 351
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999123719614237e-05,
      "loss": 1.0951,
      "step": 352
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9991167516195295e-05,
      "loss": 0.9791,
      "step": 353
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999109756035317e-05,
      "loss": 0.8426,
      "step": 354
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999102732861677e-05,
      "loss": 0.9078,
      "step": 355
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9990956820986856e-05,
      "loss": 1.0689,
      "step": 356
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999088603746422e-05,
      "loss": 1.8993,
      "step": 357
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999081497804963e-05,
      "loss": 1.0794,
      "step": 358
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999074364274389e-05,
      "loss": 1.1229,
      "step": 359
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999067203154777e-05,
      "loss": 2.1663,
      "step": 360
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999060014446206e-05,
      "loss": 1.8156,
      "step": 361
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999052798148757e-05,
      "loss": 0.9432,
      "step": 362
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999045554262509e-05,
      "loss": 0.9351,
      "step": 363
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9990382827875406e-05,
      "loss": 1.0472,
      "step": 364
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9990309837239335e-05,
      "loss": 1.0323,
      "step": 365
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999023657071769e-05,
      "loss": 1.1139,
      "step": 366
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9990163028311256e-05,
      "loss": 1.0403,
      "step": 367
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9990089210020855e-05,
      "loss": 1.2542,
      "step": 368
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.999001511584731e-05,
      "loss": 1.3477,
      "step": 369
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998994074579144e-05,
      "loss": 0.5526,
      "step": 370
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998986609985405e-05,
      "loss": 0.4528,
      "step": 371
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998979117803598e-05,
      "loss": 0.3689,
      "step": 372
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9989715980338045e-05,
      "loss": 0.8709,
      "step": 373
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9989640506761085e-05,
      "loss": 1.1167,
      "step": 374
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998956475730593e-05,
      "loss": 0.9943,
      "step": 375
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998948873197342e-05,
      "loss": 0.9485,
      "step": 376
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9989412430764384e-05,
      "loss": 0.9614,
      "step": 377
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998933585367967e-05,
      "loss": 0.9506,
      "step": 378
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998925900072012e-05,
      "loss": 1.0299,
      "step": 379
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9989181871886595e-05,
      "loss": 0.8741,
      "step": 380
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9989104467179934e-05,
      "loss": 1.0691,
      "step": 381
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9989026786601e-05,
      "loss": 1.1778,
      "step": 382
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998894883015064e-05,
      "loss": 0.929,
      "step": 383
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998887059782972e-05,
      "loss": 0.9856,
      "step": 384
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99887920896391e-05,
      "loss": 1.646,
      "step": 385
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998871330557966e-05,
      "loss": 1.6475,
      "step": 386
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998863424565226e-05,
      "loss": 0.9436,
      "step": 387
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998855490985776e-05,
      "loss": 1.1232,
      "step": 388
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998847529819706e-05,
      "loss": 1.4663,
      "step": 389
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9988395410671026e-05,
      "loss": 1.3766,
      "step": 390
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998831524728054e-05,
      "loss": 1.7138,
      "step": 391
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9988234808026494e-05,
      "loss": 1.0251,
      "step": 392
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998815409290976e-05,
      "loss": 0.983,
      "step": 393
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998807310193125e-05,
      "loss": 1.467,
      "step": 394
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998799183509184e-05,
      "loss": 1.1201,
      "step": 395
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998791029239244e-05,
      "loss": 1.0502,
      "step": 396
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998782847383395e-05,
      "loss": 0.9854,
      "step": 397
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9987746379417263e-05,
      "loss": 1.0427,
      "step": 398
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998766400914329e-05,
      "loss": 0.9387,
      "step": 399
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998758136301295e-05,
      "loss": 0.7664,
      "step": 400
    },
    {
      "epoch": 0.01,
      "eval_loss": 0.9967568516731262,
      "eval_runtime": 288.8989,
      "eval_samples_per_second": 7.446,
      "eval_steps_per_second": 1.862,
      "step": 400
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9987498441027135e-05,
      "loss": 1.0275,
      "step": 401
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9987415243186774e-05,
      "loss": 1.0794,
      "step": 402
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998733176949279e-05,
      "loss": 1.0415,
      "step": 403
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99872480199461e-05,
      "loss": 1.02,
      "step": 404
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998716399454761e-05,
      "loss": 1.0777,
      "step": 405
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998707969329828e-05,
      "loss": 0.9855,
      "step": 406
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998699511619902e-05,
      "loss": 1.0323,
      "step": 407
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9986910263250774e-05,
      "loss": 0.9186,
      "step": 408
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998682513445446e-05,
      "loss": 0.9138,
      "step": 409
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998673972981105e-05,
      "loss": 0.9507,
      "step": 410
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9986654049321456e-05,
      "loss": 1.0447,
      "step": 411
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998656809298663e-05,
      "loss": 1.1737,
      "step": 412
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998648186080754e-05,
      "loss": 1.0049,
      "step": 413
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998639535278512e-05,
      "loss": 1.0193,
      "step": 414
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998630856892033e-05,
      "loss": 1.0264,
      "step": 415
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998622150921413e-05,
      "loss": 1.0116,
      "step": 416
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9986134173667476e-05,
      "loss": 1.068,
      "step": 417
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998604656228133e-05,
      "loss": 1.2579,
      "step": 418
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9985958675056674e-05,
      "loss": 1.423,
      "step": 419
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998587051199447e-05,
      "loss": 1.0376,
      "step": 420
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998578207309568e-05,
      "loss": 1.0984,
      "step": 421
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9985693358361296e-05,
      "loss": 1.0913,
      "step": 422
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998560436779229e-05,
      "loss": 0.963,
      "step": 423
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9985515101389646e-05,
      "loss": 0.9741,
      "step": 424
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998542555915435e-05,
      "loss": 0.9893,
      "step": 425
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9985335741087394e-05,
      "loss": 0.9438,
      "step": 426
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998524564718976e-05,
      "loss": 1.0406,
      "step": 427
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9985155277462454e-05,
      "loss": 1.0179,
      "step": 428
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9985064631906465e-05,
      "loss": 1.1011,
      "step": 429
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9984973710522796e-05,
      "loss": 1.0293,
      "step": 430
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998488251331245e-05,
      "loss": 0.9017,
      "step": 431
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9984791040276435e-05,
      "loss": 0.961,
      "step": 432
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998469929141576e-05,
      "loss": 1.0012,
      "step": 433
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998460726673145e-05,
      "loss": 1.0277,
      "step": 434
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99845149662245e-05,
      "loss": 1.0168,
      "step": 435
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9984422389895936e-05,
      "loss": 1.0012,
      "step": 436
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998432953774679e-05,
      "loss": 1.1104,
      "step": 437
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9984236409778074e-05,
      "loss": 1.048,
      "step": 438
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9984143005990826e-05,
      "loss": 1.0545,
      "step": 439
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9984049326386076e-05,
      "loss": 1.0711,
      "step": 440
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9983955370964854e-05,
      "loss": 1.0356,
      "step": 441
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9983861139728196e-05,
      "loss": 0.9776,
      "step": 442
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998376663267714e-05,
      "loss": 1.0555,
      "step": 443
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998367184981274e-05,
      "loss": 0.9788,
      "step": 444
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998357679113603e-05,
      "loss": 1.185,
      "step": 445
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.9983481456648074e-05,
      "loss": 0.5851,
      "step": 446
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998338584634992e-05,
      "loss": 0.9349,
      "step": 447
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998328996024261e-05,
      "loss": 1.1246,
      "step": 448
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.998319379832722e-05,
      "loss": 0.8015,
      "step": 449
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.99830973606048e-05,
      "loss": 0.9487,
      "step": 450
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998300064707642e-05,
      "loss": 0.8586,
      "step": 451
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998290365774314e-05,
      "loss": 0.9738,
      "step": 452
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998280639260604e-05,
      "loss": 0.9544,
      "step": 453
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998270885166619e-05,
      "loss": 0.9186,
      "step": 454
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9982611034924674e-05,
      "loss": 1.0712,
      "step": 455
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998251294238256e-05,
      "loss": 0.9323,
      "step": 456
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9982414574040946e-05,
      "loss": 1.0256,
      "step": 457
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99823159299009e-05,
      "loss": 1.0425,
      "step": 458
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9982217009963524e-05,
      "loss": 1.0384,
      "step": 459
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99821178142299e-05,
      "loss": 1.0438,
      "step": 460
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9982018342701134e-05,
      "loss": 1.2086,
      "step": 461
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998191859537832e-05,
      "loss": 0.9009,
      "step": 462
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998181857226255e-05,
      "loss": 0.9238,
      "step": 463
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998171827335494e-05,
      "loss": 1.0177,
      "step": 464
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9981617698656596e-05,
      "loss": 0.7563,
      "step": 465
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998151684816862e-05,
      "loss": 0.8479,
      "step": 466
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998141572189214e-05,
      "loss": 0.9442,
      "step": 467
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998131431982826e-05,
      "loss": 1.2583,
      "step": 468
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99812126419781e-05,
      "loss": 0.8496,
      "step": 469
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9981110688342784e-05,
      "loss": 0.9856,
      "step": 470
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998100845892344e-05,
      "loss": 1.1074,
      "step": 471
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99809059537212e-05,
      "loss": 1.0568,
      "step": 472
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9980803172737187e-05,
      "loss": 1.0189,
      "step": 473
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998070011597255e-05,
      "loss": 0.9945,
      "step": 474
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99805967834284e-05,
      "loss": 1.0108,
      "step": 475
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99804931751059e-05,
      "loss": 1.0974,
      "step": 476
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998038929100619e-05,
      "loss": 1.0148,
      "step": 477
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998028513113042e-05,
      "loss": 0.9532,
      "step": 478
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.998018069547973e-05,
      "loss": 0.9243,
      "step": 479
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9980075984055276e-05,
      "loss": 0.9455,
      "step": 480
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997997099685822e-05,
      "loss": 1.0953,
      "step": 481
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997986573388971e-05,
      "loss": 1.0018,
      "step": 482
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9979760195150925e-05,
      "loss": 1.1019,
      "step": 483
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997965438064301e-05,
      "loss": 1.8503,
      "step": 484
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997954829036715e-05,
      "loss": 2.9905,
      "step": 485
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997944192432451e-05,
      "loss": 5.7033,
      "step": 486
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997933528251626e-05,
      "loss": 1.1438,
      "step": 487
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9979228364943584e-05,
      "loss": 1.084,
      "step": 488
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997912117160766e-05,
      "loss": 1.316,
      "step": 489
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9979013702509664e-05,
      "loss": 1.0081,
      "step": 490
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997890595765079e-05,
      "loss": 0.9712,
      "step": 491
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997879793703223e-05,
      "loss": 1.0164,
      "step": 492
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997868964065517e-05,
      "loss": 0.7017,
      "step": 493
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9978581068520816e-05,
      "loss": 1.0308,
      "step": 494
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997847222063036e-05,
      "loss": 1.1384,
      "step": 495
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9978363096984996e-05,
      "loss": 1.0204,
      "step": 496
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997825369758594e-05,
      "loss": 0.9767,
      "step": 497
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9978144022434395e-05,
      "loss": 1.0417,
      "step": 498
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997803407153157e-05,
      "loss": 1.0168,
      "step": 499
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997792384487867e-05,
      "loss": 1.1004,
      "step": 500
    },
    {
      "epoch": 0.02,
      "eval_loss": 0.9845504760742188,
      "eval_runtime": 308.8952,
      "eval_samples_per_second": 6.964,
      "eval_steps_per_second": 1.742,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997781334247694e-05,
      "loss": 1.0866,
      "step": 501
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9977702564327576e-05,
      "loss": 0.9923,
      "step": 502
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997759151043181e-05,
      "loss": 0.9924,
      "step": 503
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997748018079086e-05,
      "loss": 1.0721,
      "step": 504
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997736857540597e-05,
      "loss": 0.9604,
      "step": 505
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9977256694278354e-05,
      "loss": 1.2194,
      "step": 506
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9977144537409256e-05,
      "loss": 1.0647,
      "step": 507
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997703210479992e-05,
      "loss": 1.0575,
      "step": 508
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997691939645157e-05,
      "loss": 1.0612,
      "step": 509
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997680641236548e-05,
      "loss": 1.0342,
      "step": 510
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997669315254286e-05,
      "loss": 1.0266,
      "step": 511
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9976579616984993e-05,
      "loss": 1.0379,
      "step": 512
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9976465805693114e-05,
      "loss": 0.9817,
      "step": 513
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9976351718668476e-05,
      "loss": 0.8869,
      "step": 514
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997623735591236e-05,
      "loss": 1.0433,
      "step": 515
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997612271742601e-05,
      "loss": 0.9673,
      "step": 516
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99760078032107e-05,
      "loss": 0.9448,
      "step": 517
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997589261326769e-05,
      "loss": 1.1111,
      "step": 518
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9975777147598256e-05,
      "loss": 0.853,
      "step": 519
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997566140620368e-05,
      "loss": 1.0873,
      "step": 520
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9975545389085234e-05,
      "loss": 1.5826,
      "step": 521
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997542909624419e-05,
      "loss": 1.5843,
      "step": 522
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997531252768184e-05,
      "loss": 1.3157,
      "step": 523
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997519568339948e-05,
      "loss": 1.0371,
      "step": 524
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997507856339839e-05,
      "loss": 0.6945,
      "step": 525
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997496116767986e-05,
      "loss": 0.9025,
      "step": 526
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997484349624518e-05,
      "loss": 1.0725,
      "step": 527
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997472554909568e-05,
      "loss": 0.9203,
      "step": 528
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9974607326232625e-05,
      "loss": 1.0739,
      "step": 529
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997448882765734e-05,
      "loss": 0.7512,
      "step": 530
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997437005337113e-05,
      "loss": 1.0066,
      "step": 531
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9974251003375304e-05,
      "loss": 0.9411,
      "step": 532
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997413167767118e-05,
      "loss": 0.8867,
      "step": 533
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9974012076260076e-05,
      "loss": 1.1637,
      "step": 534
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9973892199143305e-05,
      "loss": 1.4102,
      "step": 535
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9973772046322195e-05,
      "loss": 1.0013,
      "step": 536
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997365161779808e-05,
      "loss": 1.2617,
      "step": 537
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997353091357228e-05,
      "loss": 1.0226,
      "step": 538
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997340993364612e-05,
      "loss": 1.1814,
      "step": 539
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9973288678020945e-05,
      "loss": 1.053,
      "step": 540
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9973167146698105e-05,
      "loss": 0.9326,
      "step": 541
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997304533967893e-05,
      "loss": 0.9954,
      "step": 542
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997292325696475e-05,
      "loss": 1.1234,
      "step": 543
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9972800898556935e-05,
      "loss": 0.7946,
      "step": 544
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997267826445684e-05,
      "loss": 1.062,
      "step": 545
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9972555354665794e-05,
      "loss": 1.1972,
      "step": 546
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997243216918517e-05,
      "loss": 1.0852,
      "step": 547
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997230870801633e-05,
      "loss": 0.9966,
      "step": 548
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997218497116063e-05,
      "loss": 1.263,
      "step": 549
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997206095861944e-05,
      "loss": 1.387,
      "step": 550
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997193667039412e-05,
      "loss": 1.3484,
      "step": 551
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997181210648606e-05,
      "loss": 1.0303,
      "step": 552
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997168726689662e-05,
      "loss": 0.9796,
      "step": 553
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9971562151627175e-05,
      "loss": 1.116,
      "step": 554
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997143676067912e-05,
      "loss": 0.9614,
      "step": 555
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997131109405384e-05,
      "loss": 0.947,
      "step": 556
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997118515175271e-05,
      "loss": 1.151,
      "step": 557
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997105893377712e-05,
      "loss": 0.5109,
      "step": 558
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997093244012848e-05,
      "loss": 0.4506,
      "step": 559
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997080567080817e-05,
      "loss": 0.3132,
      "step": 560
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997067862581759e-05,
      "loss": 0.5049,
      "step": 561
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997055130515815e-05,
      "loss": 1.0524,
      "step": 562
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997042370883126e-05,
      "loss": 0.9449,
      "step": 563
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997029583683832e-05,
      "loss": 1.0065,
      "step": 564
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997016768918075e-05,
      "loss": 0.8286,
      "step": 565
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.997003926585995e-05,
      "loss": 1.2209,
      "step": 566
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996991056687734e-05,
      "loss": 0.9944,
      "step": 567
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996978159223436e-05,
      "loss": 0.829,
      "step": 568
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9969652341932414e-05,
      "loss": 1.0127,
      "step": 569
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9969522815972944e-05,
      "loss": 0.9786,
      "step": 570
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996939301435736e-05,
      "loss": 1.0136,
      "step": 571
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996926293708711e-05,
      "loss": 0.953,
      "step": 572
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9969132584163634e-05,
      "loss": 1.1876,
      "step": 573
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9969001955588356e-05,
      "loss": 0.907,
      "step": 574
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996887105136273e-05,
      "loss": 1.0807,
      "step": 575
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996873987148819e-05,
      "loss": 0.9871,
      "step": 576
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99686084159662e-05,
      "loss": 2.0506,
      "step": 577
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9968476684798195e-05,
      "loss": 0.5178,
      "step": 578
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996834467798564e-05,
      "loss": 1.0064,
      "step": 579
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996821239552999e-05,
      "loss": 1.0122,
      "step": 580
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99680798374327e-05,
      "loss": 1.0288,
      "step": 581
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996794700369525e-05,
      "loss": 1.0822,
      "step": 582
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9967813894319085e-05,
      "loss": 1.0782,
      "step": 583
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996768050930568e-05,
      "loss": 1.0729,
      "step": 584
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996754684865651e-05,
      "loss": 1.0238,
      "step": 585
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996741291237306e-05,
      "loss": 0.9225,
      "step": 586
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99672787004568e-05,
      "loss": 1.2491,
      "step": 587
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9967144212909204e-05,
      "loss": 1.0257,
      "step": 588
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996700944973177e-05,
      "loss": 1.1085,
      "step": 589
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996687441092598e-05,
      "loss": 1.0762,
      "step": 590
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996673909649333e-05,
      "loss": 0.9597,
      "step": 591
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9966603506435306e-05,
      "loss": 1.0122,
      "step": 592
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99664676407534e-05,
      "loss": 1.026,
      "step": 593
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996633149944913e-05,
      "loss": 0.9911,
      "step": 594
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996619508252399e-05,
      "loss": 1.0355,
      "step": 595
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996605838997947e-05,
      "loss": 0.9631,
      "step": 596
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996592142181711e-05,
      "loss": 1.0802,
      "step": 597
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99657841780384e-05,
      "loss": 0.9716,
      "step": 598
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996564665864486e-05,
      "loss": 1.0427,
      "step": 599
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996550886363801e-05,
      "loss": 1.0257,
      "step": 600
    },
    {
      "epoch": 0.02,
      "eval_loss": 0.9789223074913025,
      "eval_runtime": 308.7995,
      "eval_samples_per_second": 6.966,
      "eval_steps_per_second": 1.742,
      "step": 600
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996537079301937e-05,
      "loss": 0.9975,
      "step": 601
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996523244679047e-05,
      "loss": 0.9946,
      "step": 602
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9965093824952826e-05,
      "loss": 1.1422,
      "step": 603
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9964954927507977e-05,
      "loss": 1.0361,
      "step": 604
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996481575445745e-05,
      "loss": 0.9316,
      "step": 605
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9964676305802794e-05,
      "loss": 0.9053,
      "step": 606
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996453658154553e-05,
      "loss": 0.8906,
      "step": 607
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996439658168721e-05,
      "loss": 2.8933,
      "step": 608
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996425630622939e-05,
      "loss": 1.0506,
      "step": 609
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99641157551736e-05,
      "loss": 1.1427,
      "step": 610
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9963974928521407e-05,
      "loss": 1.0648,
      "step": 611
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996383382627435e-05,
      "loss": 1.2693,
      "step": 612
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9963692448433996e-05,
      "loss": 1.0419,
      "step": 613
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996355079500191e-05,
      "loss": 1.055,
      "step": 614
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996340886597966e-05,
      "loss": 0.9107,
      "step": 615
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9963266661368784e-05,
      "loss": 0.9327,
      "step": 616
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996312418117088e-05,
      "loss": 1.0163,
      "step": 617
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996298142538752e-05,
      "loss": 1.1061,
      "step": 618
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996283839402026e-05,
      "loss": 1.0109,
      "step": 619
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99626950870707e-05,
      "loss": 1.041,
      "step": 620
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9962551504540414e-05,
      "loss": 0.9262,
      "step": 621
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996240764643099e-05,
      "loss": 0.8448,
      "step": 622
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9962263512744e-05,
      "loss": 1.0745,
      "step": 623
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996211910348106e-05,
      "loss": 0.8993,
      "step": 624
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9961974418643746e-05,
      "loss": 1.0451,
      "step": 625
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996182945823366e-05,
      "loss": 0.8029,
      "step": 626
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9961684222252415e-05,
      "loss": 1.0405,
      "step": 627
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9961538710701595e-05,
      "loss": 1.0596,
      "step": 628
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9961392923582815e-05,
      "loss": 1.006,
      "step": 629
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996124686089769e-05,
      "loss": 1.0663,
      "step": 630
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996110052264782e-05,
      "loss": 0.5662,
      "step": 631
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9960953908834825e-05,
      "loss": 0.5571,
      "step": 632
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9960807019460335e-05,
      "loss": 1.0791,
      "step": 633
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996065985452596e-05,
      "loss": 1.1459,
      "step": 634
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996051241403332e-05,
      "loss": 5.449,
      "step": 635
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9960364697984055e-05,
      "loss": 5.4546,
      "step": 636
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996021670637979e-05,
      "loss": 0.3953,
      "step": 637
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.996006843922216e-05,
      "loss": 1.0451,
      "step": 638
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9959919896512805e-05,
      "loss": 1.1229,
      "step": 639
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995977107825336e-05,
      "loss": 1.0272,
      "step": 640
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995962198444546e-05,
      "loss": 1.0769,
      "step": 641
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995947261509077e-05,
      "loss": 0.7592,
      "step": 642
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995932297019092e-05,
      "loss": 1.0298,
      "step": 643
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9959173049747574e-05,
      "loss": 1.0096,
      "step": 644
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9959022853762394e-05,
      "loss": 0.9812,
      "step": 645
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995887238223702e-05,
      "loss": 1.2525,
      "step": 646
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9958721635173125e-05,
      "loss": 0.9556,
      "step": 647
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995857061257236e-05,
      "loss": 1.0607,
      "step": 648
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995841931443641e-05,
      "loss": 1.01,
      "step": 649
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995826774076693e-05,
      "loss": 0.7448,
      "step": 650
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99581158915656e-05,
      "loss": 0.7053,
      "step": 651
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99579637668341e-05,
      "loss": 0.6826,
      "step": 652
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9957811366574116e-05,
      "loss": 0.6064,
      "step": 653
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9957658690787315e-05,
      "loss": 0.6131,
      "step": 654
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995750573947539e-05,
      "loss": 0.6392,
      "step": 655
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995735251264001e-05,
      "loss": 1.0116,
      "step": 656
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99571990102829e-05,
      "loss": 0.9327,
      "step": 657
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995704523240574e-05,
      "loss": 1.0833,
      "step": 658
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995689117901023e-05,
      "loss": 1.0715,
      "step": 659
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995673685009805e-05,
      "loss": 1.2433,
      "step": 660
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995658224567093e-05,
      "loss": 1.1899,
      "step": 661
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995642736573057e-05,
      "loss": 7.4687,
      "step": 662
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995627221027868e-05,
      "loss": 6.798,
      "step": 663
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9956116779316964e-05,
      "loss": 6.408,
      "step": 664
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9955961072847145e-05,
      "loss": 6.1311,
      "step": 665
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995580509087094e-05,
      "loss": 5.9937,
      "step": 666
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995564883339008e-05,
      "loss": 2.9751,
      "step": 667
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995549230040627e-05,
      "loss": 1.1381,
      "step": 668
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9955335491921264e-05,
      "loss": 1.0907,
      "step": 669
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995517840793677e-05,
      "loss": 1.0098,
      "step": 670
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995502104845453e-05,
      "loss": 1.0451,
      "step": 671
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995486341347629e-05,
      "loss": 0.9634,
      "step": 672
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9954705503003785e-05,
      "loss": 0.9586,
      "step": 673
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9954547317038746e-05,
      "loss": 1.0372,
      "step": 674
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995438885558294e-05,
      "loss": 1.2027,
      "step": 675
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99542301186381e-05,
      "loss": 0.9895,
      "step": 676
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995407110620599e-05,
      "loss": 1.0896,
      "step": 677
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9953911818288355e-05,
      "loss": 1.0305,
      "step": 678
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995375225488696e-05,
      "loss": 0.8958,
      "step": 679
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995359241600356e-05,
      "loss": 0.8556,
      "step": 680
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9953432301639935e-05,
      "loss": 0.9318,
      "step": 681
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9953271911797834e-05,
      "loss": 1.1007,
      "step": 682
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995311124647904e-05,
      "loss": 0.999,
      "step": 683
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995295030568532e-05,
      "loss": 0.9797,
      "step": 684
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995278908941845e-05,
      "loss": 0.9081,
      "step": 685
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9952627597680224e-05,
      "loss": 1.0406,
      "step": 686
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995246583047241e-05,
      "loss": 1.1092,
      "step": 687
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995230378779678e-05,
      "loss": 0.9795,
      "step": 688
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995214146965517e-05,
      "loss": 0.9696,
      "step": 689
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995197887604932e-05,
      "loss": 1.2503,
      "step": 690
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995181600698106e-05,
      "loss": 0.9896,
      "step": 691
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995165286245217e-05,
      "loss": 1.0376,
      "step": 692
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995148944246445e-05,
      "loss": 0.8537,
      "step": 693
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995132574701972e-05,
      "loss": 0.8631,
      "step": 694
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995116177611978e-05,
      "loss": 1.1774,
      "step": 695
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995099752976643e-05,
      "loss": 1.0297,
      "step": 696
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995083300796149e-05,
      "loss": 1.071,
      "step": 697
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995066821070679e-05,
      "loss": 0.9615,
      "step": 698
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9950503138004134e-05,
      "loss": 1.0703,
      "step": 699
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995033778985534e-05,
      "loss": 1.0865,
      "step": 700
    },
    {
      "epoch": 0.02,
      "eval_loss": 0.9759284853935242,
      "eval_runtime": 290.8289,
      "eval_samples_per_second": 7.396,
      "eval_steps_per_second": 1.85,
      "step": 700
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9950172166262246e-05,
      "loss": 1.3198,
      "step": 701
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.995000626722667e-05,
      "loss": 1.2602,
      "step": 702
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994984009275046e-05,
      "loss": 1.0985,
      "step": 703
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994967364283543e-05,
      "loss": 1.0737,
      "step": 704
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9949506917483425e-05,
      "loss": 1.0172,
      "step": 705
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9949339916696284e-05,
      "loss": 1.3928,
      "step": 706
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994917264047586e-05,
      "loss": 1.7214,
      "step": 707
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9949005088823994e-05,
      "loss": 1.6469,
      "step": 708
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994883726174253e-05,
      "loss": 1.4957,
      "step": 709
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994866915923333e-05,
      "loss": 1.37,
      "step": 710
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994850078129825e-05,
      "loss": 1.3935,
      "step": 711
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994833212793914e-05,
      "loss": 1.5346,
      "step": 712
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994816319915787e-05,
      "loss": 1.4269,
      "step": 713
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99479939949563e-05,
      "loss": 1.6038,
      "step": 714
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99478245153363e-05,
      "loss": 1.4522,
      "step": 715
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994765476029973e-05,
      "loss": 1.3598,
      "step": 716
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994748472984848e-05,
      "loss": 1.1834,
      "step": 717
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994731442398443e-05,
      "loss": 1.1222,
      "step": 718
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994714384270944e-05,
      "loss": 1.1497,
      "step": 719
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994697298602542e-05,
      "loss": 0.8855,
      "step": 720
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994680185393423e-05,
      "loss": 1.083,
      "step": 721
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994663044643777e-05,
      "loss": 1.1181,
      "step": 722
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994645876353794e-05,
      "loss": 1.2041,
      "step": 723
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9946286805236616e-05,
      "loss": 1.0189,
      "step": 724
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9946114571535716e-05,
      "loss": 1.1624,
      "step": 725
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994594206243714e-05,
      "loss": 1.0823,
      "step": 726
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9945769277942776e-05,
      "loss": 1.4095,
      "step": 727
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9945596218054545e-05,
      "loss": 3.9685,
      "step": 728
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994542288277435e-05,
      "loss": 3.2173,
      "step": 729
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9945249272104115e-05,
      "loss": 0.9638,
      "step": 730
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994507538604575e-05,
      "loss": 1.0859,
      "step": 731
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9944901224601174e-05,
      "loss": 0.9879,
      "step": 732
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9944726787772314e-05,
      "loss": 1.1589,
      "step": 733
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994455207556108e-05,
      "loss": 1.2655,
      "step": 734
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9944377087969426e-05,
      "loss": 1.0326,
      "step": 735
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9944201824999274e-05,
      "loss": 1.0201,
      "step": 736
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9944026286652544e-05,
      "loss": 1.0069,
      "step": 737
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994385047293119e-05,
      "loss": 1.1747,
      "step": 738
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9943674383837144e-05,
      "loss": 1.1272,
      "step": 739
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994349801937236e-05,
      "loss": 1.0504,
      "step": 740
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9943321379538785e-05,
      "loss": 0.9558,
      "step": 741
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994314446433835e-05,
      "loss": 0.9085,
      "step": 742
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994296727377303e-05,
      "loss": 0.9688,
      "step": 743
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994278980784478e-05,
      "loss": 1.0287,
      "step": 744
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994261206655554e-05,
      "loss": 0.9381,
      "step": 745
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9942434049907294e-05,
      "loss": 0.948,
      "step": 746
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994225575790199e-05,
      "loss": 0.9114,
      "step": 747
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.99420771905416e-05,
      "loss": 1.0369,
      "step": 748
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.994189834782811e-05,
      "loss": 0.8366,
      "step": 749
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.994171922976348e-05,
      "loss": 0.6539,
      "step": 750
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9941539836349696e-05,
      "loss": 1.0079,
      "step": 751
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.994136016758872e-05,
      "loss": 1.0984,
      "step": 752
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.994118022348256e-05,
      "loss": 1.051,
      "step": 753
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9941000004033186e-05,
      "loss": 0.6536,
      "step": 754
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.99408195092426e-05,
      "loss": 1.0081,
      "step": 755
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.994063873911278e-05,
      "loss": 0.9596,
      "step": 756
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.994045769364574e-05,
      "loss": 0.9734,
      "step": 757
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9940276372843454e-05,
      "loss": 0.9001,
      "step": 758
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9940094776707945e-05,
      "loss": 1.0639,
      "step": 759
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993991290524121e-05,
      "loss": 1.0456,
      "step": 760
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9939730758445255e-05,
      "loss": 0.9674,
      "step": 761
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993954833632209e-05,
      "loss": 1.2029,
      "step": 762
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993936563887374e-05,
      "loss": 0.8811,
      "step": 763
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993918266610221e-05,
      "loss": 1.0283,
      "step": 764
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993899941800953e-05,
      "loss": 1.3218,
      "step": 765
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993881589459771e-05,
      "loss": 1.0537,
      "step": 766
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993863209586878e-05,
      "loss": 0.9448,
      "step": 767
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993844802182478e-05,
      "loss": 1.2968,
      "step": 768
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9938263672467725e-05,
      "loss": 1.6229,
      "step": 769
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9938079047799666e-05,
      "loss": 1.5709,
      "step": 770
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993789414782264e-05,
      "loss": 1.1705,
      "step": 771
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993770897253867e-05,
      "loss": 0.9542,
      "step": 772
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993752352194982e-05,
      "loss": 1.01,
      "step": 773
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9937337796058133e-05,
      "loss": 0.9992,
      "step": 774
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993715179486565e-05,
      "loss": 1.0876,
      "step": 775
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993696551837444e-05,
      "loss": 1.0426,
      "step": 776
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993677896658654e-05,
      "loss": 1.0077,
      "step": 777
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9936592139504026e-05,
      "loss": 1.006,
      "step": 778
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993640503712895e-05,
      "loss": 0.9732,
      "step": 779
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993621765946339e-05,
      "loss": 1.0063,
      "step": 780
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.99360300065094e-05,
      "loss": 0.8799,
      "step": 781
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9935842078269055e-05,
      "loss": 1.0341,
      "step": 782
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9935653874744445e-05,
      "loss": 0.9103,
      "step": 783
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993546539593763e-05,
      "loss": 1.0255,
      "step": 784
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993527664185069e-05,
      "loss": 1.2812,
      "step": 785
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993508761248572e-05,
      "loss": 1.2012,
      "step": 786
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9934898307844804e-05,
      "loss": 1.2031,
      "step": 787
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993470872793003e-05,
      "loss": 1.1346,
      "step": 788
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993451887274349e-05,
      "loss": 1.1224,
      "step": 789
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9934328742287285e-05,
      "loss": 1.0004,
      "step": 790
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9934138336563504e-05,
      "loss": 1.0812,
      "step": 791
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9933947655574257e-05,
      "loss": 0.9117,
      "step": 792
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993375669932164e-05,
      "loss": 0.9467,
      "step": 793
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9933565467807776e-05,
      "loss": 1.0217,
      "step": 794
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993337396103477e-05,
      "loss": 3.7284,
      "step": 795
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993318217900472e-05,
      "loss": 5.8783,
      "step": 796
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993299012171977e-05,
      "loss": 5.9121,
      "step": 797
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993279778918203e-05,
      "loss": 5.8477,
      "step": 798
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993260518139361e-05,
      "loss": 5.8715,
      "step": 799
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993241229835666e-05,
      "loss": 5.8723,
      "step": 800
    },
    {
      "epoch": 0.03,
      "eval_loss": 1.0035531520843506,
      "eval_runtime": 286.4169,
      "eval_samples_per_second": 7.51,
      "eval_steps_per_second": 1.878,
      "step": 800
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993221914007329e-05,
      "loss": 5.8212,
      "step": 801
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993202570654564e-05,
      "loss": 5.8071,
      "step": 802
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993183199777584e-05,
      "loss": 1.6276,
      "step": 803
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993163801376603e-05,
      "loss": 1.1297,
      "step": 804
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993144375451837e-05,
      "loss": 1.0226,
      "step": 805
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993124922003498e-05,
      "loss": 0.9612,
      "step": 806
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993105441031802e-05,
      "loss": 1.0639,
      "step": 807
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9930859325369636e-05,
      "loss": 1.0766,
      "step": 808
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993066396519198e-05,
      "loss": 0.8815,
      "step": 809
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9930468329787214e-05,
      "loss": 0.9521,
      "step": 810
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9930272419157495e-05,
      "loss": 1.0109,
      "step": 811
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.993007623330499e-05,
      "loss": 1.077,
      "step": 812
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992987977223186e-05,
      "loss": 0.9419,
      "step": 813
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992968303594027e-05,
      "loss": 1.0663,
      "step": 814
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9929486024432406e-05,
      "loss": 1.0388,
      "step": 815
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992928873771042e-05,
      "loss": 0.9759,
      "step": 816
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992909117577651e-05,
      "loss": 0.5938,
      "step": 817
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992889333863285e-05,
      "loss": 0.7563,
      "step": 818
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9928695226281635e-05,
      "loss": 1.0307,
      "step": 819
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9928496838725035e-05,
      "loss": 1.0021,
      "step": 820
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992829817596525e-05,
      "loss": 1.005,
      "step": 821
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9928099238004465e-05,
      "loss": 0.7177,
      "step": 822
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9927900024844886e-05,
      "loss": 0.7951,
      "step": 823
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992770053648871e-05,
      "loss": 0.7398,
      "step": 824
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992750077293813e-05,
      "loss": 0.8266,
      "step": 825
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992730073419536e-05,
      "loss": 0.9131,
      "step": 826
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9927100420262605e-05,
      "loss": 1.0654,
      "step": 827
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992689983114209e-05,
      "loss": 0.7467,
      "step": 828
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9926698966836005e-05,
      "loss": 1.0499,
      "step": 829
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992649782734658e-05,
      "loss": 5.1456,
      "step": 830
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992629641267605e-05,
      "loss": 2.2626,
      "step": 831
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992609472282661e-05,
      "loss": 1.1422,
      "step": 832
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992589275780051e-05,
      "loss": 1.3367,
      "step": 833
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992569051759997e-05,
      "loss": 1.3097,
      "step": 834
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992548800222722e-05,
      "loss": 1.278,
      "step": 835
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992528521168449e-05,
      "loss": 1.2619,
      "step": 836
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9925082145974045e-05,
      "loss": 1.8006,
      "step": 837
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9924878805098094e-05,
      "loss": 0.987,
      "step": 838
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992467518905891e-05,
      "loss": 1.0752,
      "step": 839
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992447129785872e-05,
      "loss": 1.1045,
      "step": 840
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992426713149979e-05,
      "loss": 1.3264,
      "step": 841
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992406268998436e-05,
      "loss": 1.4496,
      "step": 842
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992385797331469e-05,
      "loss": 1.3956,
      "step": 843
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992365298149305e-05,
      "loss": 1.0576,
      "step": 844
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9923447714521696e-05,
      "loss": 1.0199,
      "step": 845
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.99232421724029e-05,
      "loss": 0.9795,
      "step": 846
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992303635513892e-05,
      "loss": 1.0281,
      "step": 847
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992283026273204e-05,
      "loss": 1.0085,
      "step": 848
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9922623895184526e-05,
      "loss": 1.1058,
      "step": 849
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992241725249866e-05,
      "loss": 1.0282,
      "step": 850
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992221033467672e-05,
      "loss": 0.8185,
      "step": 851
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9922003141720995e-05,
      "loss": 0.8294,
      "step": 852
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9921795673633777e-05,
      "loss": 1.2069,
      "step": 853
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992158793041735e-05,
      "loss": 0.9052,
      "step": 854
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9921379912073994e-05,
      "loss": 1.0296,
      "step": 855
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992117161860603e-05,
      "loss": 1.053,
      "step": 856
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992096305001575e-05,
      "loss": 1.1566,
      "step": 857
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9920754206305444e-05,
      "loss": 0.9749,
      "step": 858
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.992054508747743e-05,
      "loss": 0.9983,
      "step": 859
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9920335693534016e-05,
      "loss": 1.0722,
      "step": 860
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9920126024477505e-05,
      "loss": 0.9988,
      "step": 861
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991991608031022e-05,
      "loss": 0.3297,
      "step": 862
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991970586103448e-05,
      "loss": 0.6432,
      "step": 863
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991949536665259e-05,
      "loss": 0.5689,
      "step": 864
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.99192845971669e-05,
      "loss": 0.309,
      "step": 865
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991907355257972e-05,
      "loss": 0.3805,
      "step": 866
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991886223289338e-05,
      "loss": 0.6213,
      "step": 867
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991865063811021e-05,
      "loss": 1.058,
      "step": 868
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9918438768232554e-05,
      "loss": 2.0019,
      "step": 869
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991822662326276e-05,
      "loss": 1.1491,
      "step": 870
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991801420320315e-05,
      "loss": 0.7992,
      "step": 871
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9917801508056075e-05,
      "loss": 1.071,
      "step": 872
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991758853782389e-05,
      "loss": 1.0133,
      "step": 873
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991737529250893e-05,
      "loss": 0.4019,
      "step": 874
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991716177211357e-05,
      "loss": 1.2008,
      "step": 875
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991694797664016e-05,
      "loss": 1.0207,
      "step": 876
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991673390609106e-05,
      "loss": 1.1366,
      "step": 877
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9916519560468624e-05,
      "loss": 0.8884,
      "step": 878
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991630493977523e-05,
      "loss": 0.984,
      "step": 879
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991609004401324e-05,
      "loss": 1.0077,
      "step": 880
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991587487318503e-05,
      "loss": 1.0735,
      "step": 881
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991565942729298e-05,
      "loss": 0.9204,
      "step": 882
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991544370633946e-05,
      "loss": 0.9315,
      "step": 883
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991522771032686e-05,
      "loss": 0.9987,
      "step": 884
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9915011439257553e-05,
      "loss": 0.8596,
      "step": 885
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991479489313393e-05,
      "loss": 1.0373,
      "step": 886
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9914578071958394e-05,
      "loss": 0.9995,
      "step": 887
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9914360975733324e-05,
      "loss": 0.6941,
      "step": 888
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9914143604461124e-05,
      "loss": 0.482,
      "step": 889
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991392595814419e-05,
      "loss": 0.6967,
      "step": 890
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9913708036784925e-05,
      "loss": 0.8831,
      "step": 891
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991348984038574e-05,
      "loss": 0.9165,
      "step": 892
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9913271368949035e-05,
      "loss": 1.0235,
      "step": 893
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991305262247723e-05,
      "loss": 1.1501,
      "step": 894
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991283360097274e-05,
      "loss": 0.8943,
      "step": 895
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991261430443798e-05,
      "loss": 0.8168,
      "step": 896
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991239473287536e-05,
      "loss": 1.0032,
      "step": 897
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991217488628732e-05,
      "loss": 0.6643,
      "step": 898
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9911954764676286e-05,
      "loss": 0.9688,
      "step": 899
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9911734368044684e-05,
      "loss": 1.0675,
      "step": 900
    },
    {
      "epoch": 0.03,
      "eval_loss": 0.9832482933998108,
      "eval_runtime": 286.114,
      "eval_samples_per_second": 7.518,
      "eval_steps_per_second": 1.88,
      "step": 900
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991151369639495e-05,
      "loss": 1.1038,
      "step": 901
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991129274972951e-05,
      "loss": 0.9522,
      "step": 902
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9911071528050807e-05,
      "loss": 0.899,
      "step": 903
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9910850031361286e-05,
      "loss": 1.0327,
      "step": 904
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9910628259663404e-05,
      "loss": 1.0305,
      "step": 905
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.991040621295959e-05,
      "loss": 1.0655,
      "step": 906
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9910183891252304e-05,
      "loss": 1.0375,
      "step": 907
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9909961294544006e-05,
      "loss": 1.0094,
      "step": 908
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990973842283714e-05,
      "loss": 0.9754,
      "step": 909
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990951527613418e-05,
      "loss": 0.9716,
      "step": 910
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9909291854437576e-05,
      "loss": 1.0421,
      "step": 911
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.99090681577498e-05,
      "loss": 0.9665,
      "step": 912
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990884418607333e-05,
      "loss": 1.0886,
      "step": 913
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9908619939410635e-05,
      "loss": 0.8354,
      "step": 914
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9908395417764176e-05,
      "loss": 1.0642,
      "step": 915
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990817062113645e-05,
      "loss": 1.0418,
      "step": 916
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990794554952993e-05,
      "loss": 0.9147,
      "step": 917
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990772020294711e-05,
      "loss": 1.0243,
      "step": 918
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990749458139046e-05,
      "loss": 0.8944,
      "step": 919
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990726868486249e-05,
      "loss": 0.8025,
      "step": 920
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990704251336567e-05,
      "loss": 0.6817,
      "step": 921
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9906816066902516e-05,
      "loss": 1.7802,
      "step": 922
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990658934547553e-05,
      "loss": 1.8639,
      "step": 923
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990636234908721e-05,
      "loss": 1.8191,
      "step": 924
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9906135077740056e-05,
      "loss": 1.7942,
      "step": 925
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990590753143658e-05,
      "loss": 1.0075,
      "step": 926
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.99056797101793e-05,
      "loss": 1.0765,
      "step": 927
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9905451613970725e-05,
      "loss": 1.1144,
      "step": 928
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990522324281337e-05,
      "loss": 0.8783,
      "step": 929
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990499459670976e-05,
      "loss": 1.2054,
      "step": 930
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990476567566242e-05,
      "loss": 1.309,
      "step": 931
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990453647967389e-05,
      "loss": 1.0527,
      "step": 932
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9904307008746674e-05,
      "loss": 1.0586,
      "step": 933
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990407726288333e-05,
      "loss": 0.9283,
      "step": 934
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990384724208638e-05,
      "loss": 1.031,
      "step": 935
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990361694635835e-05,
      "loss": 1.1117,
      "step": 936
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990338637570182e-05,
      "loss": 0.9792,
      "step": 937
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990315553011931e-05,
      "loss": 1.0238,
      "step": 938
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990292440961337e-05,
      "loss": 0.9007,
      "step": 939
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9902693014186555e-05,
      "loss": 0.9795,
      "step": 940
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990246134384141e-05,
      "loss": 0.9988,
      "step": 941
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990222939858051e-05,
      "loss": 0.8989,
      "step": 942
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9901997178406405e-05,
      "loss": 1.3463,
      "step": 943
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9901764683321664e-05,
      "loss": 0.9825,
      "step": 944
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990153191332885e-05,
      "loss": 0.9895,
      "step": 945
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990129886843052e-05,
      "loss": 0.984,
      "step": 946
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990106554862928e-05,
      "loss": 0.9589,
      "step": 947
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990083195392767e-05,
      "loss": 1.0984,
      "step": 948
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990059808432829e-05,
      "loss": 0.9299,
      "step": 949
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9900363939833714e-05,
      "loss": 1.0476,
      "step": 950
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.990012952044654e-05,
      "loss": 1.0591,
      "step": 951
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9899894826169337e-05,
      "loss": 0.8999,
      "step": 952
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.98996598570047e-05,
      "loss": 3.0569,
      "step": 953
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989942461295523e-05,
      "loss": 0.9277,
      "step": 954
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9899189094023525e-05,
      "loss": 1.04,
      "step": 955
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989895330021217e-05,
      "loss": 1.0493,
      "step": 956
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989871723152379e-05,
      "loss": 1.0286,
      "step": 957
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989848088796097e-05,
      "loss": 0.9524,
      "step": 958
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989824426952634e-05,
      "loss": 1.0462,
      "step": 959
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.98980073762225e-05,
      "loss": 1.0895,
      "step": 960
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989777020805206e-05,
      "loss": 1.062,
      "step": 961
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989753276501765e-05,
      "loss": 0.9437,
      "step": 962
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989729504712189e-05,
      "loss": 1.0071,
      "step": 963
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9897057054367395e-05,
      "loss": 0.908,
      "step": 964
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9896818786756804e-05,
      "loss": 0.9619,
      "step": 965
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9896580244292736e-05,
      "loss": 0.617,
      "step": 966
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9896341426977834e-05,
      "loss": 0.7072,
      "step": 967
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989610233481472e-05,
      "loss": 0.9856,
      "step": 968
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9895862967806056e-05,
      "loss": 1.0616,
      "step": 969
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989562332595447e-05,
      "loss": 0.4031,
      "step": 970
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989538340926261e-05,
      "loss": 0.4459,
      "step": 971
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989514321773312e-05,
      "loss": 0.9105,
      "step": 972
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9894902751368654e-05,
      "loss": 1.2019,
      "step": 973
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9894662010171874e-05,
      "loss": 1.063,
      "step": 974
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9894420994145436e-05,
      "loss": 1.0836,
      "step": 975
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9894179703291995e-05,
      "loss": 0.938,
      "step": 976
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9893938137614215e-05,
      "loss": 1.7182,
      "step": 977
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9893696297114765e-05,
      "loss": 0.998,
      "step": 978
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989345418179632e-05,
      "loss": 0.9068,
      "step": 979
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9893211791661545e-05,
      "loss": 1.0015,
      "step": 980
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989296912671312e-05,
      "loss": 1.0304,
      "step": 981
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989272618695371e-05,
      "loss": 1.4666,
      "step": 982
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9892482972386025e-05,
      "loss": 1.1396,
      "step": 983
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989223948301273e-05,
      "loss": 1.07,
      "step": 984
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989199571883651e-05,
      "loss": 1.0957,
      "step": 985
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989175167986008e-05,
      "loss": 1.0131,
      "step": 986
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989150736608611e-05,
      "loss": 0.9387,
      "step": 987
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.98912627775173e-05,
      "loss": 1.017,
      "step": 988
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989101791415636e-05,
      "loss": 0.9977,
      "step": 989
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989077277600599e-05,
      "loss": 1.0281,
      "step": 990
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989052736306889e-05,
      "loss": 1.0954,
      "step": 991
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989028167534777e-05,
      "loss": 1.0957,
      "step": 992
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.989003571284536e-05,
      "loss": 0.9253,
      "step": 993
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988978947556435e-05,
      "loss": 1.0196,
      "step": 994
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988954296350747e-05,
      "loss": 0.904,
      "step": 995
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988929617667745e-05,
      "loss": 0.5741,
      "step": 996
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9889049115077005e-05,
      "loss": 0.8989,
      "step": 997
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988880177870886e-05,
      "loss": 1.1993,
      "step": 998
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988855416757575e-05,
      "loss": 1.035,
      "step": 999
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9888306281680405e-05,
      "loss": 1.025,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "eval_loss": 0.9734950661659241,
      "eval_runtime": 299.7473,
      "eval_samples_per_second": 7.176,
      "eval_steps_per_second": 1.795,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988805812102556e-05,
      "loss": 1.1113,
      "step": 1001
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9887809685613965e-05,
      "loss": 1.098,
      "step": 1002
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988756097544836e-05,
      "loss": 0.862,
      "step": 1003
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988731199053148e-05,
      "loss": 1.0189,
      "step": 1004
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988706273086608e-05,
      "loss": 1.1328,
      "step": 1005
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988681319645492e-05,
      "loss": 1.0056,
      "step": 1006
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9886563387300736e-05,
      "loss": 1.0506,
      "step": 1007
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9886313303406307e-05,
      "loss": 0.9781,
      "step": 1008
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988606294477438e-05,
      "loss": 1.1207,
      "step": 1009
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988581231140772e-05,
      "loss": 1.2202,
      "step": 1010
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.98855614033091e-05,
      "loss": 0.9572,
      "step": 1011
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988531022048129e-05,
      "loss": 0.7626,
      "step": 1012
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9885058762927054e-05,
      "loss": 1.0933,
      "step": 1013
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9884807030649175e-05,
      "loss": 0.4095,
      "step": 1014
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988455502365044e-05,
      "loss": 0.3238,
      "step": 1015
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9884302741933606e-05,
      "loss": 1.0389,
      "step": 1016
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9884050185501485e-05,
      "loss": 1.0175,
      "step": 1017
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988379735435685e-05,
      "loss": 0.6129,
      "step": 1018
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.98835442485025e-05,
      "loss": 0.6213,
      "step": 1019
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988329086794122e-05,
      "loss": 0.9749,
      "step": 1020
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988303721267581e-05,
      "loss": 1.1065,
      "step": 1021
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988278328270908e-05,
      "loss": 1.2167,
      "step": 1022
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988252907804383e-05,
      "loss": 1.0403,
      "step": 1023
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988227459868286e-05,
      "loss": 1.0556,
      "step": 1024
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9882019844628966e-05,
      "loss": 0.6849,
      "step": 1025
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9881764815884993e-05,
      "loss": 1.1415,
      "step": 1026
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988150951245374e-05,
      "loss": 0.5511,
      "step": 1027
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988125393433802e-05,
      "loss": 0.8831,
      "step": 1028
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9880998081540664e-05,
      "loss": 0.9677,
      "step": 1029
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988074195406449e-05,
      "loss": 0.9946,
      "step": 1030
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.988048555191233e-05,
      "loss": 1.0384,
      "step": 1031
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9880228875087004e-05,
      "loss": 1.0479,
      "step": 1032
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.987997192359136e-05,
      "loss": 1.0447,
      "step": 1033
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9879714697428234e-05,
      "loss": 0.8416,
      "step": 1034
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.987945719660045e-05,
      "loss": 0.9394,
      "step": 1035
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9879199421110865e-05,
      "loss": 0.9742,
      "step": 1036
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.987894137096233e-05,
      "loss": 1.2215,
      "step": 1037
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.987868304615768e-05,
      "loss": 1.0264,
      "step": 1038
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9878424446699764e-05,
      "loss": 1.0462,
      "step": 1039
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.987816557259145e-05,
      "loss": 1.0899,
      "step": 1040
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.98779064238356e-05,
      "loss": 1.102,
      "step": 1041
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.987764700043504e-05,
      "loss": 1.0414,
      "step": 1042
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.987738730239269e-05,
      "loss": 1.0168,
      "step": 1043
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9877127329711374e-05,
      "loss": 1.0868,
      "step": 1044
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9876867082393964e-05,
      "loss": 1.1141,
      "step": 1045
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9876606560443354e-05,
      "loss": 1.172,
      "step": 1046
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.987634576386241e-05,
      "loss": 1.4333,
      "step": 1047
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9876084692654e-05,
      "loss": 0.9353,
      "step": 1048
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9875823346821026e-05,
      "loss": 1.07,
      "step": 1049
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987556172636636e-05,
      "loss": 0.987,
      "step": 1050
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98752998312929e-05,
      "loss": 0.9829,
      "step": 1051
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987503766160353e-05,
      "loss": 0.9056,
      "step": 1052
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9874775217301136e-05,
      "loss": 1.4704,
      "step": 1053
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9874512498388634e-05,
      "loss": 1.3798,
      "step": 1054
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987424950486892e-05,
      "loss": 1.2657,
      "step": 1055
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987398623674489e-05,
      "loss": 1.2638,
      "step": 1056
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9873722694019444e-05,
      "loss": 1.126,
      "step": 1057
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9873458876695514e-05,
      "loss": 1.1468,
      "step": 1058
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9873194784775984e-05,
      "loss": 1.0899,
      "step": 1059
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987293041826379e-05,
      "loss": 0.9826,
      "step": 1060
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987266577716185e-05,
      "loss": 0.9366,
      "step": 1061
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987240086147308e-05,
      "loss": 1.173,
      "step": 1062
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9872135671200395e-05,
      "loss": 0.9772,
      "step": 1063
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9871870206346744e-05,
      "loss": 1.0039,
      "step": 1064
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9871604466915036e-05,
      "loss": 1.1604,
      "step": 1065
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987133845290822e-05,
      "loss": 0.9109,
      "step": 1066
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987107216432924e-05,
      "loss": 1.0049,
      "step": 1067
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987080560118101e-05,
      "loss": 0.95,
      "step": 1068
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987053876346649e-05,
      "loss": 1.4039,
      "step": 1069
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987027165118861e-05,
      "loss": 1.1398,
      "step": 1070
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.987000426435035e-05,
      "loss": 1.0097,
      "step": 1071
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986973660295464e-05,
      "loss": 0.9788,
      "step": 1072
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986946866700443e-05,
      "loss": 1.0314,
      "step": 1073
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986920045650269e-05,
      "loss": 0.8075,
      "step": 1074
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986893197145237e-05,
      "loss": 0.9763,
      "step": 1075
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986866321185645e-05,
      "loss": 0.99,
      "step": 1076
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986839417771788e-05,
      "loss": 1.0429,
      "step": 1077
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9868124869039646e-05,
      "loss": 1.1003,
      "step": 1078
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98678552858247e-05,
      "loss": 1.0552,
      "step": 1079
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986758542807605e-05,
      "loss": 0.9183,
      "step": 1080
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986731529579664e-05,
      "loss": 1.0602,
      "step": 1081
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986704488898948e-05,
      "loss": 0.9084,
      "step": 1082
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986677420765753e-05,
      "loss": 0.9231,
      "step": 1083
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9866503251803796e-05,
      "loss": 1.0418,
      "step": 1084
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986623202143127e-05,
      "loss": 0.9961,
      "step": 1085
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986596051654294e-05,
      "loss": 0.9858,
      "step": 1086
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98656887371418e-05,
      "loss": 0.8057,
      "step": 1087
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986541668323086e-05,
      "loss": 0.9343,
      "step": 1088
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986514435481312e-05,
      "loss": 1.0799,
      "step": 1089
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986487175189158e-05,
      "loss": 1.0532,
      "step": 1090
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986459887446926e-05,
      "loss": 0.9893,
      "step": 1091
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986432572254916e-05,
      "loss": 1.0342,
      "step": 1092
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9864052296134304e-05,
      "loss": 1.11,
      "step": 1093
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986377859522771e-05,
      "loss": 0.9471,
      "step": 1094
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9863504619832404e-05,
      "loss": 0.9256,
      "step": 1095
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98632303699514e-05,
      "loss": 0.8863,
      "step": 1096
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986295584558772e-05,
      "loss": 0.9684,
      "step": 1097
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986268104674442e-05,
      "loss": 1.3984,
      "step": 1098
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986240597342451e-05,
      "loss": 5.3965,
      "step": 1099
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9862130625631044e-05,
      "loss": 5.3798,
      "step": 1100
    },
    {
      "epoch": 0.04,
      "eval_loss": 0.9719640612602234,
      "eval_runtime": 373.683,
      "eval_samples_per_second": 5.756,
      "eval_steps_per_second": 1.44,
      "step": 1100
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986185500336705e-05,
      "loss": 3.5353,
      "step": 1101
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986157910663557e-05,
      "loss": 1.0627,
      "step": 1102
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986130293543966e-05,
      "loss": 0.8896,
      "step": 1103
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986102648978236e-05,
      "loss": 0.8021,
      "step": 1104
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986074976966673e-05,
      "loss": 0.7438,
      "step": 1105
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.986047277509581e-05,
      "loss": 1.0217,
      "step": 1106
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9860195506072676e-05,
      "loss": 1.2293,
      "step": 1107
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9859917962600374e-05,
      "loss": 1.0878,
      "step": 1108
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9859640144681985e-05,
      "loss": 0.8052,
      "step": 1109
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9859362052320556e-05,
      "loss": 0.9587,
      "step": 1110
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985908368551917e-05,
      "loss": 1.0993,
      "step": 1111
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9858805044280895e-05,
      "loss": 1.0229,
      "step": 1112
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9858526128608815e-05,
      "loss": 0.9004,
      "step": 1113
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9858246938505996e-05,
      "loss": 0.7753,
      "step": 1114
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985796747397553e-05,
      "loss": 1.0487,
      "step": 1115
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98576877350205e-05,
      "loss": 0.983,
      "step": 1116
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985740772164399e-05,
      "loss": 0.9663,
      "step": 1117
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98571274338491e-05,
      "loss": 0.921,
      "step": 1118
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9856846871638915e-05,
      "loss": 0.9105,
      "step": 1119
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985656603501654e-05,
      "loss": 1.0629,
      "step": 1120
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985628492398507e-05,
      "loss": 0.8816,
      "step": 1121
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9856003538547614e-05,
      "loss": 0.9998,
      "step": 1122
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985572187870727e-05,
      "loss": 1.0261,
      "step": 1123
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985543994446715e-05,
      "loss": 0.9282,
      "step": 1124
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9855157735830375e-05,
      "loss": 1.0524,
      "step": 1125
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985487525280005e-05,
      "loss": 1.3857,
      "step": 1126
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9854592495379305e-05,
      "loss": 1.1359,
      "step": 1127
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9854309463571245e-05,
      "loss": 1.0022,
      "step": 1128
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9854026157379005e-05,
      "loss": 1.0458,
      "step": 1129
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9853742576805715e-05,
      "loss": 0.9502,
      "step": 1130
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98534587218545e-05,
      "loss": 1.0168,
      "step": 1131
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9853174592528496e-05,
      "loss": 1.0601,
      "step": 1132
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9852890188830835e-05,
      "loss": 1.0097,
      "step": 1133
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9852605510764673e-05,
      "loss": 1.0672,
      "step": 1134
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985232055833313e-05,
      "loss": 1.0086,
      "step": 1135
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9852035331539374e-05,
      "loss": 1.0098,
      "step": 1136
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985174983038653e-05,
      "loss": 1.1098,
      "step": 1137
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9851464054877764e-05,
      "loss": 1.0769,
      "step": 1138
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985117800501624e-05,
      "loss": 0.9359,
      "step": 1139
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985089168080509e-05,
      "loss": 1.0192,
      "step": 1140
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98506050822475e-05,
      "loss": 0.9949,
      "step": 1141
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985031820934663e-05,
      "loss": 1.0946,
      "step": 1142
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.985003106210563e-05,
      "loss": 1.0889,
      "step": 1143
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984974364052768e-05,
      "loss": 0.9721,
      "step": 1144
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9849455944615955e-05,
      "loss": 1.1035,
      "step": 1145
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984916797437363e-05,
      "loss": 0.9047,
      "step": 1146
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984887972980389e-05,
      "loss": 1.1304,
      "step": 1147
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9848591210909914e-05,
      "loss": 5.3089,
      "step": 1148
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984830241769488e-05,
      "loss": 5.5864,
      "step": 1149
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984801335016198e-05,
      "loss": 0.8834,
      "step": 1150
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984772400831441e-05,
      "loss": 0.8813,
      "step": 1151
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9847434392155356e-05,
      "loss": 0.825,
      "step": 1152
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984714450168801e-05,
      "loss": 1.0825,
      "step": 1153
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98468543369156e-05,
      "loss": 0.9352,
      "step": 1154
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98465638978413e-05,
      "loss": 0.9671,
      "step": 1155
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9846273184468336e-05,
      "loss": 0.9595,
      "step": 1156
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98459821967999e-05,
      "loss": 1.622,
      "step": 1157
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984569093483922e-05,
      "loss": 3.1409,
      "step": 1158
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98453993985895e-05,
      "loss": 3.1059,
      "step": 1159
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9845107588053966e-05,
      "loss": 3.1042,
      "step": 1160
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9844815503235835e-05,
      "loss": 3.0861,
      "step": 1161
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984452314413833e-05,
      "loss": 3.0862,
      "step": 1162
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984423051076469e-05,
      "loss": 3.0704,
      "step": 1163
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9843937603118135e-05,
      "loss": 3.0699,
      "step": 1164
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98436444212019e-05,
      "loss": 3.0665,
      "step": 1165
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984335096501922e-05,
      "loss": 3.0623,
      "step": 1166
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984305723457334e-05,
      "loss": 3.0588,
      "step": 1167
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98427632298675e-05,
      "loss": 3.0528,
      "step": 1168
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9842468950904945e-05,
      "loss": 1.6186,
      "step": 1169
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9842174397688925e-05,
      "loss": 1.1385,
      "step": 1170
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984187957022269e-05,
      "loss": 0.9388,
      "step": 1171
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98415844685095e-05,
      "loss": 1.1529,
      "step": 1172
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98412890925526e-05,
      "loss": 1.1394,
      "step": 1173
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984099344235527e-05,
      "loss": 0.9572,
      "step": 1174
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9840697517920766e-05,
      "loss": 0.9929,
      "step": 1175
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9840401319252336e-05,
      "loss": 1.0256,
      "step": 1176
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.984010484635328e-05,
      "loss": 0.9115,
      "step": 1177
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983980809922686e-05,
      "loss": 1.0245,
      "step": 1178
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9839511077876344e-05,
      "loss": 1.3408,
      "step": 1179
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983921378230502e-05,
      "loss": 0.7245,
      "step": 1180
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983891621251616e-05,
      "loss": 0.4785,
      "step": 1181
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983861836851306e-05,
      "loss": 1.2116,
      "step": 1182
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9838320250298996e-05,
      "loss": 1.1713,
      "step": 1183
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9838021857877283e-05,
      "loss": 1.0022,
      "step": 1184
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9837723191251185e-05,
      "loss": 1.0523,
      "step": 1185
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9837424250424014e-05,
      "loss": 1.0957,
      "step": 1186
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983712503539908e-05,
      "loss": 0.9155,
      "step": 1187
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9836825546179664e-05,
      "loss": 0.9837,
      "step": 1188
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98365257827691e-05,
      "loss": 1.0753,
      "step": 1189
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983622574517066e-05,
      "loss": 0.9765,
      "step": 1190
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983592543338769e-05,
      "loss": 1.0414,
      "step": 1191
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983562484742349e-05,
      "loss": 0.9707,
      "step": 1192
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983532398728139e-05,
      "loss": 1.0076,
      "step": 1193
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983502285296469e-05,
      "loss": 0.9692,
      "step": 1194
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9834721444476734e-05,
      "loss": 1.0429,
      "step": 1195
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983441976182084e-05,
      "loss": 0.8168,
      "step": 1196
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9834117805000344e-05,
      "loss": 0.9048,
      "step": 1197
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983381557401858e-05,
      "loss": 1.0388,
      "step": 1198
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9833513068878875e-05,
      "loss": 1.0072,
      "step": 1199
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9833210289584576e-05,
      "loss": 0.9587,
      "step": 1200
    },
    {
      "epoch": 0.04,
      "eval_loss": 0.960885763168335,
      "eval_runtime": 373.4131,
      "eval_samples_per_second": 5.76,
      "eval_steps_per_second": 1.441,
      "step": 1200
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9832907236139024e-05,
      "loss": 0.9822,
      "step": 1201
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983260390854557e-05,
      "loss": 0.8111,
      "step": 1202
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983230030680755e-05,
      "loss": 1.0734,
      "step": 1203
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9831996430928326e-05,
      "loss": 1.0698,
      "step": 1204
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983169228091125e-05,
      "loss": 1.0753,
      "step": 1205
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983138785675968e-05,
      "loss": 0.935,
      "step": 1206
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9831083158476976e-05,
      "loss": 1.1086,
      "step": 1207
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9830778186066505e-05,
      "loss": 1.1362,
      "step": 1208
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983047293953163e-05,
      "loss": 0.9187,
      "step": 1209
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.983016741887572e-05,
      "loss": 0.8926,
      "step": 1210
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982986162410216e-05,
      "loss": 0.8638,
      "step": 1211
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98295555552143e-05,
      "loss": 0.9501,
      "step": 1212
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982924921221555e-05,
      "loss": 1.7165,
      "step": 1213
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9828942595109263e-05,
      "loss": 0.6233,
      "step": 1214
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982863570389884e-05,
      "loss": 0.8953,
      "step": 1215
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982832853858767e-05,
      "loss": 1.0904,
      "step": 1216
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982802109917914e-05,
      "loss": 0.7956,
      "step": 1217
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9827713385676645e-05,
      "loss": 0.546,
      "step": 1218
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982740539808358e-05,
      "loss": 0.4636,
      "step": 1219
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982709713640335e-05,
      "loss": 0.4982,
      "step": 1220
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982678860063935e-05,
      "loss": 0.4477,
      "step": 1221
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982647979079499e-05,
      "loss": 0.459,
      "step": 1222
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9826170706873686e-05,
      "loss": 0.4419,
      "step": 1223
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9825861348878844e-05,
      "loss": 0.4679,
      "step": 1224
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9825551716813875e-05,
      "loss": 0.4099,
      "step": 1225
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98252418106822e-05,
      "loss": 0.7367,
      "step": 1226
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982493163048725e-05,
      "loss": 0.9091,
      "step": 1227
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9824621176232436e-05,
      "loss": 1.0747,
      "step": 1228
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982431044792119e-05,
      "loss": 1.0403,
      "step": 1229
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982399944555695e-05,
      "loss": 1.5088,
      "step": 1230
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982368816914314e-05,
      "loss": 1.6626,
      "step": 1231
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982337661868319e-05,
      "loss": 1.212,
      "step": 1232
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9823064794180553e-05,
      "loss": 1.0315,
      "step": 1233
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9822752695638665e-05,
      "loss": 0.8981,
      "step": 1234
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982244032306098e-05,
      "loss": 0.9613,
      "step": 1235
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982212767645093e-05,
      "loss": 1.1094,
      "step": 1236
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982181475581198e-05,
      "loss": 1.0018,
      "step": 1237
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.982150156114759e-05,
      "loss": 1.1949,
      "step": 1238
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98211880924612e-05,
      "loss": 1.2943,
      "step": 1239
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9820874349756274e-05,
      "loss": 0.9217,
      "step": 1240
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9820560333036285e-05,
      "loss": 0.9649,
      "step": 1241
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9820246042304695e-05,
      "loss": 1.0746,
      "step": 1242
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981993147756498e-05,
      "loss": 0.893,
      "step": 1243
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98196166388206e-05,
      "loss": 0.7825,
      "step": 1244
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9819301526075035e-05,
      "loss": 0.6444,
      "step": 1245
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981898613933177e-05,
      "loss": 1.0549,
      "step": 1246
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9818670478594285e-05,
      "loss": 1.1322,
      "step": 1247
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9818354543866064e-05,
      "loss": 1.0649,
      "step": 1248
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981803833515059e-05,
      "loss": 0.983,
      "step": 1249
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981772185245135e-05,
      "loss": 0.6899,
      "step": 1250
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981740509577185e-05,
      "loss": 0.6298,
      "step": 1251
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981708806511559e-05,
      "loss": 1.1007,
      "step": 1252
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981677076048605e-05,
      "loss": 0.9672,
      "step": 1253
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981645318188676e-05,
      "loss": 0.9566,
      "step": 1254
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9816135329321194e-05,
      "loss": 0.3964,
      "step": 1255
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9815817202792886e-05,
      "loss": 0.3656,
      "step": 1256
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981549880230534e-05,
      "loss": 0.9123,
      "step": 1257
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981518012786207e-05,
      "loss": 0.998,
      "step": 1258
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98148611794666e-05,
      "loss": 1.2456,
      "step": 1259
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981454195712244e-05,
      "loss": 1.6273,
      "step": 1260
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9814222460833115e-05,
      "loss": 1.526,
      "step": 1261
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9813902690602164e-05,
      "loss": 1.4927,
      "step": 1262
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9813582646433114e-05,
      "loss": 1.6195,
      "step": 1263
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981326232832949e-05,
      "loss": 1.5887,
      "step": 1264
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981294173629484e-05,
      "loss": 1.6555,
      "step": 1265
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981262087033268e-05,
      "loss": 1.6323,
      "step": 1266
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981229973044659e-05,
      "loss": 1.6385,
      "step": 1267
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9811978316640076e-05,
      "loss": 1.6356,
      "step": 1268
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9811656628916714e-05,
      "loss": 1.6267,
      "step": 1269
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981133466728004e-05,
      "loss": 1.6412,
      "step": 1270
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9811012431733616e-05,
      "loss": 1.6205,
      "step": 1271
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981068992228099e-05,
      "loss": 1.2787,
      "step": 1272
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9810367138925746e-05,
      "loss": 1.1116,
      "step": 1273
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.981004408167142e-05,
      "loss": 1.0445,
      "step": 1274
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980972075052159e-05,
      "loss": 0.9609,
      "step": 1275
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980939714547983e-05,
      "loss": 1.0706,
      "step": 1276
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98090732665497e-05,
      "loss": 1.02,
      "step": 1277
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980874911373479e-05,
      "loss": 0.9308,
      "step": 1278
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9808424687038664e-05,
      "loss": 0.9699,
      "step": 1279
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980809998646492e-05,
      "loss": 0.8678,
      "step": 1280
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980777501201712e-05,
      "loss": 1.063,
      "step": 1281
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980744976369888e-05,
      "loss": 1.0023,
      "step": 1282
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980712424151377e-05,
      "loss": 0.7438,
      "step": 1283
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980679844546539e-05,
      "loss": 1.0074,
      "step": 1284
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9806472375557334e-05,
      "loss": 1.1127,
      "step": 1285
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98061460317932e-05,
      "loss": 1.0486,
      "step": 1286
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.98058194141766e-05,
      "loss": 0.9734,
      "step": 1287
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980549252271114e-05,
      "loss": 0.8879,
      "step": 1288
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9805165357400415e-05,
      "loss": 1.1676,
      "step": 1289
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980483791824805e-05,
      "loss": 0.9939,
      "step": 1290
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9804510205257647e-05,
      "loss": 1.0102,
      "step": 1291
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980418221843284e-05,
      "loss": 1.0471,
      "step": 1292
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9803853957777244e-05,
      "loss": 1.0721,
      "step": 1293
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980352542329447e-05,
      "loss": 0.909,
      "step": 1294
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980319661498817e-05,
      "loss": 0.8067,
      "step": 1295
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980286753286195e-05,
      "loss": 0.9173,
      "step": 1296
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9802538176919453e-05,
      "loss": 1.0631,
      "step": 1297
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9802208547164315e-05,
      "loss": 0.9453,
      "step": 1298
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980187864360017e-05,
      "loss": 0.2468,
      "step": 1299
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980154846623068e-05,
      "loss": 0.8916,
      "step": 1300
    },
    {
      "epoch": 0.04,
      "eval_loss": 0.9709475040435791,
      "eval_runtime": 388.6114,
      "eval_samples_per_second": 5.535,
      "eval_steps_per_second": 1.384,
      "step": 1300
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9801218015059456e-05,
      "loss": 1.1455,
      "step": 1301
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980088729009017e-05,
      "loss": 0.9706,
      "step": 1302
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.980055629132647e-05,
      "loss": 0.7933,
      "step": 1303
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9800225018772005e-05,
      "loss": 0.9857,
      "step": 1304
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9799893472430436e-05,
      "loss": 1.0344,
      "step": 1305
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979956165230543e-05,
      "loss": 1.0649,
      "step": 1306
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979922955840063e-05,
      "loss": 0.9724,
      "step": 1307
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979889719071972e-05,
      "loss": 1.0094,
      "step": 1308
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979856454926637e-05,
      "loss": 1.0704,
      "step": 1309
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9798231634044234e-05,
      "loss": 0.9043,
      "step": 1310
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979789844505701e-05,
      "loss": 1.2134,
      "step": 1311
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9797564982308365e-05,
      "loss": 1.8016,
      "step": 1312
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979723124580197e-05,
      "loss": 1.3687,
      "step": 1313
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979689723554153e-05,
      "loss": 1.1302,
      "step": 1314
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9796562951530724e-05,
      "loss": 1.0531,
      "step": 1315
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9796228393773236e-05,
      "loss": 0.7541,
      "step": 1316
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9795893562272765e-05,
      "loss": 0.9844,
      "step": 1317
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979555845703301e-05,
      "loss": 0.7212,
      "step": 1318
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979522307805766e-05,
      "loss": 0.5992,
      "step": 1319
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979488742535043e-05,
      "loss": 0.9858,
      "step": 1320
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979455149891502e-05,
      "loss": 0.8671,
      "step": 1321
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979421529875514e-05,
      "loss": 0.9352,
      "step": 1322
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.97938788248745e-05,
      "loss": 0.8404,
      "step": 1323
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979354207727682e-05,
      "loss": 0.6903,
      "step": 1324
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979320505596581e-05,
      "loss": 1.0803,
      "step": 1325
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9792867760945186e-05,
      "loss": 0.7976,
      "step": 1326
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979253019221868e-05,
      "loss": 0.9382,
      "step": 1327
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979219234979002e-05,
      "loss": 1.068,
      "step": 1328
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979185423366294e-05,
      "loss": 1.1783,
      "step": 1329
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9791515843841165e-05,
      "loss": 1.0501,
      "step": 1330
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979117718032842e-05,
      "loss": 1.0846,
      "step": 1331
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979083824312847e-05,
      "loss": 1.1015,
      "step": 1332
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979049903224503e-05,
      "loss": 1.061,
      "step": 1333
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979015954768186e-05,
      "loss": 0.8835,
      "step": 1334
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.978981978944271e-05,
      "loss": 1.1084,
      "step": 1335
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9789479757531324e-05,
      "loss": 1.4507,
      "step": 1336
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.978913945195145e-05,
      "loss": 0.9376,
      "step": 1337
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.978879887270686e-05,
      "loss": 1.0241,
      "step": 1338
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9788458019801304e-05,
      "loss": 1.0186,
      "step": 1339
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.978811689323854e-05,
      "loss": 1.0776,
      "step": 1340
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.978777549302236e-05,
      "loss": 1.0953,
      "step": 1341
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.97874338191565e-05,
      "loss": 1.188,
      "step": 1342
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.978709187164475e-05,
      "loss": 1.0942,
      "step": 1343
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9786749650490874e-05,
      "loss": 0.9831,
      "step": 1344
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.978640715569867e-05,
      "loss": 1.0115,
      "step": 1345
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.978606438727189e-05,
      "loss": 1.4371,
      "step": 1346
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.978572134521435e-05,
      "loss": 0.9423,
      "step": 1347
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.978537802952981e-05,
      "loss": 0.9665,
      "step": 1348
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9785034440222076e-05,
      "loss": 0.9124,
      "step": 1349
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9784690577294936e-05,
      "loss": 1.0346,
      "step": 1350
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.978434644075218e-05,
      "loss": 1.0874,
      "step": 1351
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9784002030597624e-05,
      "loss": 0.9703,
      "step": 1352
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.978365734683505e-05,
      "loss": 1.01,
      "step": 1353
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9783312389468285e-05,
      "loss": 0.9495,
      "step": 1354
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9782967158501126e-05,
      "loss": 1.0022,
      "step": 1355
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.978262165393737e-05,
      "loss": 1.1435,
      "step": 1356
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.978227587578086e-05,
      "loss": 1.0142,
      "step": 1357
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.978192982403539e-05,
      "loss": 1.1047,
      "step": 1358
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.978158349870479e-05,
      "loss": 0.8851,
      "step": 1359
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.978123689979287e-05,
      "loss": 0.9842,
      "step": 1360
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9780890027303486e-05,
      "loss": 1.0318,
      "step": 1361
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.978054288124044e-05,
      "loss": 1.0066,
      "step": 1362
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9780195461607574e-05,
      "loss": 0.8913,
      "step": 1363
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977984776840873e-05,
      "loss": 1.0903,
      "step": 1364
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977949980164773e-05,
      "loss": 1.1022,
      "step": 1365
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977915156132843e-05,
      "loss": 0.9392,
      "step": 1366
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977880304745467e-05,
      "loss": 1.0969,
      "step": 1367
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97784542600303e-05,
      "loss": 0.8413,
      "step": 1368
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9778105199059164e-05,
      "loss": 0.924,
      "step": 1369
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9777755864545116e-05,
      "loss": 0.9146,
      "step": 1370
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9777406256492024e-05,
      "loss": 0.9359,
      "step": 1371
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977705637490373e-05,
      "loss": 1.0429,
      "step": 1372
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9776706219784115e-05,
      "loss": 1.0387,
      "step": 1373
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977635579113703e-05,
      "loss": 1.022,
      "step": 1374
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977600508896635e-05,
      "loss": 1.034,
      "step": 1375
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977565411327595e-05,
      "loss": 0.9886,
      "step": 1376
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97753028640697e-05,
      "loss": 0.9831,
      "step": 1377
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9774951341351475e-05,
      "loss": 0.9397,
      "step": 1378
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9774599545125157e-05,
      "loss": 0.9957,
      "step": 1379
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977424747539463e-05,
      "loss": 1.1078,
      "step": 1380
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9773895132163785e-05,
      "loss": 0.9999,
      "step": 1381
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977354251543651e-05,
      "loss": 1.0317,
      "step": 1382
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97731896252167e-05,
      "loss": 0.919,
      "step": 1383
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977283646150824e-05,
      "loss": 1.0567,
      "step": 1384
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977248302431504e-05,
      "loss": 0.7626,
      "step": 1385
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9772129313641e-05,
      "loss": 1.019,
      "step": 1386
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9771775329490024e-05,
      "loss": 1.0202,
      "step": 1387
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977142107186602e-05,
      "loss": 1.1994,
      "step": 1388
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9771066540772893e-05,
      "loss": 1.0147,
      "step": 1389
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977071173621457e-05,
      "loss": 1.0682,
      "step": 1390
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.977035665819495e-05,
      "loss": 1.0545,
      "step": 1391
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9770001306717964e-05,
      "loss": 1.1179,
      "step": 1392
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9769645681787534e-05,
      "loss": 0.9332,
      "step": 1393
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9769289783407594e-05,
      "loss": 1.0748,
      "step": 1394
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976893361158206e-05,
      "loss": 1.1557,
      "step": 1395
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9768577166314865e-05,
      "loss": 0.981,
      "step": 1396
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976822044760995e-05,
      "loss": 0.9584,
      "step": 1397
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976786345547125e-05,
      "loss": 1.0634,
      "step": 1398
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976750618990271e-05,
      "loss": 0.9605,
      "step": 1399
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976714865090827e-05,
      "loss": 1.0955,
      "step": 1400
    },
    {
      "epoch": 0.05,
      "eval_loss": 0.9706699252128601,
      "eval_runtime": 277.7627,
      "eval_samples_per_second": 7.744,
      "eval_steps_per_second": 1.937,
      "step": 1400
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976679083849188e-05,
      "loss": 1.0327,
      "step": 1401
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976643275265749e-05,
      "loss": 0.5041,
      "step": 1402
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976607439340905e-05,
      "loss": 0.1208,
      "step": 1403
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976571576075051e-05,
      "loss": 0.7269,
      "step": 1404
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976535685468584e-05,
      "loss": 0.9696,
      "step": 1405
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9764997675219e-05,
      "loss": 1.1368,
      "step": 1406
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976463822235396e-05,
      "loss": 1.1095,
      "step": 1407
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976427849609468e-05,
      "loss": 0.9157,
      "step": 1408
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9763918496445125e-05,
      "loss": 0.961,
      "step": 1409
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976355822340929e-05,
      "loss": 0.9396,
      "step": 1410
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9763197676991134e-05,
      "loss": 0.9132,
      "step": 1411
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976283685719465e-05,
      "loss": 1.0231,
      "step": 1412
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976247576402381e-05,
      "loss": 1.0287,
      "step": 1413
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9762114397482605e-05,
      "loss": 0.774,
      "step": 1414
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976175275757503e-05,
      "loss": 0.9443,
      "step": 1415
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976139084430507e-05,
      "loss": 1.0228,
      "step": 1416
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.976102865767672e-05,
      "loss": 2.0289,
      "step": 1417
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9760666197693986e-05,
      "loss": 1.3567,
      "step": 1418
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9760303464360856e-05,
      "loss": 0.993,
      "step": 1419
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975994045768135e-05,
      "loss": 0.9866,
      "step": 1420
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975957717765947e-05,
      "loss": 0.8149,
      "step": 1421
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9759213624299215e-05,
      "loss": 1.0718,
      "step": 1422
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975884979760462e-05,
      "loss": 1.0381,
      "step": 1423
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975848569757968e-05,
      "loss": 0.9972,
      "step": 1424
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9758121324228435e-05,
      "loss": 0.9215,
      "step": 1425
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975775667755489e-05,
      "loss": 0.9134,
      "step": 1426
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975739175756308e-05,
      "loss": 1.0004,
      "step": 1427
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975702656425704e-05,
      "loss": 1.0012,
      "step": 1428
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975666109764078e-05,
      "loss": 1.0799,
      "step": 1429
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9756295357718354e-05,
      "loss": 0.9739,
      "step": 1430
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975592934449379e-05,
      "loss": 1.1545,
      "step": 1431
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975556305797113e-05,
      "loss": 1.0423,
      "step": 1432
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9755196498154425e-05,
      "loss": 0.7184,
      "step": 1433
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975482966504772e-05,
      "loss": 0.595,
      "step": 1434
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975446255865506e-05,
      "loss": 0.962,
      "step": 1435
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9754095178980486e-05,
      "loss": 1.1056,
      "step": 1436
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975372752602808e-05,
      "loss": 1.0772,
      "step": 1437
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9753359599801884e-05,
      "loss": 1.0101,
      "step": 1438
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975299140030596e-05,
      "loss": 0.9838,
      "step": 1439
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975262292754438e-05,
      "loss": 1.9783,
      "step": 1440
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975225418152121e-05,
      "loss": 1.3386,
      "step": 1441
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975188516224051e-05,
      "loss": 4.3328,
      "step": 1442
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9751515869706367e-05,
      "loss": 0.9601,
      "step": 1443
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975114630392286e-05,
      "loss": 1.0544,
      "step": 1444
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975077646489405e-05,
      "loss": 0.9658,
      "step": 1445
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.975040635262404e-05,
      "loss": 1.0182,
      "step": 1446
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97500359671169e-05,
      "loss": 1.011,
      "step": 1447
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9749665308376744e-05,
      "loss": 1.0953,
      "step": 1448
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974929437640764e-05,
      "loss": 1.1326,
      "step": 1449
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974892317121368e-05,
      "loss": 0.9989,
      "step": 1450
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974855169279898e-05,
      "loss": 1.0383,
      "step": 1451
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974817994116764e-05,
      "loss": 1.4369,
      "step": 1452
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974780791632375e-05,
      "loss": 1.1963,
      "step": 1453
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974743561827142e-05,
      "loss": 0.9523,
      "step": 1454
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974706304701476e-05,
      "loss": 0.3763,
      "step": 1455
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97466902025579e-05,
      "loss": 1.0655,
      "step": 1456
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9746317084904944e-05,
      "loss": 0.993,
      "step": 1457
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974594369406e-05,
      "loss": 0.989,
      "step": 1458
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974557003002721e-05,
      "loss": 1.0831,
      "step": 1459
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974519609281069e-05,
      "loss": 0.6683,
      "step": 1460
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974482188241456e-05,
      "loss": 1.0585,
      "step": 1461
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9744447398842966e-05,
      "loss": 1.0161,
      "step": 1462
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974407264210003e-05,
      "loss": 1.0817,
      "step": 1463
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97436976121899e-05,
      "loss": 1.0359,
      "step": 1464
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97433223091167e-05,
      "loss": 0.9982,
      "step": 1465
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97429467328846e-05,
      "loss": 1.0226,
      "step": 1466
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974257088349772e-05,
      "loss": 1.1008,
      "step": 1467
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9742194760960214e-05,
      "loss": 1.1163,
      "step": 1468
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974181836527625e-05,
      "loss": 0.6973,
      "step": 1469
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9741441696449974e-05,
      "loss": 1.1521,
      "step": 1470
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9741064754485537e-05,
      "loss": 0.9275,
      "step": 1471
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974068753938711e-05,
      "loss": 1.1368,
      "step": 1472
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.974031005115885e-05,
      "loss": 0.7919,
      "step": 1473
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973993228980494e-05,
      "loss": 1.0786,
      "step": 1474
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9739554255329525e-05,
      "loss": 1.0088,
      "step": 1475
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9739175947736794e-05,
      "loss": 1.0562,
      "step": 1476
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973879736703093e-05,
      "loss": 0.9582,
      "step": 1477
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97384185132161e-05,
      "loss": 1.0386,
      "step": 1478
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973803938629649e-05,
      "loss": 1.0136,
      "step": 1479
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973765998627628e-05,
      "loss": 0.9613,
      "step": 1480
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9737280313159676e-05,
      "loss": 1.0263,
      "step": 1481
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973690036695084e-05,
      "loss": 0.989,
      "step": 1482
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9736520147654e-05,
      "loss": 0.8783,
      "step": 1483
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9736139655273334e-05,
      "loss": 0.9968,
      "step": 1484
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973575888981304e-05,
      "loss": 0.9983,
      "step": 1485
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973537785127734e-05,
      "loss": 1.0272,
      "step": 1486
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973499653967042e-05,
      "loss": 0.9104,
      "step": 1487
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97346149549965e-05,
      "loss": 1.1625,
      "step": 1488
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973423309725979e-05,
      "loss": 1.0316,
      "step": 1489
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973385096646451e-05,
      "loss": 0.9642,
      "step": 1490
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973346856261487e-05,
      "loss": 0.9862,
      "step": 1491
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97330858857151e-05,
      "loss": 1.0428,
      "step": 1492
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973270293576942e-05,
      "loss": 1.1983,
      "step": 1493
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973231971278205e-05,
      "loss": 1.0869,
      "step": 1494
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973193621675724e-05,
      "loss": 1.077,
      "step": 1495
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973155244769921e-05,
      "loss": 1.3441,
      "step": 1496
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973116840561219e-05,
      "loss": 0.9228,
      "step": 1497
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.973078409050045e-05,
      "loss": 0.8932,
      "step": 1498
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97303995023682e-05,
      "loss": 1.0115,
      "step": 1499
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9730014641219694e-05,
      "loss": 1.041,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "eval_loss": 0.9728406071662903,
      "eval_runtime": 384.8964,
      "eval_samples_per_second": 5.589,
      "eval_steps_per_second": 1.398,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9729629507059196e-05,
      "loss": 1.0754,
      "step": 1501
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972924409989094e-05,
      "loss": 0.9507,
      "step": 1502
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972885841971918e-05,
      "loss": 0.9796,
      "step": 1503
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9728472466548194e-05,
      "loss": 1.0,
      "step": 1504
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972808624038222e-05,
      "loss": 0.9353,
      "step": 1505
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9727699741225534e-05,
      "loss": 1.2348,
      "step": 1506
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9727312969082406e-05,
      "loss": 0.9028,
      "step": 1507
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972692592395709e-05,
      "loss": 1.0407,
      "step": 1508
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972653860585388e-05,
      "loss": 0.9672,
      "step": 1509
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9726151014777036e-05,
      "loss": 1.308,
      "step": 1510
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9725763150730844e-05,
      "loss": 0.9447,
      "step": 1511
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9725375013719584e-05,
      "loss": 0.9216,
      "step": 1512
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972498660374754e-05,
      "loss": 1.1896,
      "step": 1513
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9724597920819e-05,
      "loss": 0.9683,
      "step": 1514
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9724208964938256e-05,
      "loss": 1.0122,
      "step": 1515
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97238197361096e-05,
      "loss": 1.0438,
      "step": 1516
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972343023433733e-05,
      "loss": 0.9651,
      "step": 1517
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9723040459625747e-05,
      "loss": 0.9372,
      "step": 1518
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9722650411979154e-05,
      "loss": 0.9692,
      "step": 1519
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972226009140186e-05,
      "loss": 1.0268,
      "step": 1520
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972186949789817e-05,
      "loss": 0.9772,
      "step": 1521
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972147863147239e-05,
      "loss": 0.9697,
      "step": 1522
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972108749212885e-05,
      "loss": 0.8654,
      "step": 1523
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9720696079871855e-05,
      "loss": 0.8747,
      "step": 1524
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.972030439470573e-05,
      "loss": 1.0657,
      "step": 1525
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9719912436634796e-05,
      "loss": 1.019,
      "step": 1526
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971952020566339e-05,
      "loss": 0.8703,
      "step": 1527
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9719127701795844e-05,
      "loss": 0.9719,
      "step": 1528
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971873492503647e-05,
      "loss": 1.9989,
      "step": 1529
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9718341875389625e-05,
      "loss": 2.7773,
      "step": 1530
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9717948552859636e-05,
      "loss": 1.0151,
      "step": 1531
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971755495745085e-05,
      "loss": 1.189,
      "step": 1532
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971716108916761e-05,
      "loss": 1.4368,
      "step": 1533
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971676694801427e-05,
      "loss": 0.9868,
      "step": 1534
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971637253399518e-05,
      "loss": 1.1043,
      "step": 1535
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971597784711469e-05,
      "loss": 0.7679,
      "step": 1536
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971558288737715e-05,
      "loss": 0.9899,
      "step": 1537
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971518765478694e-05,
      "loss": 3.4116,
      "step": 1538
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971479214934841e-05,
      "loss": 3.8675,
      "step": 1539
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971439637106592e-05,
      "loss": 1.2898,
      "step": 1540
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971400031994385e-05,
      "loss": 0.9012,
      "step": 1541
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971360399598658e-05,
      "loss": 0.711,
      "step": 1542
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971320739919847e-05,
      "loss": 0.9619,
      "step": 1543
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9712810529583894e-05,
      "loss": 0.9806,
      "step": 1544
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971241338714726e-05,
      "loss": 0.948,
      "step": 1545
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9712015971892924e-05,
      "loss": 1.0218,
      "step": 1546
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971161828382529e-05,
      "loss": 1.0762,
      "step": 1547
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971122032294874e-05,
      "loss": 1.1812,
      "step": 1548
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971082208926767e-05,
      "loss": 0.9539,
      "step": 1549
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971042358278648e-05,
      "loss": 1.0463,
      "step": 1550
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.971002480350956e-05,
      "loss": 1.027,
      "step": 1551
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9709625751441326e-05,
      "loss": 1.0016,
      "step": 1552
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970922642658616e-05,
      "loss": 0.8587,
      "step": 1553
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970882682894851e-05,
      "loss": 1.1557,
      "step": 1554
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970842695853275e-05,
      "loss": 1.1506,
      "step": 1555
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9708026815343314e-05,
      "loss": 1.0802,
      "step": 1556
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97076263993846e-05,
      "loss": 0.9325,
      "step": 1557
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9707225710661055e-05,
      "loss": 1.016,
      "step": 1558
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970682474917709e-05,
      "loss": 0.995,
      "step": 1559
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970642351493713e-05,
      "loss": 1.0325,
      "step": 1560
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.97060220079456e-05,
      "loss": 1.1201,
      "step": 1561
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970562022820694e-05,
      "loss": 1.1073,
      "step": 1562
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9705218175725586e-05,
      "loss": 1.1122,
      "step": 1563
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970481585050598e-05,
      "loss": 1.0197,
      "step": 1564
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970441325255254e-05,
      "loss": 1.0442,
      "step": 1565
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9704010381869745e-05,
      "loss": 1.0413,
      "step": 1566
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970360723846202e-05,
      "loss": 0.9988,
      "step": 1567
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970320382233382e-05,
      "loss": 0.5318,
      "step": 1568
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970280013348961e-05,
      "loss": 0.8365,
      "step": 1569
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970239617193383e-05,
      "loss": 0.5525,
      "step": 1570
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970199193767094e-05,
      "loss": 0.6358,
      "step": 1571
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9701587430705415e-05,
      "loss": 0.7772,
      "step": 1572
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9701182651041716e-05,
      "loss": 0.9843,
      "step": 1573
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.970077759868431e-05,
      "loss": 1.068,
      "step": 1574
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9700372273637666e-05,
      "loss": 1.0402,
      "step": 1575
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969996667590626e-05,
      "loss": 1.0503,
      "step": 1576
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969956080549457e-05,
      "loss": 0.9423,
      "step": 1577
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969915466240709e-05,
      "loss": 1.0438,
      "step": 1578
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969874824664828e-05,
      "loss": 0.8971,
      "step": 1579
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969834155822264e-05,
      "loss": 0.9941,
      "step": 1580
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9697934597134664e-05,
      "loss": 0.6356,
      "step": 1581
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9697527363388834e-05,
      "loss": 0.9707,
      "step": 1582
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969711985698965e-05,
      "loss": 0.9014,
      "step": 1583
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9696712077941615e-05,
      "loss": 1.0082,
      "step": 1584
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9696304026249224e-05,
      "loss": 0.7725,
      "step": 1585
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9695895701916985e-05,
      "loss": 0.5903,
      "step": 1586
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9695487104949405e-05,
      "loss": 1.046,
      "step": 1587
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9695078235351e-05,
      "loss": 1.057,
      "step": 1588
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969466909312627e-05,
      "loss": 1.0429,
      "step": 1589
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9694259678279756e-05,
      "loss": 1.0146,
      "step": 1590
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9693849990815946e-05,
      "loss": 1.076,
      "step": 1591
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969344003073939e-05,
      "loss": 1.0381,
      "step": 1592
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9693029798054595e-05,
      "loss": 0.9265,
      "step": 1593
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969261929276611e-05,
      "loss": 1.077,
      "step": 1594
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9692208514878444e-05,
      "loss": 0.9889,
      "step": 1595
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969179746439615e-05,
      "loss": 0.9326,
      "step": 1596
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969138614132376e-05,
      "loss": 1.2396,
      "step": 1597
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9690974545665805e-05,
      "loss": 1.2951,
      "step": 1598
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.969056267742684e-05,
      "loss": 1.2,
      "step": 1599
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9690150536611416e-05,
      "loss": 1.1362,
      "step": 1600
    },
    {
      "epoch": 0.05,
      "eval_loss": 0.9817221760749817,
      "eval_runtime": 254.343,
      "eval_samples_per_second": 8.457,
      "eval_steps_per_second": 2.115,
      "step": 1600
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968973812322407e-05,
      "loss": 0.9943,
      "step": 1601
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9689325437269375e-05,
      "loss": 0.8558,
      "step": 1602
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968891247875186e-05,
      "loss": 1.0321,
      "step": 1603
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.96884992476761e-05,
      "loss": 1.0551,
      "step": 1604
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968808574404665e-05,
      "loss": 1.0341,
      "step": 1605
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968767196786809e-05,
      "loss": 1.0794,
      "step": 1606
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968725791914497e-05,
      "loss": 1.0763,
      "step": 1607
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968684359788187e-05,
      "loss": 0.4145,
      "step": 1608
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9686429004083355e-05,
      "loss": 0.9713,
      "step": 1609
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968601413775402e-05,
      "loss": 1.0031,
      "step": 1610
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968559899889843e-05,
      "loss": 0.5285,
      "step": 1611
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9685183587521165e-05,
      "loss": 0.1614,
      "step": 1612
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968476790362683e-05,
      "loss": 0.7268,
      "step": 1613
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968435194722e-05,
      "loss": 1.002,
      "step": 1614
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968393571830526e-05,
      "loss": 1.0245,
      "step": 1615
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968351921688722e-05,
      "loss": 1.0337,
      "step": 1616
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968310244297047e-05,
      "loss": 1.032,
      "step": 1617
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9682685396559627e-05,
      "loss": 0.9639,
      "step": 1618
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968226807765926e-05,
      "loss": 1.049,
      "step": 1619
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968185048627402e-05,
      "loss": 1.1604,
      "step": 1620
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968143262240847e-05,
      "loss": 1.123,
      "step": 1621
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.968101448606727e-05,
      "loss": 1.2904,
      "step": 1622
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9680596077255004e-05,
      "loss": 1.2232,
      "step": 1623
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9680177395976305e-05,
      "loss": 1.1542,
      "step": 1624
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9679758442235776e-05,
      "loss": 1.125,
      "step": 1625
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9679339216038075e-05,
      "loss": 1.0809,
      "step": 1626
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967891971738781e-05,
      "loss": 1.0062,
      "step": 1627
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.96784999462896e-05,
      "loss": 1.0138,
      "step": 1628
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967807990274811e-05,
      "loss": 1.0553,
      "step": 1629
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967765958676795e-05,
      "loss": 0.9199,
      "step": 1630
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9677238998353774e-05,
      "loss": 1.0964,
      "step": 1631
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967681813751022e-05,
      "loss": 1.0597,
      "step": 1632
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9676397004241944e-05,
      "loss": 1.0531,
      "step": 1633
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9675975598553584e-05,
      "loss": 1.2406,
      "step": 1634
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967555392044979e-05,
      "loss": 1.0199,
      "step": 1635
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967513196993523e-05,
      "loss": 0.9632,
      "step": 1636
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967470974701455e-05,
      "loss": 1.4493,
      "step": 1637
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967428725169242e-05,
      "loss": 2.087,
      "step": 1638
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.96738644839735e-05,
      "loss": 0.9708,
      "step": 1639
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967344144386246e-05,
      "loss": 1.1024,
      "step": 1640
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967301813136396e-05,
      "loss": 1.0238,
      "step": 1641
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9672594546482686e-05,
      "loss": 1.0597,
      "step": 1642
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967217068922331e-05,
      "loss": 0.9847,
      "step": 1643
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967174655959051e-05,
      "loss": 1.0266,
      "step": 1644
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967132215758896e-05,
      "loss": 0.9769,
      "step": 1645
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967089748322337e-05,
      "loss": 0.9521,
      "step": 1646
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9670472536498405e-05,
      "loss": 0.9573,
      "step": 1647
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.967004731741876e-05,
      "loss": 1.0246,
      "step": 1648
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.966962182598914e-05,
      "loss": 1.0363,
      "step": 1649
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966919606221423e-05,
      "loss": 0.951,
      "step": 1650
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966877002609874e-05,
      "loss": 0.7816,
      "step": 1651
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966834371764736e-05,
      "loss": 0.7541,
      "step": 1652
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9667917136864816e-05,
      "loss": 0.935,
      "step": 1653
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9667490283755805e-05,
      "loss": 0.443,
      "step": 1654
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966706315832504e-05,
      "loss": 0.3529,
      "step": 1655
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966663576057723e-05,
      "loss": 0.2889,
      "step": 1656
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966620809051711e-05,
      "loss": 0.5268,
      "step": 1657
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966578014814939e-05,
      "loss": 1.0737,
      "step": 1658
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966535193347879e-05,
      "loss": 1.0187,
      "step": 1659
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966492344651005e-05,
      "loss": 8.0024,
      "step": 1660
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9664494687247895e-05,
      "loss": 6.7479,
      "step": 1661
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9664065655697056e-05,
      "loss": 8.4277,
      "step": 1662
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966363635186227e-05,
      "loss": 6.7526,
      "step": 1663
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966320677574827e-05,
      "loss": 1.9002,
      "step": 1664
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966277692735982e-05,
      "loss": 1.0552,
      "step": 1665
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966234680670164e-05,
      "loss": 1.126,
      "step": 1666
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.96619164137785e-05,
      "loss": 0.9823,
      "step": 1667
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9661485748595126e-05,
      "loss": 1.4444,
      "step": 1668
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9661054811156296e-05,
      "loss": 0.8386,
      "step": 1669
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966062360146676e-05,
      "loss": 0.9784,
      "step": 1670
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.966019211953128e-05,
      "loss": 1.0769,
      "step": 1671
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965976036535461e-05,
      "loss": 1.1803,
      "step": 1672
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9659328338941525e-05,
      "loss": 1.0157,
      "step": 1673
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965889604029679e-05,
      "loss": 1.0441,
      "step": 1674
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965846346942518e-05,
      "loss": 0.9194,
      "step": 1675
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965803062633147e-05,
      "loss": 0.9906,
      "step": 1676
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965759751102044e-05,
      "loss": 0.9547,
      "step": 1677
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965716412349687e-05,
      "loss": 1.0579,
      "step": 1678
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965673046376555e-05,
      "loss": 1.014,
      "step": 1679
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9656296531831256e-05,
      "loss": 0.9563,
      "step": 1680
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965586232769879e-05,
      "loss": 1.0234,
      "step": 1681
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965542785137294e-05,
      "loss": 0.7083,
      "step": 1682
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9654993102858496e-05,
      "loss": 0.6686,
      "step": 1683
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965455808216026e-05,
      "loss": 1.3832,
      "step": 1684
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965412278928305e-05,
      "loss": 0.5831,
      "step": 1685
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965368722423166e-05,
      "loss": 0.8385,
      "step": 1686
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965325138701089e-05,
      "loss": 0.9857,
      "step": 1687
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9652815277625566e-05,
      "loss": 1.0601,
      "step": 1688
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.96523788960805e-05,
      "loss": 1.0189,
      "step": 1689
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9651942242380504e-05,
      "loss": 1.3857,
      "step": 1690
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965150531653039e-05,
      "loss": 1.3856,
      "step": 1691
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9651068118535006e-05,
      "loss": 0.9788,
      "step": 1692
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965063064839915e-05,
      "loss": 1.0745,
      "step": 1693
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.965019290612768e-05,
      "loss": 0.8555,
      "step": 1694
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964975489172541e-05,
      "loss": 0.8601,
      "step": 1695
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964931660519718e-05,
      "loss": 1.0048,
      "step": 1696
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964887804654783e-05,
      "loss": 0.8365,
      "step": 1697
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.96484392157822e-05,
      "loss": 0.9401,
      "step": 1698
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964800011290514e-05,
      "loss": 0.912,
      "step": 1699
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964756073792148e-05,
      "loss": 0.9898,
      "step": 1700
    },
    {
      "epoch": 0.06,
      "eval_loss": 0.969132125377655,
      "eval_runtime": 338.7833,
      "eval_samples_per_second": 6.349,
      "eval_steps_per_second": 1.588,
      "step": 1700
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.96471210908361e-05,
      "loss": 1.0333,
      "step": 1701
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9646681171653824e-05,
      "loss": 1.0074,
      "step": 1702
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964624098037953e-05,
      "loss": 1.0056,
      "step": 1703
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964580051701806e-05,
      "loss": 1.0389,
      "step": 1704
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9645359781574294e-05,
      "loss": 0.8308,
      "step": 1705
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964491877405308e-05,
      "loss": 0.9456,
      "step": 1706
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9644477494459304e-05,
      "loss": 1.0233,
      "step": 1707
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964403594279783e-05,
      "loss": 0.9492,
      "step": 1708
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9643594119073524e-05,
      "loss": 1.3328,
      "step": 1709
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964315202329127e-05,
      "loss": 0.9127,
      "step": 1710
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964270965545596e-05,
      "loss": 0.9579,
      "step": 1711
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9642267015572464e-05,
      "loss": 1.0894,
      "step": 1712
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964182410364567e-05,
      "loss": 0.9579,
      "step": 1713
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964138091968047e-05,
      "loss": 1.0091,
      "step": 1714
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9640937463681755e-05,
      "loss": 1.2508,
      "step": 1715
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964049373565443e-05,
      "loss": 1.2461,
      "step": 1716
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.964004973560338e-05,
      "loss": 1.6601,
      "step": 1717
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9639605463533514e-05,
      "loss": 1.6161,
      "step": 1718
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963916091944973e-05,
      "loss": 0.9546,
      "step": 1719
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963871610335695e-05,
      "loss": 1.0137,
      "step": 1720
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9638271015260074e-05,
      "loss": 1.0219,
      "step": 1721
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963782565516401e-05,
      "loss": 1.0666,
      "step": 1722
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9637380023073684e-05,
      "loss": 1.144,
      "step": 1723
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963693411899402e-05,
      "loss": 1.1526,
      "step": 1724
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963648794292992e-05,
      "loss": 1.2056,
      "step": 1725
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9636041494886334e-05,
      "loss": 0.9895,
      "step": 1726
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963559477486818e-05,
      "loss": 0.9514,
      "step": 1727
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963514778288039e-05,
      "loss": 0.9441,
      "step": 1728
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.96347005189279e-05,
      "loss": 1.089,
      "step": 1729
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963425298301564e-05,
      "loss": 0.8773,
      "step": 1730
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963380517514856e-05,
      "loss": 1.0073,
      "step": 1731
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.96333570953316e-05,
      "loss": 0.8075,
      "step": 1732
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963290874356971e-05,
      "loss": 0.9427,
      "step": 1733
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963246011986783e-05,
      "loss": 1.1137,
      "step": 1734
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9632011224230924e-05,
      "loss": 1.0577,
      "step": 1735
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963156205666394e-05,
      "loss": 0.9742,
      "step": 1736
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963111261717184e-05,
      "loss": 0.7015,
      "step": 1737
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963066290575959e-05,
      "loss": 0.9625,
      "step": 1738
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.963021292243215e-05,
      "loss": 1.071,
      "step": 1739
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962976266719449e-05,
      "loss": 0.9823,
      "step": 1740
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9629312140051574e-05,
      "loss": 0.9389,
      "step": 1741
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962886134100838e-05,
      "loss": 0.7352,
      "step": 1742
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9628410270069884e-05,
      "loss": 0.966,
      "step": 1743
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962795892724107e-05,
      "loss": 0.9184,
      "step": 1744
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962750731252692e-05,
      "loss": 0.7261,
      "step": 1745
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962705542593241e-05,
      "loss": 1.5175,
      "step": 1746
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962660326746254e-05,
      "loss": 3.4503,
      "step": 1747
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9626150837122295e-05,
      "loss": 3.127,
      "step": 1748
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962569813491668e-05,
      "loss": 3.1469,
      "step": 1749
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962524516085067e-05,
      "loss": 1.7431,
      "step": 1750
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9624791914929294e-05,
      "loss": 1.0317,
      "step": 1751
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9624338397157536e-05,
      "loss": 0.9366,
      "step": 1752
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962388460754041e-05,
      "loss": 1.1032,
      "step": 1753
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9623430546082926e-05,
      "loss": 0.9766,
      "step": 1754
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9622976212790096e-05,
      "loss": 1.0437,
      "step": 1755
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962252160766693e-05,
      "loss": 1.0331,
      "step": 1756
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962206673071846e-05,
      "loss": 1.0539,
      "step": 1757
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9621611581949686e-05,
      "loss": 1.0927,
      "step": 1758
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962115616136566e-05,
      "loss": 1.2316,
      "step": 1759
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962070046897139e-05,
      "loss": 0.9607,
      "step": 1760
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.962024450477192e-05,
      "loss": 0.8528,
      "step": 1761
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9619788268772264e-05,
      "loss": 0.2426,
      "step": 1762
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9619331760977475e-05,
      "loss": 1.2277,
      "step": 1763
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9618874981392596e-05,
      "loss": 0.9902,
      "step": 1764
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961841793002265e-05,
      "loss": 0.9432,
      "step": 1765
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.96179606068727e-05,
      "loss": 0.902,
      "step": 1766
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.96175030119478e-05,
      "loss": 1.114,
      "step": 1767
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9617045145252983e-05,
      "loss": 1.3514,
      "step": 1768
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961658700679331e-05,
      "loss": 0.9533,
      "step": 1769
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961612859657384e-05,
      "loss": 1.0412,
      "step": 1770
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9615669914599636e-05,
      "loss": 1.0514,
      "step": 1771
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961521096087577e-05,
      "loss": 0.903,
      "step": 1772
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961475173540729e-05,
      "loss": 0.9029,
      "step": 1773
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961429223819928e-05,
      "loss": 0.9013,
      "step": 1774
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.96138324692568e-05,
      "loss": 0.9206,
      "step": 1775
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9613372428584934e-05,
      "loss": 0.8634,
      "step": 1776
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961291211618877e-05,
      "loss": 1.1974,
      "step": 1777
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961245153207337e-05,
      "loss": 0.8354,
      "step": 1778
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961199067624383e-05,
      "loss": 0.995,
      "step": 1779
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9611529548705236e-05,
      "loss": 1.0538,
      "step": 1780
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961106814946268e-05,
      "loss": 1.7806,
      "step": 1781
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961060647852126e-05,
      "loss": 1.2198,
      "step": 1782
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.961014453588606e-05,
      "loss": 1.0059,
      "step": 1783
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960968232156219e-05,
      "loss": 0.8188,
      "step": 1784
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960921983555474e-05,
      "loss": 0.9596,
      "step": 1785
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960875707786884e-05,
      "loss": 1.0665,
      "step": 1786
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960829404850958e-05,
      "loss": 0.3211,
      "step": 1787
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960783074748208e-05,
      "loss": 0.1984,
      "step": 1788
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960736717479144e-05,
      "loss": 0.2369,
      "step": 1789
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960690333044279e-05,
      "loss": 0.6787,
      "step": 1790
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960643921444125e-05,
      "loss": 1.0843,
      "step": 1791
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9605974826791944e-05,
      "loss": 1.1137,
      "step": 1792
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9605510167500005e-05,
      "loss": 0.9691,
      "step": 1793
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960504523657054e-05,
      "loss": 1.0397,
      "step": 1794
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9604580034008706e-05,
      "loss": 0.7967,
      "step": 1795
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9604114559819626e-05,
      "loss": 0.7106,
      "step": 1796
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960364881400844e-05,
      "loss": 1.0339,
      "step": 1797
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.96031827965803e-05,
      "loss": 0.9758,
      "step": 1798
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9602716507540335e-05,
      "loss": 1.1729,
      "step": 1799
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960224994689371e-05,
      "loss": 0.9571,
      "step": 1800
    },
    {
      "epoch": 0.06,
      "eval_loss": 0.9705952405929565,
      "eval_runtime": 318.5076,
      "eval_samples_per_second": 6.753,
      "eval_steps_per_second": 1.689,
      "step": 1800
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960178311464555e-05,
      "loss": 1.3212,
      "step": 1801
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960131601080104e-05,
      "loss": 2.4568,
      "step": 1802
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.960084863536531e-05,
      "loss": 2.1943,
      "step": 1803
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9600380988343534e-05,
      "loss": 2.1523,
      "step": 1804
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.959991306974088e-05,
      "loss": 2.0587,
      "step": 1805
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9599444879562494e-05,
      "loss": 2.0411,
      "step": 1806
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.959897641781356e-05,
      "loss": 1.6872,
      "step": 1807
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9598507684499246e-05,
      "loss": 1.1127,
      "step": 1808
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9598038679624724e-05,
      "loss": 1.0001,
      "step": 1809
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9597569403195174e-05,
      "loss": 0.9221,
      "step": 1810
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9597099855215784e-05,
      "loss": 1.0328,
      "step": 1811
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9596630035691726e-05,
      "loss": 1.0431,
      "step": 1812
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9596159944628184e-05,
      "loss": 0.9884,
      "step": 1813
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.959568958203037e-05,
      "loss": 1.0445,
      "step": 1814
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.959521894790344e-05,
      "loss": 1.1106,
      "step": 1815
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.959474804225263e-05,
      "loss": 0.9437,
      "step": 1816
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.959427686508311e-05,
      "loss": 1.1665,
      "step": 1817
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9593805416400096e-05,
      "loss": 0.9676,
      "step": 1818
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9593333696208785e-05,
      "loss": 1.0174,
      "step": 1819
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.959286170451439e-05,
      "loss": 0.968,
      "step": 1820
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9592389441322115e-05,
      "loss": 0.9266,
      "step": 1821
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9591916906637184e-05,
      "loss": 1.0276,
      "step": 1822
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9591444100464804e-05,
      "loss": 1.1469,
      "step": 1823
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.95909710228102e-05,
      "loss": 0.4562,
      "step": 1824
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.959049767367859e-05,
      "loss": 0.3489,
      "step": 1825
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.959002405307521e-05,
      "loss": 0.3481,
      "step": 1826
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958955016100527e-05,
      "loss": 1.0544,
      "step": 1827
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958907599747402e-05,
      "loss": 0.8931,
      "step": 1828
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958860156248668e-05,
      "loss": 0.9778,
      "step": 1829
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.95881268560485e-05,
      "loss": 1.0301,
      "step": 1830
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958765187816472e-05,
      "loss": 0.9595,
      "step": 1831
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958717662884057e-05,
      "loss": 0.8636,
      "step": 1832
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958670110808131e-05,
      "loss": 0.9368,
      "step": 1833
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958622531589219e-05,
      "loss": 0.8953,
      "step": 1834
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958574925227846e-05,
      "loss": 1.0642,
      "step": 1835
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958527291724536e-05,
      "loss": 1.0492,
      "step": 1836
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9584796310798176e-05,
      "loss": 1.0195,
      "step": 1837
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958431943294215e-05,
      "loss": 1.1029,
      "step": 1838
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958384228368255e-05,
      "loss": 0.9577,
      "step": 1839
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958336486302464e-05,
      "loss": 1.0498,
      "step": 1840
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9582887170973716e-05,
      "loss": 0.8328,
      "step": 1841
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958240920753502e-05,
      "loss": 0.7538,
      "step": 1842
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9581930972713835e-05,
      "loss": 1.4796,
      "step": 1843
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958145246651545e-05,
      "loss": 1.5185,
      "step": 1844
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9580973688945155e-05,
      "loss": 1.5321,
      "step": 1845
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958049464000821e-05,
      "loss": 1.4869,
      "step": 1846
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.958001531970993e-05,
      "loss": 1.4443,
      "step": 1847
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.957953572805558e-05,
      "loss": 1.4265,
      "step": 1848
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.957905586505048e-05,
      "loss": 1.4069,
      "step": 1849
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9578575730699916e-05,
      "loss": 1.1981,
      "step": 1850
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.957809532500919e-05,
      "loss": 1.0623,
      "step": 1851
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.95776146479836e-05,
      "loss": 1.0764,
      "step": 1852
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.957713369962846e-05,
      "loss": 0.9631,
      "step": 1853
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.957665247994908e-05,
      "loss": 0.9781,
      "step": 1854
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.957617098895076e-05,
      "loss": 1.0183,
      "step": 1855
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9575689226638834e-05,
      "loss": 0.9941,
      "step": 1856
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.95752071930186e-05,
      "loss": 1.1472,
      "step": 1857
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.95747248880954e-05,
      "loss": 0.9036,
      "step": 1858
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9574242311874555e-05,
      "loss": 1.0072,
      "step": 1859
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9573759464361376e-05,
      "loss": 1.0828,
      "step": 1860
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.95732763455612e-05,
      "loss": 0.9375,
      "step": 1861
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9572792955479374e-05,
      "loss": 1.1716,
      "step": 1862
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9572309294121225e-05,
      "loss": 0.9239,
      "step": 1863
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.95718253614921e-05,
      "loss": 1.1708,
      "step": 1864
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.957134115759732e-05,
      "loss": 0.9947,
      "step": 1865
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.957085668244225e-05,
      "loss": 0.7812,
      "step": 1866
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9570371936032233e-05,
      "loss": 1.0636,
      "step": 1867
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956988691837262e-05,
      "loss": 1.0979,
      "step": 1868
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956940162946877e-05,
      "loss": 1.0255,
      "step": 1869
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956891606932603e-05,
      "loss": 0.9941,
      "step": 1870
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956843023794977e-05,
      "loss": 1.0674,
      "step": 1871
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956794413534536e-05,
      "loss": 0.9336,
      "step": 1872
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9567457761518146e-05,
      "loss": 0.8708,
      "step": 1873
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9566971116473504e-05,
      "loss": 0.7944,
      "step": 1874
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956648420021682e-05,
      "loss": 0.6372,
      "step": 1875
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9565997012753454e-05,
      "loss": 0.9903,
      "step": 1876
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9565509554088794e-05,
      "loss": 1.0579,
      "step": 1877
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956502182422822e-05,
      "loss": 0.9117,
      "step": 1878
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956453382317711e-05,
      "loss": 1.0752,
      "step": 1879
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9564045550940856e-05,
      "loss": 1.0344,
      "step": 1880
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9563557007524854e-05,
      "loss": 1.0073,
      "step": 1881
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956306819293448e-05,
      "loss": 0.9551,
      "step": 1882
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956257910717515e-05,
      "loss": 0.9711,
      "step": 1883
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956208975025226e-05,
      "loss": 0.9521,
      "step": 1884
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956160012217119e-05,
      "loss": 0.961,
      "step": 1885
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.956111022293738e-05,
      "loss": 1.0823,
      "step": 1886
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9560620052556215e-05,
      "loss": 0.9248,
      "step": 1887
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9560129611033115e-05,
      "loss": 0.6185,
      "step": 1888
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.955963889837348e-05,
      "loss": 0.9288,
      "step": 1889
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9559147914582754e-05,
      "loss": 0.9603,
      "step": 1890
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9558656659666335e-05,
      "loss": 0.9331,
      "step": 1891
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.955816513362965e-05,
      "loss": 0.9989,
      "step": 1892
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.955767333647814e-05,
      "loss": 1.081,
      "step": 1893
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9557181268217227e-05,
      "loss": 0.9692,
      "step": 1894
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9556688928852325e-05,
      "loss": 0.9809,
      "step": 1895
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9556196318388884e-05,
      "loss": 1.0741,
      "step": 1896
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.955570343683235e-05,
      "loss": 0.7715,
      "step": 1897
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9555210284188156e-05,
      "loss": 0.9101,
      "step": 1898
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9554716860461745e-05,
      "loss": 1.1469,
      "step": 1899
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.955422316565857e-05,
      "loss": 1.025,
      "step": 1900
    },
    {
      "epoch": 0.06,
      "eval_loss": 0.97469562292099,
      "eval_runtime": 372.8621,
      "eval_samples_per_second": 5.769,
      "eval_steps_per_second": 1.443,
      "step": 1900
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.955372919978407e-05,
      "loss": 0.9755,
      "step": 1901
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.955323496284371e-05,
      "loss": 1.1381,
      "step": 1902
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.955274045484294e-05,
      "loss": 0.613,
      "step": 1903
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9552245675787225e-05,
      "loss": 1.0287,
      "step": 1904
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9551750625682014e-05,
      "loss": 0.8106,
      "step": 1905
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9551255304532784e-05,
      "loss": 1.0035,
      "step": 1906
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9550759712345006e-05,
      "loss": 1.0089,
      "step": 1907
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.955026384912415e-05,
      "loss": 0.795,
      "step": 1908
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.954976771487567e-05,
      "loss": 1.0158,
      "step": 1909
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9549271309605075e-05,
      "loss": 1.0131,
      "step": 1910
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.954877463331782e-05,
      "loss": 1.0234,
      "step": 1911
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9548277686019396e-05,
      "loss": 1.0523,
      "step": 1912
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.95477804677153e-05,
      "loss": 0.9238,
      "step": 1913
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9547282978411e-05,
      "loss": 0.9993,
      "step": 1914
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9546785218112004e-05,
      "loss": 0.8089,
      "step": 1915
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.954628718682381e-05,
      "loss": 0.7369,
      "step": 1916
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.95457888845519e-05,
      "loss": 0.9586,
      "step": 1917
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.954529031130179e-05,
      "loss": 1.0196,
      "step": 1918
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9544791467078974e-05,
      "loss": 0.995,
      "step": 1919
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9544292351888966e-05,
      "loss": 0.9712,
      "step": 1920
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.954379296573727e-05,
      "loss": 0.8103,
      "step": 1921
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9543293308629415e-05,
      "loss": 0.2603,
      "step": 1922
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.954279338057089e-05,
      "loss": 0.9053,
      "step": 1923
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9542293181567234e-05,
      "loss": 0.921,
      "step": 1924
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.954179271162396e-05,
      "loss": 0.973,
      "step": 1925
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9541291970746603e-05,
      "loss": 0.4652,
      "step": 1926
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.954079095894069e-05,
      "loss": 0.9109,
      "step": 1927
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.954028967621173e-05,
      "loss": 0.7283,
      "step": 1928
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953978812256529e-05,
      "loss": 0.1288,
      "step": 1929
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953928629800688e-05,
      "loss": 0.0856,
      "step": 1930
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9538784202542055e-05,
      "loss": 0.0473,
      "step": 1931
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953828183617635e-05,
      "loss": 0.885,
      "step": 1932
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953777919891532e-05,
      "loss": 1.076,
      "step": 1933
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953727629076451e-05,
      "loss": 0.9592,
      "step": 1934
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953677311172946e-05,
      "loss": 1.079,
      "step": 1935
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9536269661815746e-05,
      "loss": 1.3693,
      "step": 1936
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953576594102891e-05,
      "loss": 0.8405,
      "step": 1937
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9535261949374524e-05,
      "loss": 0.9565,
      "step": 1938
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9534757686858144e-05,
      "loss": 0.6503,
      "step": 1939
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953425315348534e-05,
      "loss": 1.0261,
      "step": 1940
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953374834926168e-05,
      "loss": 1.0757,
      "step": 1941
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953324327419274e-05,
      "loss": 0.949,
      "step": 1942
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953273792828409e-05,
      "loss": 0.9505,
      "step": 1943
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9532232311541314e-05,
      "loss": 0.9798,
      "step": 1944
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953172642396999e-05,
      "loss": 0.7895,
      "step": 1945
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9531220265575714e-05,
      "loss": 1.1307,
      "step": 1946
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9530713836364065e-05,
      "loss": 1.0144,
      "step": 1947
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.953020713634062e-05,
      "loss": 1.0383,
      "step": 1948
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.9529700165511e-05,
      "loss": 0.9562,
      "step": 1949
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952919292388078e-05,
      "loss": 0.9956,
      "step": 1950
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9528685411455576e-05,
      "loss": 0.8281,
      "step": 1951
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952817762824099e-05,
      "loss": 0.9424,
      "step": 1952
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952766957424261e-05,
      "loss": 0.9436,
      "step": 1953
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952716124946606e-05,
      "loss": 0.9714,
      "step": 1954
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952665265391695e-05,
      "loss": 1.0402,
      "step": 1955
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9526143787600885e-05,
      "loss": 0.7521,
      "step": 1956
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.95256346505235e-05,
      "loss": 1.0698,
      "step": 1957
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9525125242690396e-05,
      "loss": 0.9659,
      "step": 1958
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952461556410721e-05,
      "loss": 1.033,
      "step": 1959
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952410561477957e-05,
      "loss": 1.087,
      "step": 1960
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.95235953947131e-05,
      "loss": 1.0146,
      "step": 1961
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952308490391342e-05,
      "loss": 0.8877,
      "step": 1962
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.95225741423862e-05,
      "loss": 0.9322,
      "step": 1963
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952206311013705e-05,
      "loss": 1.0066,
      "step": 1964
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9521551807171614e-05,
      "loss": 0.9071,
      "step": 1965
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952104023349555e-05,
      "loss": 1.8599,
      "step": 1966
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.952052838911449e-05,
      "loss": 0.9741,
      "step": 1967
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9520016274034095e-05,
      "loss": 0.9339,
      "step": 1968
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.951950388826002e-05,
      "loss": 1.0426,
      "step": 1969
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.951899123179792e-05,
      "loss": 1.0421,
      "step": 1970
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.951847830465345e-05,
      "loss": 1.0784,
      "step": 1971
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.951796510683227e-05,
      "loss": 1.1452,
      "step": 1972
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.951745163834005e-05,
      "loss": 1.0039,
      "step": 1973
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.951693789918246e-05,
      "loss": 0.891,
      "step": 1974
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9516423889365176e-05,
      "loss": 1.2564,
      "step": 1975
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9515909608893865e-05,
      "loss": 0.7873,
      "step": 1976
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.95153950577742e-05,
      "loss": 1.1434,
      "step": 1977
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.951488023601188e-05,
      "loss": 1.083,
      "step": 1978
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9514365143612565e-05,
      "loss": 0.8484,
      "step": 1979
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.951384978058196e-05,
      "loss": 1.5933,
      "step": 1980
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.951333414692575e-05,
      "loss": 2.3786,
      "step": 1981
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.951281824264962e-05,
      "loss": 0.8993,
      "step": 1982
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9512302067759264e-05,
      "loss": 1.1638,
      "step": 1983
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9511785622260396e-05,
      "loss": 0.9947,
      "step": 1984
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.95112689061587e-05,
      "loss": 1.0299,
      "step": 1985
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9510751919459895e-05,
      "loss": 0.9923,
      "step": 1986
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9510234662169686e-05,
      "loss": 0.9104,
      "step": 1987
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950971713429377e-05,
      "loss": 0.9995,
      "step": 1988
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950919933583787e-05,
      "loss": 1.1821,
      "step": 1989
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950868126680771e-05,
      "loss": 0.9905,
      "step": 1990
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9508162927209e-05,
      "loss": 0.9413,
      "step": 1991
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9507644317047455e-05,
      "loss": 1.0624,
      "step": 1992
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950712543632882e-05,
      "loss": 1.6918,
      "step": 1993
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9506606285058795e-05,
      "loss": 0.9903,
      "step": 1994
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9506086863243144e-05,
      "loss": 0.8224,
      "step": 1995
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950556717088758e-05,
      "loss": 0.9957,
      "step": 1996
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950504720799785e-05,
      "loss": 0.9611,
      "step": 1997
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950452697457968e-05,
      "loss": 1.637,
      "step": 1998
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9504006470638836e-05,
      "loss": 1.0157,
      "step": 1999
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9503485696181045e-05,
      "loss": 2.0042,
      "step": 2000
    },
    {
      "epoch": 0.07,
      "eval_loss": 0.9700034260749817,
      "eval_runtime": 394.4097,
      "eval_samples_per_second": 5.454,
      "eval_steps_per_second": 1.364,
      "step": 2000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950296465121207e-05,
      "loss": 1.1226,
      "step": 2001
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950244333573765e-05,
      "loss": 1.0893,
      "step": 2002
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950192174976354e-05,
      "loss": 1.2914,
      "step": 2003
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9501399893295506e-05,
      "loss": 1.2657,
      "step": 2004
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950087776633931e-05,
      "loss": 1.2766,
      "step": 2005
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.950035536890072e-05,
      "loss": 1.2522,
      "step": 2006
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949983270098549e-05,
      "loss": 1.1146,
      "step": 2007
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94993097625994e-05,
      "loss": 1.0478,
      "step": 2008
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949878655374822e-05,
      "loss": 1.0554,
      "step": 2009
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949826307443772e-05,
      "loss": 0.903,
      "step": 2010
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94977393246737e-05,
      "loss": 0.9445,
      "step": 2011
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949721530446191e-05,
      "loss": 1.1377,
      "step": 2012
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949669101380816e-05,
      "loss": 2.0,
      "step": 2013
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949616645271823e-05,
      "loss": 1.4152,
      "step": 2014
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949564162119791e-05,
      "loss": 1.0261,
      "step": 2015
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949511651925299e-05,
      "loss": 0.8683,
      "step": 2016
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9494591146889276e-05,
      "loss": 1.1752,
      "step": 2017
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9494065504112575e-05,
      "loss": 1.8985,
      "step": 2018
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949353959092866e-05,
      "loss": 1.9257,
      "step": 2019
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9493013407343366e-05,
      "loss": 0.9505,
      "step": 2020
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949248695336249e-05,
      "loss": 0.8719,
      "step": 2021
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949196022899185e-05,
      "loss": 0.89,
      "step": 2022
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949143323423725e-05,
      "loss": 0.9317,
      "step": 2023
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9490905969104514e-05,
      "loss": 1.0108,
      "step": 2024
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.949037843359947e-05,
      "loss": 0.9833,
      "step": 2025
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948985062772793e-05,
      "loss": 1.059,
      "step": 2026
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948932255149573e-05,
      "loss": 0.9866,
      "step": 2027
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94887942049087e-05,
      "loss": 0.9897,
      "step": 2028
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948826558797265e-05,
      "loss": 1.0157,
      "step": 2029
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9487736700693446e-05,
      "loss": 1.0186,
      "step": 2030
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948720754307692e-05,
      "loss": 1.0471,
      "step": 2031
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94866781151289e-05,
      "loss": 0.9309,
      "step": 2032
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948614841685524e-05,
      "loss": 1.0723,
      "step": 2033
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94856184482618e-05,
      "loss": 0.9658,
      "step": 2034
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9485088209354403e-05,
      "loss": 1.0141,
      "step": 2035
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948455770013892e-05,
      "loss": 0.931,
      "step": 2036
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9484026920621205e-05,
      "loss": 1.0697,
      "step": 2037
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948349587080713e-05,
      "loss": 0.9488,
      "step": 2038
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948296455070253e-05,
      "loss": 0.9973,
      "step": 2039
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94824329603133e-05,
      "loss": 1.0139,
      "step": 2040
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9481901099645275e-05,
      "loss": 0.7921,
      "step": 2041
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948136896870437e-05,
      "loss": 0.8358,
      "step": 2042
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948083656749642e-05,
      "loss": 1.0265,
      "step": 2043
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.948030389602733e-05,
      "loss": 0.9578,
      "step": 2044
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947977095430295e-05,
      "loss": 0.8841,
      "step": 2045
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94792377423292e-05,
      "loss": 1.0986,
      "step": 2046
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947870426011194e-05,
      "loss": 1.068,
      "step": 2047
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9478170507657076e-05,
      "loss": 0.8848,
      "step": 2048
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947763648497048e-05,
      "loss": 0.9744,
      "step": 2049
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947710219205808e-05,
      "loss": 1.0402,
      "step": 2050
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947656762892574e-05,
      "loss": 1.0524,
      "step": 2051
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9476032795579375e-05,
      "loss": 0.9523,
      "step": 2052
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94754976920249e-05,
      "loss": 2.9165,
      "step": 2053
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947496231826821e-05,
      "loss": 2.4561,
      "step": 2054
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947442667431522e-05,
      "loss": 1.0039,
      "step": 2055
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947389076017184e-05,
      "loss": 1.0973,
      "step": 2056
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947335457584398e-05,
      "loss": 0.8869,
      "step": 2057
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947281812133757e-05,
      "loss": 0.6934,
      "step": 2058
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9472281396658535e-05,
      "loss": 0.9718,
      "step": 2059
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94717444018128e-05,
      "loss": 0.9535,
      "step": 2060
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947120713680628e-05,
      "loss": 0.8392,
      "step": 2061
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9470669601644916e-05,
      "loss": 1.0808,
      "step": 2062
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.947013179633464e-05,
      "loss": 0.9043,
      "step": 2063
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946959372088139e-05,
      "loss": 1.2519,
      "step": 2064
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9469055375291114e-05,
      "loss": 0.767,
      "step": 2065
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946851675956974e-05,
      "loss": 0.9181,
      "step": 2066
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9467977873723225e-05,
      "loss": 0.9626,
      "step": 2067
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946743871775751e-05,
      "loss": 0.9452,
      "step": 2068
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9466899291678556e-05,
      "loss": 1.1195,
      "step": 2069
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946635959549231e-05,
      "loss": 0.9542,
      "step": 2070
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946581962920473e-05,
      "loss": 1.0629,
      "step": 2071
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946527939282179e-05,
      "loss": 1.0338,
      "step": 2072
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9464738886349437e-05,
      "loss": 0.8698,
      "step": 2073
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946419810979366e-05,
      "loss": 0.9966,
      "step": 2074
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94636570631604e-05,
      "loss": 1.4957,
      "step": 2075
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946311574645565e-05,
      "loss": 1.0641,
      "step": 2076
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946257415968538e-05,
      "loss": 0.9673,
      "step": 2077
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9462032302855576e-05,
      "loss": 1.0034,
      "step": 2078
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9461490175972206e-05,
      "loss": 1.0195,
      "step": 2079
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946094777904127e-05,
      "loss": 1.2112,
      "step": 2080
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.946040511206874e-05,
      "loss": 0.7135,
      "step": 2081
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.945986217506062e-05,
      "loss": 1.0247,
      "step": 2082
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.945931896802291e-05,
      "loss": 0.9684,
      "step": 2083
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.945877549096159e-05,
      "loss": 1.2075,
      "step": 2084
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9458231743882656e-05,
      "loss": 0.9651,
      "step": 2085
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9457687726792125e-05,
      "loss": 0.9649,
      "step": 2086
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9457143439696005e-05,
      "loss": 1.2671,
      "step": 2087
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.945659888260029e-05,
      "loss": 0.9887,
      "step": 2088
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9456054055511005e-05,
      "loss": 0.8647,
      "step": 2089
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.945550895843416e-05,
      "loss": 0.574,
      "step": 2090
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9454963591375777e-05,
      "loss": 0.7752,
      "step": 2091
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.945441795434187e-05,
      "loss": 0.9509,
      "step": 2092
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.945387204733846e-05,
      "loss": 1.0011,
      "step": 2093
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.945332587037158e-05,
      "loss": 1.0085,
      "step": 2094
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.945277942344726e-05,
      "loss": 1.8374,
      "step": 2095
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9452232706571535e-05,
      "loss": 1.7547,
      "step": 2096
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.945168571975043e-05,
      "loss": 0.9547,
      "step": 2097
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.945113846298999e-05,
      "loss": 0.9939,
      "step": 2098
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9450590936296264e-05,
      "loss": 0.9841,
      "step": 2099
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9450043139675286e-05,
      "loss": 0.8857,
      "step": 2100
    },
    {
      "epoch": 0.07,
      "eval_loss": 0.9579198360443115,
      "eval_runtime": 322.2639,
      "eval_samples_per_second": 6.675,
      "eval_steps_per_second": 1.669,
      "step": 2100
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9449495073133104e-05,
      "loss": 0.5438,
      "step": 2101
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9448946736675775e-05,
      "loss": 1.0372,
      "step": 2102
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9448398130309345e-05,
      "loss": 1.0237,
      "step": 2103
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.944784925403988e-05,
      "loss": 1.1998,
      "step": 2104
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.944730010787343e-05,
      "loss": 0.9505,
      "step": 2105
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9446750691816056e-05,
      "loss": 0.8634,
      "step": 2106
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.944620100587383e-05,
      "loss": 0.9956,
      "step": 2107
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.944565105005282e-05,
      "loss": 1.149,
      "step": 2108
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94451008243591e-05,
      "loss": 0.7212,
      "step": 2109
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9444550328798736e-05,
      "loss": 1.0262,
      "step": 2110
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9443999563377815e-05,
      "loss": 0.9625,
      "step": 2111
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.944344852810241e-05,
      "loss": 0.9896,
      "step": 2112
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9442897222978603e-05,
      "loss": 0.9591,
      "step": 2113
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.944234564801249e-05,
      "loss": 0.9425,
      "step": 2114
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9441793803210145e-05,
      "loss": 1.0791,
      "step": 2115
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.944124168857767e-05,
      "loss": 0.8407,
      "step": 2116
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.944068930412117e-05,
      "loss": 1.0175,
      "step": 2117
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.944013664984672e-05,
      "loss": 0.2582,
      "step": 2118
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9439583725760435e-05,
      "loss": 0.0665,
      "step": 2119
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.943903053186843e-05,
      "loss": 0.7765,
      "step": 2120
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.943847706817678e-05,
      "loss": 1.0479,
      "step": 2121
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9437923334691624e-05,
      "loss": 0.7264,
      "step": 2122
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.943736933141907e-05,
      "loss": 0.7625,
      "step": 2123
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.943681505836523e-05,
      "loss": 1.1327,
      "step": 2124
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.943626051553621e-05,
      "loss": 0.9687,
      "step": 2125
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.943570570293816e-05,
      "loss": 1.0171,
      "step": 2126
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9435150620577185e-05,
      "loss": 0.9673,
      "step": 2127
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.943459526845942e-05,
      "loss": 1.0035,
      "step": 2128
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9434039646590984e-05,
      "loss": 1.0711,
      "step": 2129
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.943348375497803e-05,
      "loss": 2.1862,
      "step": 2130
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9432927593626676e-05,
      "loss": 0.9636,
      "step": 2131
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.943237116254308e-05,
      "loss": 0.9645,
      "step": 2132
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9431814461733375e-05,
      "loss": 0.8808,
      "step": 2133
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94312574912037e-05,
      "loss": 0.6965,
      "step": 2134
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9430700250960224e-05,
      "loss": 0.8517,
      "step": 2135
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9430142741009075e-05,
      "loss": 1.0179,
      "step": 2136
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.942958496135642e-05,
      "loss": 1.0302,
      "step": 2137
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9429026912008426e-05,
      "loss": 1.0105,
      "step": 2138
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.942846859297124e-05,
      "loss": 1.0461,
      "step": 2139
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9427910004251023e-05,
      "loss": 0.7591,
      "step": 2140
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.942735114585396e-05,
      "loss": 1.0725,
      "step": 2141
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.942679201778619e-05,
      "loss": 1.0239,
      "step": 2142
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.942623262005393e-05,
      "loss": 0.9477,
      "step": 2143
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94256729526633e-05,
      "loss": 0.9361,
      "step": 2144
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9425113015620525e-05,
      "loss": 0.952,
      "step": 2145
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9424552808931775e-05,
      "loss": 1.0321,
      "step": 2146
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.942399233260322e-05,
      "loss": 1.1245,
      "step": 2147
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.942343158664106e-05,
      "loss": 1.2402,
      "step": 2148
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.942287057105148e-05,
      "loss": 1.1377,
      "step": 2149
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.942230928584068e-05,
      "loss": 0.9803,
      "step": 2150
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.942174773101485e-05,
      "loss": 1.003,
      "step": 2151
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.942118590658019e-05,
      "loss": 0.9624,
      "step": 2152
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9420623812542906e-05,
      "loss": 0.9771,
      "step": 2153
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94200614489092e-05,
      "loss": 0.9832,
      "step": 2154
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9419498815685286e-05,
      "loss": 0.9096,
      "step": 2155
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.941893591287736e-05,
      "loss": 1.0287,
      "step": 2156
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.941837274049166e-05,
      "loss": 0.8813,
      "step": 2157
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.941780929853438e-05,
      "loss": 1.3419,
      "step": 2158
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.941724558701176e-05,
      "loss": 1.3326,
      "step": 2159
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9416681605930005e-05,
      "loss": 1.1339,
      "step": 2160
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.941611735529535e-05,
      "loss": 1.062,
      "step": 2161
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.941555283511404e-05,
      "loss": 0.9952,
      "step": 2162
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.941498804539227e-05,
      "loss": 0.9525,
      "step": 2163
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.941442298613631e-05,
      "loss": 1.9246,
      "step": 2164
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.941385765735239e-05,
      "loss": 0.9599,
      "step": 2165
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9413292059046736e-05,
      "loss": 0.9967,
      "step": 2166
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.94127261912256e-05,
      "loss": 0.9854,
      "step": 2167
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9412160053895234e-05,
      "loss": 1.1172,
      "step": 2168
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.941159364706188e-05,
      "loss": 1.026,
      "step": 2169
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9411026970731805e-05,
      "loss": 1.0519,
      "step": 2170
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9410460024911254e-05,
      "loss": 4.1576,
      "step": 2171
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9409892809606486e-05,
      "loss": 5.6816,
      "step": 2172
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9409325324823755e-05,
      "loss": 5.3057,
      "step": 2173
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9408757570569345e-05,
      "loss": 5.0603,
      "step": 2174
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9408189546849515e-05,
      "loss": 2.641,
      "step": 2175
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9407621253670533e-05,
      "loss": 0.9654,
      "step": 2176
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.940705269103867e-05,
      "loss": 0.9233,
      "step": 2177
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.940648385896021e-05,
      "loss": 0.7882,
      "step": 2178
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.940591475744143e-05,
      "loss": 1.1012,
      "step": 2179
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9405345386488614e-05,
      "loss": 0.5643,
      "step": 2180
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.940477574610805e-05,
      "loss": 0.4126,
      "step": 2181
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.940420583630602e-05,
      "loss": 0.3709,
      "step": 2182
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.940363565708882e-05,
      "loss": 0.9588,
      "step": 2183
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.940306520846274e-05,
      "loss": 1.0267,
      "step": 2184
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9402494490434086e-05,
      "loss": 0.9884,
      "step": 2185
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.940192350300915e-05,
      "loss": 1.0112,
      "step": 2186
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9401352246194246e-05,
      "loss": 0.9229,
      "step": 2187
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9400780719995666e-05,
      "loss": 0.9794,
      "step": 2188
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9400208924419724e-05,
      "loss": 1.0006,
      "step": 2189
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.939963685947274e-05,
      "loss": 0.759,
      "step": 2190
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.939906452516103e-05,
      "loss": 0.7719,
      "step": 2191
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.939849192149091e-05,
      "loss": 0.5818,
      "step": 2192
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.939791904846869e-05,
      "loss": 0.9629,
      "step": 2193
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.93973459061007e-05,
      "loss": 0.9179,
      "step": 2194
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.939677249439328e-05,
      "loss": 0.9943,
      "step": 2195
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.939619881335274e-05,
      "loss": 1.1332,
      "step": 2196
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9395624862985435e-05,
      "loss": 1.0189,
      "step": 2197
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.939505064329769e-05,
      "loss": 1.0336,
      "step": 2198
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9394476154295844e-05,
      "loss": 2.1914,
      "step": 2199
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9393901395986233e-05,
      "loss": 1.0495,
      "step": 2200
    },
    {
      "epoch": 0.07,
      "eval_loss": 0.9599708318710327,
      "eval_runtime": 327.3696,
      "eval_samples_per_second": 6.571,
      "eval_steps_per_second": 1.643,
      "step": 2200
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9393326368375216e-05,
      "loss": 0.9038,
      "step": 2201
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.939275107146912e-05,
      "loss": 0.6945,
      "step": 2202
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.939217550527433e-05,
      "loss": 0.927,
      "step": 2203
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9391599669797165e-05,
      "loss": 0.9647,
      "step": 2204
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9391023565044e-05,
      "loss": 0.9551,
      "step": 2205
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.93904471910212e-05,
      "loss": 0.9485,
      "step": 2206
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938987054773512e-05,
      "loss": 0.9996,
      "step": 2207
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938929363519212e-05,
      "loss": 1.0433,
      "step": 2208
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938871645339858e-05,
      "loss": 1.0905,
      "step": 2209
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938813900236085e-05,
      "loss": 1.0534,
      "step": 2210
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9387561282085344e-05,
      "loss": 1.0286,
      "step": 2211
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938698329257841e-05,
      "loss": 1.7744,
      "step": 2212
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938640503384644e-05,
      "loss": 5.8246,
      "step": 2213
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.93858265058958e-05,
      "loss": 6.9339,
      "step": 2214
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938524770873291e-05,
      "loss": 6.1171,
      "step": 2215
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9384668642364126e-05,
      "loss": 3.9928,
      "step": 2216
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9384089306795866e-05,
      "loss": 1.0207,
      "step": 2217
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938350970203452e-05,
      "loss": 1.0789,
      "step": 2218
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938292982808647e-05,
      "loss": 1.7634,
      "step": 2219
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938234968495814e-05,
      "loss": 1.1787,
      "step": 2220
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938176927265591e-05,
      "loss": 1.0604,
      "step": 2221
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938118859118621e-05,
      "loss": 1.0579,
      "step": 2222
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9380607640555444e-05,
      "loss": 1.0701,
      "step": 2223
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.938002642077002e-05,
      "loss": 1.0802,
      "step": 2224
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9379444931836365e-05,
      "loss": 0.8872,
      "step": 2225
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.937886317376089e-05,
      "loss": 1.0084,
      "step": 2226
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.937828114655002e-05,
      "loss": 1.0363,
      "step": 2227
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9377698850210176e-05,
      "loss": 0.956,
      "step": 2228
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.937711628474779e-05,
      "loss": 0.9577,
      "step": 2229
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9376533450169295e-05,
      "loss": 1.0459,
      "step": 2230
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9375950346481124e-05,
      "loss": 0.9979,
      "step": 2231
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.937536697368971e-05,
      "loss": 0.9786,
      "step": 2232
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.93747833318015e-05,
      "loss": 0.8911,
      "step": 2233
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.937419942082294e-05,
      "loss": 0.9335,
      "step": 2234
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.937361524076046e-05,
      "loss": 0.8954,
      "step": 2235
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.937303079162053e-05,
      "loss": 0.7257,
      "step": 2236
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9372446073409595e-05,
      "loss": 0.5031,
      "step": 2237
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.93718610861341e-05,
      "loss": 0.779,
      "step": 2238
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9371275829800514e-05,
      "loss": 0.9896,
      "step": 2239
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.937069030441529e-05,
      "loss": 0.9959,
      "step": 2240
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.937010450998489e-05,
      "loss": 1.0583,
      "step": 2241
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.93695184465158e-05,
      "loss": 0.9678,
      "step": 2242
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.936893211401448e-05,
      "loss": 0.9278,
      "step": 2243
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.936834551248739e-05,
      "loss": 0.7907,
      "step": 2244
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.936775864194101e-05,
      "loss": 0.8329,
      "step": 2245
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.936717150238184e-05,
      "loss": 0.7498,
      "step": 2246
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9366584093816345e-05,
      "loss": 0.7833,
      "step": 2247
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9365996416251e-05,
      "loss": 0.6747,
      "step": 2248
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.936540846969232e-05,
      "loss": 0.9213,
      "step": 2249
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.936482025414677e-05,
      "loss": 0.8,
      "step": 2250
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.936423176962086e-05,
      "loss": 1.0415,
      "step": 2251
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.936364301612108e-05,
      "loss": 1.0413,
      "step": 2252
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.936305399365392e-05,
      "loss": 0.8955,
      "step": 2253
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.936246470222591e-05,
      "loss": 0.9564,
      "step": 2254
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9361875141843525e-05,
      "loss": 1.095,
      "step": 2255
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.93612853125133e-05,
      "loss": 1.135,
      "step": 2256
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.936069521424171e-05,
      "loss": 1.0479,
      "step": 2257
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.936010484703532e-05,
      "loss": 0.9187,
      "step": 2258
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9359514210900605e-05,
      "loss": 1.0023,
      "step": 2259
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9358923305844104e-05,
      "loss": 1.1477,
      "step": 2260
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.935833213187234e-05,
      "loss": 5.7744,
      "step": 2261
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.935774068899184e-05,
      "loss": 5.7131,
      "step": 2262
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9357148977209125e-05,
      "loss": 1.1928,
      "step": 2263
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.935655699653073e-05,
      "loss": 0.7065,
      "step": 2264
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.93559647469632e-05,
      "loss": 1.0659,
      "step": 2265
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9355372228513065e-05,
      "loss": 0.9873,
      "step": 2266
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.935477944118687e-05,
      "loss": 0.9671,
      "step": 2267
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.935418638499115e-05,
      "loss": 1.0376,
      "step": 2268
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9353593059932455e-05,
      "loss": 1.5398,
      "step": 2269
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9352999466017345e-05,
      "loss": 1.2162,
      "step": 2270
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.935240560325237e-05,
      "loss": 0.91,
      "step": 2271
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.935181147164408e-05,
      "loss": 0.9151,
      "step": 2272
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.935121707119904e-05,
      "loss": 0.8356,
      "step": 2273
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.935062240192381e-05,
      "loss": 1.5099,
      "step": 2274
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.935002746382494e-05,
      "loss": 1.7609,
      "step": 2275
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.934943225690902e-05,
      "loss": 1.0574,
      "step": 2276
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.934883678118262e-05,
      "loss": 1.0542,
      "step": 2277
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.93482410366523e-05,
      "loss": 0.8656,
      "step": 2278
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.934764502332465e-05,
      "loss": 0.959,
      "step": 2279
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9347048741206235e-05,
      "loss": 0.9026,
      "step": 2280
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.934645219030365e-05,
      "loss": 1.099,
      "step": 2281
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.934585537062348e-05,
      "loss": 0.9111,
      "step": 2282
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.934525828217231e-05,
      "loss": 1.1509,
      "step": 2283
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.934466092495673e-05,
      "loss": 0.8686,
      "step": 2284
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.934406329898334e-05,
      "loss": 0.8891,
      "step": 2285
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.934346540425873e-05,
      "loss": 1.0703,
      "step": 2286
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9342867240789506e-05,
      "loss": 0.9686,
      "step": 2287
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.934226880858227e-05,
      "loss": 0.9068,
      "step": 2288
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.934167010764363e-05,
      "loss": 1.0195,
      "step": 2289
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.93410711379802e-05,
      "loss": 1.2331,
      "step": 2290
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9340471899598586e-05,
      "loss": 1.0009,
      "step": 2291
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9339872392505407e-05,
      "loss": 0.9987,
      "step": 2292
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.933927261670727e-05,
      "loss": 0.989,
      "step": 2293
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.933867257221081e-05,
      "loss": 0.9832,
      "step": 2294
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9338072259022646e-05,
      "loss": 1.1871,
      "step": 2295
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9337471677149404e-05,
      "loss": 1.1798,
      "step": 2296
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.933687082659771e-05,
      "loss": 1.0671,
      "step": 2297
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9336269707374215e-05,
      "loss": 1.079,
      "step": 2298
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.933566831948554e-05,
      "loss": 1.13,
      "step": 2299
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9335066662938334e-05,
      "loss": 0.9313,
      "step": 2300
    },
    {
      "epoch": 0.08,
      "eval_loss": 0.9834993481636047,
      "eval_runtime": 370.1415,
      "eval_samples_per_second": 5.811,
      "eval_steps_per_second": 1.453,
      "step": 2300
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9334464737739224e-05,
      "loss": 1.0544,
      "step": 2301
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9333862543894866e-05,
      "loss": 1.0117,
      "step": 2302
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.93332600814119e-05,
      "loss": 1.0226,
      "step": 2303
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.933265735029699e-05,
      "loss": 1.0191,
      "step": 2304
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.933205435055679e-05,
      "loss": 1.0393,
      "step": 2305
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.933145108219795e-05,
      "loss": 0.558,
      "step": 2306
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.933084754522711e-05,
      "loss": 0.8831,
      "step": 2307
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9330243739650964e-05,
      "loss": 0.9176,
      "step": 2308
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.932963966547617e-05,
      "loss": 1.0968,
      "step": 2309
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9329035322709386e-05,
      "loss": 1.147,
      "step": 2310
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.93284307113573e-05,
      "loss": 1.0087,
      "step": 2311
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.932782583142657e-05,
      "loss": 0.9341,
      "step": 2312
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9327220682923884e-05,
      "loss": 1.0213,
      "step": 2313
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9326615265855925e-05,
      "loss": 1.058,
      "step": 2314
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.932600958022936e-05,
      "loss": 0.8711,
      "step": 2315
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9325403626050896e-05,
      "loss": 0.9936,
      "step": 2316
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.932479740332722e-05,
      "loss": 0.9052,
      "step": 2317
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9324190912065003e-05,
      "loss": 1.0288,
      "step": 2318
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.932358415227096e-05,
      "loss": 0.9601,
      "step": 2319
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.93229771239518e-05,
      "loss": 0.9796,
      "step": 2320
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.932236982711419e-05,
      "loss": 1.0309,
      "step": 2321
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.932176226176486e-05,
      "loss": 1.0526,
      "step": 2322
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.932115442791052e-05,
      "loss": 1.0474,
      "step": 2323
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.932054632555787e-05,
      "loss": 1.6741,
      "step": 2324
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.931993795471361e-05,
      "loss": 1.0493,
      "step": 2325
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.931932931538449e-05,
      "loss": 0.8648,
      "step": 2326
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.93187204075772e-05,
      "loss": 0.8991,
      "step": 2327
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9318111231298475e-05,
      "loss": 1.0242,
      "step": 2328
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.931750178655504e-05,
      "loss": 0.966,
      "step": 2329
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9316892073353616e-05,
      "loss": 0.9343,
      "step": 2330
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9316282091700944e-05,
      "loss": 0.8834,
      "step": 2331
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.931567184160375e-05,
      "loss": 0.7359,
      "step": 2332
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.931506132306877e-05,
      "loss": 0.6719,
      "step": 2333
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.931445053610275e-05,
      "loss": 0.6248,
      "step": 2334
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9313839480712436e-05,
      "loss": 0.9481,
      "step": 2335
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.931322815690457e-05,
      "loss": 1.0667,
      "step": 2336
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9312616564685895e-05,
      "loss": 1.0191,
      "step": 2337
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9312004704063167e-05,
      "loss": 1.3862,
      "step": 2338
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9311392575043134e-05,
      "loss": 1.0844,
      "step": 2339
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9310780177632575e-05,
      "loss": 1.2153,
      "step": 2340
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.931016751183822e-05,
      "loss": 1.1834,
      "step": 2341
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.930955457766686e-05,
      "loss": 1.2698,
      "step": 2342
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.930894137512525e-05,
      "loss": 1.0259,
      "step": 2343
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9308327904220164e-05,
      "loss": 0.8681,
      "step": 2344
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9307714164958366e-05,
      "loss": 0.9295,
      "step": 2345
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.930710015734663e-05,
      "loss": 1.0345,
      "step": 2346
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.930648588139175e-05,
      "loss": 1.0319,
      "step": 2347
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.93058713371005e-05,
      "loss": 1.0078,
      "step": 2348
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9305256524479653e-05,
      "loss": 1.006,
      "step": 2349
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.930464144353602e-05,
      "loss": 1.1006,
      "step": 2350
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.930402609427636e-05,
      "loss": 1.1354,
      "step": 2351
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.930341047670749e-05,
      "loss": 1.0807,
      "step": 2352
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.93027945908362e-05,
      "loss": 0.9583,
      "step": 2353
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9302178436669286e-05,
      "loss": 0.8544,
      "step": 2354
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.930156201421356e-05,
      "loss": 0.9589,
      "step": 2355
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9300945323475815e-05,
      "loss": 0.9899,
      "step": 2356
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9300328364462864e-05,
      "loss": 1.924,
      "step": 2357
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929971113718153e-05,
      "loss": 0.9546,
      "step": 2358
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92990936416386e-05,
      "loss": 0.8097,
      "step": 2359
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929847587784091e-05,
      "loss": 1.1129,
      "step": 2360
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929785784579528e-05,
      "loss": 0.9999,
      "step": 2361
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929723954550853e-05,
      "loss": 0.9442,
      "step": 2362
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929662097698748e-05,
      "loss": 0.9558,
      "step": 2363
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929600214023896e-05,
      "loss": 0.9969,
      "step": 2364
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929538303526982e-05,
      "loss": 0.9809,
      "step": 2365
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929476366208687e-05,
      "loss": 1.0774,
      "step": 2366
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929414402069695e-05,
      "loss": 1.064,
      "step": 2367
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929352411110693e-05,
      "loss": 1.0711,
      "step": 2368
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9292903933323614e-05,
      "loss": 1.0517,
      "step": 2369
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929228348735387e-05,
      "loss": 1.1276,
      "step": 2370
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929166277320455e-05,
      "loss": 0.931,
      "step": 2371
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9291041790882495e-05,
      "loss": 0.9499,
      "step": 2372
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.929042054039458e-05,
      "loss": 0.9467,
      "step": 2373
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.928979902174763e-05,
      "loss": 0.9433,
      "step": 2374
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9289177234948535e-05,
      "loss": 1.0272,
      "step": 2375
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.928855518000416e-05,
      "loss": 1.0485,
      "step": 2376
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9287932856921346e-05,
      "loss": 1.0334,
      "step": 2377
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.928731026570699e-05,
      "loss": 1.7007,
      "step": 2378
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9286687406367954e-05,
      "loss": 4.0556,
      "step": 2379
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9286064278911114e-05,
      "loss": 0.8854,
      "step": 2380
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.928544088334335e-05,
      "loss": 1.0251,
      "step": 2381
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.928481721967154e-05,
      "loss": 1.0594,
      "step": 2382
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.928419328790259e-05,
      "loss": 1.2142,
      "step": 2383
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9283569088043354e-05,
      "loss": 1.0448,
      "step": 2384
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.928294462010075e-05,
      "loss": 0.97,
      "step": 2385
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.928231988408166e-05,
      "loss": 1.0528,
      "step": 2386
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.928169487999299e-05,
      "loss": 0.9345,
      "step": 2387
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9281069607841624e-05,
      "loss": 1.0428,
      "step": 2388
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.928044406763447e-05,
      "loss": 1.077,
      "step": 2389
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9279818259378454e-05,
      "loss": 1.1485,
      "step": 2390
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.927919218308046e-05,
      "loss": 0.9791,
      "step": 2391
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9278565838747403e-05,
      "loss": 1.0139,
      "step": 2392
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.927793922638621e-05,
      "loss": 1.1787,
      "step": 2393
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.927731234600379e-05,
      "loss": 1.0674,
      "step": 2394
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.927668519760706e-05,
      "loss": 1.0048,
      "step": 2395
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.927605778120296e-05,
      "loss": 1.0324,
      "step": 2396
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9275430096798405e-05,
      "loss": 0.9664,
      "step": 2397
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.927480214440032e-05,
      "loss": 1.0778,
      "step": 2398
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9274173924015634e-05,
      "loss": 1.0049,
      "step": 2399
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92735454356513e-05,
      "loss": 1.0383,
      "step": 2400
    },
    {
      "epoch": 0.08,
      "eval_loss": 0.9561118483543396,
      "eval_runtime": 389.5933,
      "eval_samples_per_second": 5.521,
      "eval_steps_per_second": 1.381,
      "step": 2400
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9272916679314255e-05,
      "loss": 0.7773,
      "step": 2401
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9272287655011415e-05,
      "loss": 0.4313,
      "step": 2402
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.927165836274975e-05,
      "loss": 0.6814,
      "step": 2403
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92710288025362e-05,
      "loss": 0.4338,
      "step": 2404
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9270398974377706e-05,
      "loss": 1.0527,
      "step": 2405
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.926976887828124e-05,
      "loss": 1.0705,
      "step": 2406
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9269138514253736e-05,
      "loss": 0.9072,
      "step": 2407
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9268507882302176e-05,
      "loss": 1.013,
      "step": 2408
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92678769824335e-05,
      "loss": 0.7728,
      "step": 2409
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9267245814654684e-05,
      "loss": 1.0266,
      "step": 2410
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92666143789727e-05,
      "loss": 1.002,
      "step": 2411
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9265982675394504e-05,
      "loss": 2.3593,
      "step": 2412
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9265350703927084e-05,
      "loss": 0.8666,
      "step": 2413
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.926471846457742e-05,
      "loss": 0.6043,
      "step": 2414
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.926408595735248e-05,
      "loss": 0.5586,
      "step": 2415
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.926345318225924e-05,
      "loss": 0.5885,
      "step": 2416
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.926282013930471e-05,
      "loss": 0.4616,
      "step": 2417
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.926218682849586e-05,
      "loss": 0.8601,
      "step": 2418
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.926155324983969e-05,
      "loss": 1.1257,
      "step": 2419
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.926091940334318e-05,
      "loss": 1.1344,
      "step": 2420
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9260285289013346e-05,
      "loss": 0.9858,
      "step": 2421
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.925965090685718e-05,
      "loss": 0.8112,
      "step": 2422
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.925901625688168e-05,
      "loss": 0.939,
      "step": 2423
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9258381339093876e-05,
      "loss": 0.9973,
      "step": 2424
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.925774615350075e-05,
      "loss": 1.0428,
      "step": 2425
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.925711070010932e-05,
      "loss": 0.9362,
      "step": 2426
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9256474978926617e-05,
      "loss": 1.0735,
      "step": 2427
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9255838989959634e-05,
      "loss": 0.9245,
      "step": 2428
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.925520273321541e-05,
      "loss": 0.9491,
      "step": 2429
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.925456620870097e-05,
      "loss": 0.6212,
      "step": 2430
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.925392941642333e-05,
      "loss": 0.8496,
      "step": 2431
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.925329235638953e-05,
      "loss": 0.9994,
      "step": 2432
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92526550286066e-05,
      "loss": 1.011,
      "step": 2433
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9252017433081565e-05,
      "loss": 0.9594,
      "step": 2434
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.925137956982149e-05,
      "loss": 0.9124,
      "step": 2435
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.925074143883339e-05,
      "loss": 1.0291,
      "step": 2436
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9250103040124316e-05,
      "loss": 1.1053,
      "step": 2437
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9249464373701326e-05,
      "loss": 0.9021,
      "step": 2438
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9248825439571465e-05,
      "loss": 0.9577,
      "step": 2439
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.924818623774178e-05,
      "loss": 0.9647,
      "step": 2440
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.924754676821934e-05,
      "loss": 0.9258,
      "step": 2441
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92469070310112e-05,
      "loss": 2.4956,
      "step": 2442
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.924626702612441e-05,
      "loss": 5.8455,
      "step": 2443
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9245626753566066e-05,
      "loss": 5.811,
      "step": 2444
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92449862133432e-05,
      "loss": 5.7834,
      "step": 2445
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9244345405462903e-05,
      "loss": 1.5272,
      "step": 2446
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.924370432993225e-05,
      "loss": 1.0311,
      "step": 2447
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9243062986758315e-05,
      "loss": 0.9503,
      "step": 2448
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.924242137594818e-05,
      "loss": 0.8954,
      "step": 2449
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9241779497508924e-05,
      "loss": 0.9598,
      "step": 2450
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.924113735144763e-05,
      "loss": 1.0326,
      "step": 2451
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92404949377714e-05,
      "loss": 1.0169,
      "step": 2452
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9239852256487314e-05,
      "loss": 1.033,
      "step": 2453
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9239209307602474e-05,
      "loss": 1.0125,
      "step": 2454
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9238566091123974e-05,
      "loss": 0.9556,
      "step": 2455
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9237922607058917e-05,
      "loss": 0.9559,
      "step": 2456
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92372788554144e-05,
      "loss": 0.7013,
      "step": 2457
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.923663483619755e-05,
      "loss": 0.5882,
      "step": 2458
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.923599054941545e-05,
      "loss": 0.6108,
      "step": 2459
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9235345995075234e-05,
      "loss": 0.5725,
      "step": 2460
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9234701173184005e-05,
      "loss": 0.8438,
      "step": 2461
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.923405608374889e-05,
      "loss": 0.8517,
      "step": 2462
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9233410726777e-05,
      "loss": 0.9438,
      "step": 2463
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9232765102275466e-05,
      "loss": 0.9748,
      "step": 2464
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9232119210251424e-05,
      "loss": 0.9623,
      "step": 2465
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.923147305071199e-05,
      "loss": 0.88,
      "step": 2466
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.923082662366431e-05,
      "loss": 1.066,
      "step": 2467
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9230179929115506e-05,
      "loss": 0.9964,
      "step": 2468
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.922953296707272e-05,
      "loss": 1.0706,
      "step": 2469
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9228885737543115e-05,
      "loss": 0.9608,
      "step": 2470
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.922823824053381e-05,
      "loss": 0.9583,
      "step": 2471
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9227590476051966e-05,
      "loss": 0.8946,
      "step": 2472
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.922694244410473e-05,
      "loss": 0.9593,
      "step": 2473
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9226294144699265e-05,
      "loss": 0.9786,
      "step": 2474
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9225645577842714e-05,
      "loss": 0.9443,
      "step": 2475
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.922499674354225e-05,
      "loss": 1.088,
      "step": 2476
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.922434764180503e-05,
      "loss": 0.9466,
      "step": 2477
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.922369827263821e-05,
      "loss": 0.9111,
      "step": 2478
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.922304863604898e-05,
      "loss": 0.8749,
      "step": 2479
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92223987320445e-05,
      "loss": 0.7827,
      "step": 2480
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.922174856063193e-05,
      "loss": 1.1009,
      "step": 2481
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.922109812181848e-05,
      "loss": 0.9536,
      "step": 2482
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92204474156113e-05,
      "loss": 0.8037,
      "step": 2483
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92197964420176e-05,
      "loss": 0.8275,
      "step": 2484
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9219145201044545e-05,
      "loss": 0.9661,
      "step": 2485
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.921849369269934e-05,
      "loss": 1.0123,
      "step": 2486
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.921784191698917e-05,
      "loss": 0.8871,
      "step": 2487
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.921718987392122e-05,
      "loss": 0.7357,
      "step": 2488
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9216537563502715e-05,
      "loss": 0.9076,
      "step": 2489
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9215884985740834e-05,
      "loss": 0.9674,
      "step": 2490
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9215232140642784e-05,
      "loss": 1.013,
      "step": 2491
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9214579028215776e-05,
      "loss": 0.9167,
      "step": 2492
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9213925648467027e-05,
      "loss": 1.153,
      "step": 2493
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9213272001403746e-05,
      "loss": 0.745,
      "step": 2494
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9212618087033136e-05,
      "loss": 1.0816,
      "step": 2495
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9211963905362436e-05,
      "loss": 1.0136,
      "step": 2496
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.921130945639886e-05,
      "loss": 0.8988,
      "step": 2497
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.921065474014962e-05,
      "loss": 0.9432,
      "step": 2498
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.920999975662196e-05,
      "loss": 4.5987,
      "step": 2499
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.920934450582311e-05,
      "loss": 5.8635,
      "step": 2500
    },
    {
      "epoch": 0.08,
      "eval_loss": 0.9598649144172668,
      "eval_runtime": 304.1004,
      "eval_samples_per_second": 7.073,
      "eval_steps_per_second": 1.769,
      "step": 2500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92086889877603e-05,
      "loss": 5.905,
      "step": 2501
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.920803320244076e-05,
      "loss": 5.7072,
      "step": 2502
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.920737714987175e-05,
      "loss": 5.861,
      "step": 2503
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9206720830060486e-05,
      "loss": 4.349,
      "step": 2504
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.920606424301423e-05,
      "loss": 1.0545,
      "step": 2505
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9205407388740236e-05,
      "loss": 1.1474,
      "step": 2506
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9204750267245736e-05,
      "loss": 1.0774,
      "step": 2507
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9204092878537996e-05,
      "loss": 0.9305,
      "step": 2508
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.920343522262427e-05,
      "loss": 1.0217,
      "step": 2509
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9202777299511826e-05,
      "loss": 0.9735,
      "step": 2510
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9202119109207924e-05,
      "loss": 1.0479,
      "step": 2511
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9201460651719825e-05,
      "loss": 0.826,
      "step": 2512
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.92008019270548e-05,
      "loss": 0.8114,
      "step": 2513
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.920014293522013e-05,
      "loss": 1.0093,
      "step": 2514
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919948367622307e-05,
      "loss": 1.1322,
      "step": 2515
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919882415007092e-05,
      "loss": 0.9508,
      "step": 2516
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919816435677095e-05,
      "loss": 0.9589,
      "step": 2517
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919750429633044e-05,
      "loss": 0.9126,
      "step": 2518
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919684396875669e-05,
      "loss": 0.9477,
      "step": 2519
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919618337405698e-05,
      "loss": 1.228,
      "step": 2520
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919552251223861e-05,
      "loss": 0.9089,
      "step": 2521
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919486138330886e-05,
      "loss": 0.9907,
      "step": 2522
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919419998727505e-05,
      "loss": 1.005,
      "step": 2523
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919353832414446e-05,
      "loss": 0.966,
      "step": 2524
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919287639392441e-05,
      "loss": 1.0129,
      "step": 2525
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.91922141966222e-05,
      "loss": 0.9512,
      "step": 2526
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9191551732245154e-05,
      "loss": 1.2365,
      "step": 2527
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919088900080057e-05,
      "loss": 1.0442,
      "step": 2528
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.919022600229577e-05,
      "loss": 1.0277,
      "step": 2529
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9189562736738073e-05,
      "loss": 0.9726,
      "step": 2530
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9188899204134794e-05,
      "loss": 0.9101,
      "step": 2531
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9188235404493266e-05,
      "loss": 0.9493,
      "step": 2532
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.918757133782082e-05,
      "loss": 1.1436,
      "step": 2533
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.918690700412478e-05,
      "loss": 1.005,
      "step": 2534
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9186242403412494e-05,
      "loss": 0.93,
      "step": 2535
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.918557753569127e-05,
      "loss": 0.8151,
      "step": 2536
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.918491240096849e-05,
      "loss": 1.0729,
      "step": 2537
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.918424699925145e-05,
      "loss": 0.9812,
      "step": 2538
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9183581330547536e-05,
      "loss": 0.8843,
      "step": 2539
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9182915394864074e-05,
      "loss": 0.7285,
      "step": 2540
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.918224919220842e-05,
      "loss": 0.7456,
      "step": 2541
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9181582722587925e-05,
      "loss": 0.7801,
      "step": 2542
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.918091598600996e-05,
      "loss": 2.9582,
      "step": 2543
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9180248982481876e-05,
      "loss": 0.9993,
      "step": 2544
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.917958171201104e-05,
      "loss": 0.7526,
      "step": 2545
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9178914174604806e-05,
      "loss": 1.0462,
      "step": 2546
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.917824637027056e-05,
      "loss": 0.9976,
      "step": 2547
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9177578299015676e-05,
      "loss": 0.6869,
      "step": 2548
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9176909960847516e-05,
      "loss": 0.8792,
      "step": 2549
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.917624135577346e-05,
      "loss": 0.9011,
      "step": 2550
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.91755724838009e-05,
      "loss": 1.0465,
      "step": 2551
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9174903344937204e-05,
      "loss": 0.872,
      "step": 2552
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.917423393918977e-05,
      "loss": 1.0849,
      "step": 2553
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9173564266566e-05,
      "loss": 0.7478,
      "step": 2554
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9172894327073257e-05,
      "loss": 0.9269,
      "step": 2555
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9172224120718955e-05,
      "loss": 1.0427,
      "step": 2556
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9171553647510494e-05,
      "loss": 1.0034,
      "step": 2557
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9170882907455275e-05,
      "loss": 0.9965,
      "step": 2558
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.91702119005607e-05,
      "loss": 0.8718,
      "step": 2559
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9169540626834176e-05,
      "loss": 0.9968,
      "step": 2560
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916886908628312e-05,
      "loss": 1.0188,
      "step": 2561
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916819727891493e-05,
      "loss": 0.9676,
      "step": 2562
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916752520473704e-05,
      "loss": 0.9233,
      "step": 2563
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916685286375686e-05,
      "loss": 0.8801,
      "step": 2564
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916618025598182e-05,
      "loss": 0.964,
      "step": 2565
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916550738141934e-05,
      "loss": 1.0554,
      "step": 2566
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916483424007684e-05,
      "loss": 1.058,
      "step": 2567
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916416083196177e-05,
      "loss": 0.806,
      "step": 2568
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916348715708155e-05,
      "loss": 0.9611,
      "step": 2569
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916281321544362e-05,
      "loss": 1.0365,
      "step": 2570
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916213900705543e-05,
      "loss": 0.8355,
      "step": 2571
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.916146453192441e-05,
      "loss": 1.0465,
      "step": 2572
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9160789790058e-05,
      "loss": 1.1252,
      "step": 2573
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9160114781463675e-05,
      "loss": 1.0887,
      "step": 2574
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.915943950614886e-05,
      "loss": 0.9327,
      "step": 2575
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.915876396412102e-05,
      "loss": 0.9894,
      "step": 2576
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9158088155387625e-05,
      "loss": 1.0502,
      "step": 2577
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.915741207995612e-05,
      "loss": 0.7435,
      "step": 2578
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.915673573783398e-05,
      "loss": 1.0309,
      "step": 2579
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.915605912902866e-05,
      "loss": 0.9805,
      "step": 2580
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9155382253547635e-05,
      "loss": 1.1644,
      "step": 2581
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.915470511139837e-05,
      "loss": 0.9826,
      "step": 2582
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9154027702588356e-05,
      "loss": 0.9707,
      "step": 2583
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9153350027125064e-05,
      "loss": 0.9655,
      "step": 2584
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.915267208501597e-05,
      "loss": 1.0911,
      "step": 2585
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.915199387626857e-05,
      "loss": 0.8644,
      "step": 2586
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.915131540089033e-05,
      "loss": 0.9419,
      "step": 2587
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.915063665888876e-05,
      "loss": 0.8748,
      "step": 2588
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.914995765027135e-05,
      "loss": 0.9282,
      "step": 2589
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9149278375045596e-05,
      "loss": 1.061,
      "step": 2590
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.914859883321899e-05,
      "loss": 0.8925,
      "step": 2591
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.914791902479904e-05,
      "loss": 0.982,
      "step": 2592
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.914723894979325e-05,
      "loss": 1.0666,
      "step": 2593
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.914655860820913e-05,
      "loss": 0.967,
      "step": 2594
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9145878000054176e-05,
      "loss": 0.9484,
      "step": 2595
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9145197125335916e-05,
      "loss": 1.0062,
      "step": 2596
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.914451598406187e-05,
      "loss": 0.8678,
      "step": 2597
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.914383457623955e-05,
      "loss": 1.0557,
      "step": 2598
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.914315290187648e-05,
      "loss": 1.0766,
      "step": 2599
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.914247096098019e-05,
      "loss": 0.9663,
      "step": 2600
    },
    {
      "epoch": 0.09,
      "eval_loss": 0.9582121968269348,
      "eval_runtime": 295.6759,
      "eval_samples_per_second": 7.275,
      "eval_steps_per_second": 1.82,
      "step": 2600
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.914178875355819e-05,
      "loss": 0.9373,
      "step": 2601
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9141106279618043e-05,
      "loss": 0.9734,
      "step": 2602
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.914042353916726e-05,
      "loss": 1.1168,
      "step": 2603
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9139740532213375e-05,
      "loss": 1.1564,
      "step": 2604
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9139057258763945e-05,
      "loss": 0.8477,
      "step": 2605
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9138373718826504e-05,
      "loss": 1.0675,
      "step": 2606
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.91376899124086e-05,
      "loss": 0.8982,
      "step": 2607
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.913700583951778e-05,
      "loss": 0.8794,
      "step": 2608
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9136321500161595e-05,
      "loss": 0.997,
      "step": 2609
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.913563689434761e-05,
      "loss": 0.9736,
      "step": 2610
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.913495202208337e-05,
      "loss": 1.0047,
      "step": 2611
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.913426688337644e-05,
      "loss": 0.9775,
      "step": 2612
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.913358147823438e-05,
      "loss": 1.0181,
      "step": 2613
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.913289580666477e-05,
      "loss": 0.992,
      "step": 2614
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.913220986867517e-05,
      "loss": 1.0158,
      "step": 2615
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9131523664273146e-05,
      "loss": 0.8749,
      "step": 2616
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.91308371934663e-05,
      "loss": 1.0414,
      "step": 2617
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.913015045626217e-05,
      "loss": 1.0357,
      "step": 2618
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.912946345266837e-05,
      "loss": 1.1486,
      "step": 2619
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.912877618269247e-05,
      "loss": 1.0326,
      "step": 2620
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.912808864634206e-05,
      "loss": 1.0663,
      "step": 2621
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9127400843624724e-05,
      "loss": 1.0678,
      "step": 2622
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9126712774548074e-05,
      "loss": 0.9435,
      "step": 2623
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.912602443911969e-05,
      "loss": 0.9484,
      "step": 2624
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.912533583734718e-05,
      "loss": 0.4583,
      "step": 2625
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9124646969238124e-05,
      "loss": 0.2564,
      "step": 2626
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.912395783480015e-05,
      "loss": 0.852,
      "step": 2627
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.912326843404087e-05,
      "loss": 0.8408,
      "step": 2628
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.912257876696787e-05,
      "loss": 1.4016,
      "step": 2629
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9121888833588795e-05,
      "loss": 0.9687,
      "step": 2630
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9121198633911226e-05,
      "loss": 0.5765,
      "step": 2631
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9120508167942815e-05,
      "loss": 0.6468,
      "step": 2632
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.911981743569116e-05,
      "loss": 0.8463,
      "step": 2633
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.911912643716391e-05,
      "loss": 1.0521,
      "step": 2634
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9118435172368673e-05,
      "loss": 0.8938,
      "step": 2635
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.911774364131308e-05,
      "loss": 0.9172,
      "step": 2636
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.911705184400479e-05,
      "loss": 1.0458,
      "step": 2637
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.911635978045142e-05,
      "loss": 1.0042,
      "step": 2638
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9115667450660604e-05,
      "loss": 1.0106,
      "step": 2639
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9114974854639996e-05,
      "loss": 0.8652,
      "step": 2640
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.911428199239725e-05,
      "loss": 0.9902,
      "step": 2641
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.911358886393999e-05,
      "loss": 0.9884,
      "step": 2642
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.91128954692759e-05,
      "loss": 0.4132,
      "step": 2643
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.911220180841262e-05,
      "loss": 0.9756,
      "step": 2644
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.911150788135779e-05,
      "loss": 0.8951,
      "step": 2645
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9110813688119103e-05,
      "loss": 0.8055,
      "step": 2646
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9110119228704196e-05,
      "loss": 1.1137,
      "step": 2647
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.910942450312075e-05,
      "loss": 1.0023,
      "step": 2648
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.910872951137644e-05,
      "loss": 1.0442,
      "step": 2649
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.910803425347892e-05,
      "loss": 0.9923,
      "step": 2650
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9107338729435874e-05,
      "loss": 1.1529,
      "step": 2651
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.910664293925499e-05,
      "loss": 1.0055,
      "step": 2652
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.910594688294393e-05,
      "loss": 0.9812,
      "step": 2653
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9105250560510394e-05,
      "loss": 0.9172,
      "step": 2654
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.910455397196206e-05,
      "loss": 0.9038,
      "step": 2655
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.910385711730663e-05,
      "loss": 0.8649,
      "step": 2656
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.910315999655179e-05,
      "loss": 0.985,
      "step": 2657
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.910246260970523e-05,
      "loss": 1.0602,
      "step": 2658
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.910176495677465e-05,
      "loss": 0.9417,
      "step": 2659
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.910106703776777e-05,
      "loss": 1.0375,
      "step": 2660
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9100368852692267e-05,
      "loss": 1.0813,
      "step": 2661
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9099670401555875e-05,
      "loss": 1.0171,
      "step": 2662
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9098971684366286e-05,
      "loss": 2.8443,
      "step": 2663
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.909827270113122e-05,
      "loss": 1.6206,
      "step": 2664
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9097573451858394e-05,
      "loss": 0.4938,
      "step": 2665
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9096873936555534e-05,
      "loss": 0.6585,
      "step": 2666
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.909617415523034e-05,
      "loss": 0.9408,
      "step": 2667
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9095474107890564e-05,
      "loss": 1.1709,
      "step": 2668
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9094773794543924e-05,
      "loss": 0.9116,
      "step": 2669
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.909407321519815e-05,
      "loss": 1.2001,
      "step": 2670
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.909337236986097e-05,
      "loss": 1.0079,
      "step": 2671
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.909267125854014e-05,
      "loss": 0.9377,
      "step": 2672
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.909196988124338e-05,
      "loss": 0.7623,
      "step": 2673
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.909126823797844e-05,
      "loss": 0.9777,
      "step": 2674
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.909056632875307e-05,
      "loss": 0.9684,
      "step": 2675
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9089864153575016e-05,
      "loss": 1.0375,
      "step": 2676
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.908916171245203e-05,
      "loss": 1.0597,
      "step": 2677
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.908845900539186e-05,
      "loss": 0.9558,
      "step": 2678
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.908775603240228e-05,
      "loss": 0.9814,
      "step": 2679
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.908705279349103e-05,
      "loss": 0.9219,
      "step": 2680
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9086349288665886e-05,
      "loss": 0.9103,
      "step": 2681
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.908564551793462e-05,
      "loss": 0.9575,
      "step": 2682
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9084941481304985e-05,
      "loss": 0.8387,
      "step": 2683
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.908423717878476e-05,
      "loss": 1.0132,
      "step": 2684
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.908353261038174e-05,
      "loss": 0.9967,
      "step": 2685
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.908282777610367e-05,
      "loss": 0.9603,
      "step": 2686
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9082122675958356e-05,
      "loss": 0.9915,
      "step": 2687
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.908141730995357e-05,
      "loss": 1.0666,
      "step": 2688
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.90807116780971e-05,
      "loss": 1.0244,
      "step": 2689
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.908000578039674e-05,
      "loss": 0.9645,
      "step": 2690
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.907929961686029e-05,
      "loss": 0.9001,
      "step": 2691
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.907859318749552e-05,
      "loss": 0.9721,
      "step": 2692
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9077886492310255e-05,
      "loss": 2.1558,
      "step": 2693
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9077179531312286e-05,
      "loss": 1.0912,
      "step": 2694
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.907647230450942e-05,
      "loss": 0.8582,
      "step": 2695
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.907576481190946e-05,
      "loss": 1.6804,
      "step": 2696
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.907505705352023e-05,
      "loss": 1.1748,
      "step": 2697
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.907434902934952e-05,
      "loss": 0.9509,
      "step": 2698
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9073640739405167e-05,
      "loss": 1.0591,
      "step": 2699
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.907293218369499e-05,
      "loss": 0.9739,
      "step": 2700
    },
    {
      "epoch": 0.09,
      "eval_loss": 0.959994912147522,
      "eval_runtime": 342.7316,
      "eval_samples_per_second": 6.276,
      "eval_steps_per_second": 1.57,
      "step": 2700
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9072223362226786e-05,
      "loss": 0.8416,
      "step": 2701
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9071514275008414e-05,
      "loss": 0.9024,
      "step": 2702
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9070804922047685e-05,
      "loss": 0.936,
      "step": 2703
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9070095303352425e-05,
      "loss": 0.8816,
      "step": 2704
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.906938541893048e-05,
      "loss": 0.9093,
      "step": 2705
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.906867526878968e-05,
      "loss": 1.0569,
      "step": 2706
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.906796485293787e-05,
      "loss": 0.8354,
      "step": 2707
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9067254171382884e-05,
      "loss": 1.5236,
      "step": 2708
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9066543224132575e-05,
      "loss": 1.1651,
      "step": 2709
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.906583201119479e-05,
      "loss": 0.9688,
      "step": 2710
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.906512053257739e-05,
      "loss": 1.1035,
      "step": 2711
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.906440878828821e-05,
      "loss": 0.9741,
      "step": 2712
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.906369677833511e-05,
      "loss": 0.9834,
      "step": 2713
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9062984502725964e-05,
      "loss": 1.1155,
      "step": 2714
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.906227196146863e-05,
      "loss": 0.9638,
      "step": 2715
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.906155915457098e-05,
      "loss": 1.0254,
      "step": 2716
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.906084608204087e-05,
      "loss": 1.0337,
      "step": 2717
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9060132743886177e-05,
      "loss": 0.9138,
      "step": 2718
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905941914011478e-05,
      "loss": 1.0409,
      "step": 2719
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905870527073455e-05,
      "loss": 0.9508,
      "step": 2720
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905799113575338e-05,
      "loss": 1.0561,
      "step": 2721
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905727673517914e-05,
      "loss": 1.0311,
      "step": 2722
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905656206901972e-05,
      "loss": 1.0053,
      "step": 2723
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9055847137283015e-05,
      "loss": 0.9773,
      "step": 2724
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905513193997692e-05,
      "loss": 0.8944,
      "step": 2725
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905441647710932e-05,
      "loss": 1.0638,
      "step": 2726
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905370074868813e-05,
      "loss": 0.9982,
      "step": 2727
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905298475472123e-05,
      "loss": 1.0842,
      "step": 2728
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9052268495216536e-05,
      "loss": 1.1171,
      "step": 2729
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905155197018196e-05,
      "loss": 0.9077,
      "step": 2730
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9050835179625395e-05,
      "loss": 1.0493,
      "step": 2731
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.905011812355478e-05,
      "loss": 0.9548,
      "step": 2732
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.904940080197801e-05,
      "loss": 1.369,
      "step": 2733
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.904868321490301e-05,
      "loss": 2.467,
      "step": 2734
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.904796536233771e-05,
      "loss": 1.8712,
      "step": 2735
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.904724724429002e-05,
      "loss": 1.4133,
      "step": 2736
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.904652886076788e-05,
      "loss": 1.0388,
      "step": 2737
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.904581021177922e-05,
      "loss": 1.1946,
      "step": 2738
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.904509129733196e-05,
      "loss": 1.0403,
      "step": 2739
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.904437211743405e-05,
      "loss": 1.0673,
      "step": 2740
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.904365267209343e-05,
      "loss": 1.0601,
      "step": 2741
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.904293296131803e-05,
      "loss": 0.9344,
      "step": 2742
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9042212985115813e-05,
      "loss": 1.066,
      "step": 2743
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.904149274349472e-05,
      "loss": 1.1525,
      "step": 2744
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9040772236462695e-05,
      "loss": 0.9562,
      "step": 2745
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.90400514640277e-05,
      "loss": 0.9808,
      "step": 2746
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9039330426197684e-05,
      "loss": 0.9094,
      "step": 2747
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.903860912298062e-05,
      "loss": 1.2482,
      "step": 2748
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.903788755438446e-05,
      "loss": 1.0086,
      "step": 2749
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.903716572041718e-05,
      "loss": 0.799,
      "step": 2750
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9036443621086734e-05,
      "loss": 1.1839,
      "step": 2751
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.90357212564011e-05,
      "loss": 1.0266,
      "step": 2752
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.903499862636826e-05,
      "loss": 0.7671,
      "step": 2753
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9034275730996194e-05,
      "loss": 1.0178,
      "step": 2754
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.903355257029286e-05,
      "loss": 1.0508,
      "step": 2755
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.903282914426627e-05,
      "loss": 1.1618,
      "step": 2756
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9032105452924395e-05,
      "loss": 1.073,
      "step": 2757
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.903138149627522e-05,
      "loss": 1.0024,
      "step": 2758
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.903065727432675e-05,
      "loss": 0.911,
      "step": 2759
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.902993278708697e-05,
      "loss": 1.0125,
      "step": 2760
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.902920803456388e-05,
      "loss": 1.063,
      "step": 2761
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.902848301676549e-05,
      "loss": 0.8709,
      "step": 2762
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.902775773369979e-05,
      "loss": 1.0413,
      "step": 2763
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.90270321853748e-05,
      "loss": 0.9238,
      "step": 2764
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9026306371798524e-05,
      "loss": 0.9058,
      "step": 2765
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9025580292978965e-05,
      "loss": 0.9979,
      "step": 2766
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.902485394892416e-05,
      "loss": 1.0335,
      "step": 2767
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.902412733964211e-05,
      "loss": 0.8472,
      "step": 2768
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.902340046514085e-05,
      "loss": 0.9879,
      "step": 2769
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.902267332542838e-05,
      "loss": 0.9487,
      "step": 2770
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.902194592051276e-05,
      "loss": 0.5531,
      "step": 2771
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9021218250401994e-05,
      "loss": 0.7712,
      "step": 2772
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.902049031510413e-05,
      "loss": 1.0398,
      "step": 2773
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9019762114627204e-05,
      "loss": 0.7579,
      "step": 2774
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.901903364897925e-05,
      "loss": 1.064,
      "step": 2775
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9018304918168315e-05,
      "loss": 0.8982,
      "step": 2776
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.901757592220243e-05,
      "loss": 0.9228,
      "step": 2777
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.901684666108965e-05,
      "loss": 0.954,
      "step": 2778
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.901611713483805e-05,
      "loss": 1.0003,
      "step": 2779
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.901538734345564e-05,
      "loss": 1.2969,
      "step": 2780
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.901465728695051e-05,
      "loss": 1.0127,
      "step": 2781
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.901392696533071e-05,
      "loss": 0.9745,
      "step": 2782
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9013196378604296e-05,
      "loss": 1.0129,
      "step": 2783
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.901246552677935e-05,
      "loss": 1.1789,
      "step": 2784
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.901173440986392e-05,
      "loss": 1.3628,
      "step": 2785
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.901100302786609e-05,
      "loss": 0.9565,
      "step": 2786
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.901027138079393e-05,
      "loss": 4.8619,
      "step": 2787
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9009539468655516e-05,
      "loss": 1.9135,
      "step": 2788
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9008807291458924e-05,
      "loss": 0.8999,
      "step": 2789
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.900807484921225e-05,
      "loss": 0.9483,
      "step": 2790
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.900734214192358e-05,
      "loss": 0.7361,
      "step": 2791
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.900660916960099e-05,
      "loss": 1.0791,
      "step": 2792
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9005875932252576e-05,
      "loss": 0.9563,
      "step": 2793
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.900514242988644e-05,
      "loss": 0.9595,
      "step": 2794
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.900440866251067e-05,
      "loss": 0.9773,
      "step": 2795
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9003674630133365e-05,
      "loss": 0.9868,
      "step": 2796
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.900294033276265e-05,
      "loss": 1.0727,
      "step": 2797
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.90022057704066e-05,
      "loss": 1.0108,
      "step": 2798
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9001470943073344e-05,
      "loss": 1.2056,
      "step": 2799
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9000735850770994e-05,
      "loss": 1.2509,
      "step": 2800
    },
    {
      "epoch": 0.09,
      "eval_loss": 0.9518626928329468,
      "eval_runtime": 356.764,
      "eval_samples_per_second": 6.029,
      "eval_steps_per_second": 1.508,
      "step": 2800
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.900000049350766e-05,
      "loss": 1.8124,
      "step": 2801
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.899926487129146e-05,
      "loss": 1.15,
      "step": 2802
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.899852898413052e-05,
      "loss": 0.9514,
      "step": 2803
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.899779283203296e-05,
      "loss": 0.1051,
      "step": 2804
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.89970564150069e-05,
      "loss": 0.8301,
      "step": 2805
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.899631973306049e-05,
      "loss": 0.9707,
      "step": 2806
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8995582786201846e-05,
      "loss": 1.0009,
      "step": 2807
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.899484557443911e-05,
      "loss": 0.7373,
      "step": 2808
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8994108097780414e-05,
      "loss": 0.9713,
      "step": 2809
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8993370356233904e-05,
      "loss": 0.8911,
      "step": 2810
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8992632349807724e-05,
      "loss": 0.7419,
      "step": 2811
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.899189407851001e-05,
      "loss": 0.5803,
      "step": 2812
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8991155542348945e-05,
      "loss": 1.0212,
      "step": 2813
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.899041674133265e-05,
      "loss": 1.1003,
      "step": 2814
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.89896776754693e-05,
      "loss": 1.0439,
      "step": 2815
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.898893834476704e-05,
      "loss": 0.8091,
      "step": 2816
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8988198749234036e-05,
      "loss": 0.9955,
      "step": 2817
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8987458888878465e-05,
      "loss": 0.8225,
      "step": 2818
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.898671876370847e-05,
      "loss": 0.7509,
      "step": 2819
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8985978373732256e-05,
      "loss": 1.0264,
      "step": 2820
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.898523771895797e-05,
      "loss": 1.0387,
      "step": 2821
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.898449679939379e-05,
      "loss": 0.8835,
      "step": 2822
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.89837556150479e-05,
      "loss": 0.9787,
      "step": 2823
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8983014165928496e-05,
      "loss": 0.9666,
      "step": 2824
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.898227245204374e-05,
      "loss": 1.092,
      "step": 2825
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.898153047340184e-05,
      "loss": 1.0636,
      "step": 2826
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8980788230010974e-05,
      "loss": 1.028,
      "step": 2827
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8980045721879355e-05,
      "loss": 1.3796,
      "step": 2828
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.897930294901515e-05,
      "loss": 0.897,
      "step": 2829
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.897855991142658e-05,
      "loss": 0.9117,
      "step": 2830
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8977816609121844e-05,
      "loss": 1.1802,
      "step": 2831
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8977073042109154e-05,
      "loss": 1.3865,
      "step": 2832
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8976329210396696e-05,
      "loss": 0.9837,
      "step": 2833
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8975585113992716e-05,
      "loss": 1.0674,
      "step": 2834
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.89748407529054e-05,
      "loss": 1.003,
      "step": 2835
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.897409612714298e-05,
      "loss": 0.897,
      "step": 2836
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.897335123671366e-05,
      "loss": 0.9187,
      "step": 2837
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.897260608162569e-05,
      "loss": 0.9696,
      "step": 2838
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8971860661887274e-05,
      "loss": 1.1332,
      "step": 2839
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.897111497750665e-05,
      "loss": 1.0664,
      "step": 2840
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.897036902849205e-05,
      "loss": 0.9785,
      "step": 2841
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.896962281485171e-05,
      "loss": 0.9637,
      "step": 2842
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8968876336593864e-05,
      "loss": 0.8586,
      "step": 2843
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.896812959372676e-05,
      "loss": 0.935,
      "step": 2844
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.896738258625863e-05,
      "loss": 1.0262,
      "step": 2845
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8966635314197726e-05,
      "loss": 1.1293,
      "step": 2846
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.896588777755231e-05,
      "loss": 1.0169,
      "step": 2847
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.896513997633061e-05,
      "loss": 1.2171,
      "step": 2848
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.8964391910540906e-05,
      "loss": 1.0414,
      "step": 2849
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8963643580191444e-05,
      "loss": 1.0897,
      "step": 2850
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8962894985290485e-05,
      "loss": 0.9843,
      "step": 2851
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8962146125846296e-05,
      "loss": 1.0511,
      "step": 2852
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8961397001867146e-05,
      "loss": 0.9074,
      "step": 2853
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.89606476133613e-05,
      "loss": 1.0395,
      "step": 2854
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.895989796033703e-05,
      "loss": 1.0365,
      "step": 2855
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.895914804280262e-05,
      "loss": 1.039,
      "step": 2856
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.895839786076635e-05,
      "loss": 0.8977,
      "step": 2857
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8957647414236485e-05,
      "loss": 1.0267,
      "step": 2858
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.895689670322133e-05,
      "loss": 2.9909,
      "step": 2859
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8956145727729156e-05,
      "loss": 1.0741,
      "step": 2860
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.895539448776827e-05,
      "loss": 0.9618,
      "step": 2861
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.895464298334695e-05,
      "loss": 0.9038,
      "step": 2862
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.89538912144735e-05,
      "loss": 0.8919,
      "step": 2863
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8953139181156214e-05,
      "loss": 1.0851,
      "step": 2864
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8952386883403415e-05,
      "loss": 1.0782,
      "step": 2865
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8951634321223375e-05,
      "loss": 1.016,
      "step": 2866
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.895088149462442e-05,
      "loss": 1.0573,
      "step": 2867
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.895012840361487e-05,
      "loss": 1.091,
      "step": 2868
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.894937504820302e-05,
      "loss": 1.1905,
      "step": 2869
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.89486214283972e-05,
      "loss": 1.0128,
      "step": 2870
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.894786754420572e-05,
      "loss": 0.9368,
      "step": 2871
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.894711339563691e-05,
      "loss": 1.0165,
      "step": 2872
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8946358982699105e-05,
      "loss": 1.1656,
      "step": 2873
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.894560430540061e-05,
      "loss": 1.065,
      "step": 2874
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.894484936374977e-05,
      "loss": 1.1771,
      "step": 2875
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8944094157754914e-05,
      "loss": 0.715,
      "step": 2876
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.894333868742439e-05,
      "loss": 1.0453,
      "step": 2877
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8942582952766524e-05,
      "loss": 0.9771,
      "step": 2878
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.894182695378967e-05,
      "loss": 1.1164,
      "step": 2879
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.894107069050217e-05,
      "loss": 1.0473,
      "step": 2880
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8940314162912376e-05,
      "loss": 0.9582,
      "step": 2881
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.893955737102863e-05,
      "loss": 0.8336,
      "step": 2882
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8938800314859304e-05,
      "loss": 0.8209,
      "step": 2883
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8938042994412736e-05,
      "loss": 1.0115,
      "step": 2884
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8937285409697296e-05,
      "loss": 0.9208,
      "step": 2885
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.893652756072136e-05,
      "loss": 1.0239,
      "step": 2886
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.893576944749327e-05,
      "loss": 1.0049,
      "step": 2887
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.893501107002141e-05,
      "loss": 1.0585,
      "step": 2888
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.893425242831415e-05,
      "loss": 1.1467,
      "step": 2889
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.893349352237986e-05,
      "loss": 0.9894,
      "step": 2890
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.893273435222693e-05,
      "loss": 1.1529,
      "step": 2891
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8931974917863734e-05,
      "loss": 0.9917,
      "step": 2892
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.893121521929865e-05,
      "loss": 1.0499,
      "step": 2893
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8930455256540075e-05,
      "loss": 0.5153,
      "step": 2894
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.892969502959639e-05,
      "loss": 0.9172,
      "step": 2895
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.892893453847599e-05,
      "loss": 0.9419,
      "step": 2896
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.892817378318728e-05,
      "loss": 1.0247,
      "step": 2897
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.892741276373865e-05,
      "loss": 0.9924,
      "step": 2898
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8926651480138505e-05,
      "loss": 0.6484,
      "step": 2899
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.892588993239524e-05,
      "loss": 1.0148,
      "step": 2900
    },
    {
      "epoch": 0.1,
      "eval_loss": 0.9518806338310242,
      "eval_runtime": 275.533,
      "eval_samples_per_second": 7.807,
      "eval_steps_per_second": 1.953,
      "step": 2900
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.892512812051727e-05,
      "loss": 3.7687,
      "step": 2901
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.892436604451301e-05,
      "loss": 0.9722,
      "step": 2902
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8923603704390866e-05,
      "loss": 1.1693,
      "step": 2903
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.892284110015925e-05,
      "loss": 1.0848,
      "step": 2904
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8922078231826585e-05,
      "loss": 0.8848,
      "step": 2905
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.89213150994013e-05,
      "loss": 0.9867,
      "step": 2906
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.892055170289181e-05,
      "loss": 0.9902,
      "step": 2907
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.891978804230656e-05,
      "loss": 0.9371,
      "step": 2908
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8919024117653957e-05,
      "loss": 0.7351,
      "step": 2909
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.891825992894244e-05,
      "loss": 0.8046,
      "step": 2910
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8917495476180456e-05,
      "loss": 0.9321,
      "step": 2911
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8916730759376436e-05,
      "loss": 0.805,
      "step": 2912
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.891596577853882e-05,
      "loss": 0.9318,
      "step": 2913
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8915200533676065e-05,
      "loss": 1.0743,
      "step": 2914
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8914435024796615e-05,
      "loss": 0.978,
      "step": 2915
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8913669251908914e-05,
      "loss": 1.1294,
      "step": 2916
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.891290321502141e-05,
      "loss": 1.013,
      "step": 2917
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.891213691414259e-05,
      "loss": 0.864,
      "step": 2918
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8911370349280885e-05,
      "loss": 0.9663,
      "step": 2919
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.891060352044475e-05,
      "loss": 1.0256,
      "step": 2920
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8909836427642685e-05,
      "loss": 0.9475,
      "step": 2921
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.890906907088314e-05,
      "loss": 1.0395,
      "step": 2922
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8908301450174576e-05,
      "loss": 1.0467,
      "step": 2923
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.890753356552549e-05,
      "loss": 0.9309,
      "step": 2924
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.890676541694433e-05,
      "loss": 0.9791,
      "step": 2925
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8905997004439605e-05,
      "loss": 0.9949,
      "step": 2926
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8905228328019794e-05,
      "loss": 1.0772,
      "step": 2927
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8904459387693363e-05,
      "loss": 1.069,
      "step": 2928
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.890369018346882e-05,
      "loss": 1.068,
      "step": 2929
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.890292071535464e-05,
      "loss": 0.7019,
      "step": 2930
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.890215098335933e-05,
      "loss": 0.1411,
      "step": 2931
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.89013809874914e-05,
      "loss": 0.083,
      "step": 2932
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8900610727759315e-05,
      "loss": 0.643,
      "step": 2933
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889984020417161e-05,
      "loss": 0.6037,
      "step": 2934
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889906941673678e-05,
      "loss": 1.0056,
      "step": 2935
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889829836546333e-05,
      "loss": 0.6609,
      "step": 2936
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889752705035978e-05,
      "loss": 0.3879,
      "step": 2937
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889675547143464e-05,
      "loss": 1.0452,
      "step": 2938
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889598362869644e-05,
      "loss": 1.1164,
      "step": 2939
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889521152215368e-05,
      "loss": 1.2395,
      "step": 2940
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.88944391518149e-05,
      "loss": 1.0476,
      "step": 2941
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889366651768861e-05,
      "loss": 1.1329,
      "step": 2942
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889289361978337e-05,
      "loss": 1.0124,
      "step": 2943
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889212045810767e-05,
      "loss": 1.0453,
      "step": 2944
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889134703267009e-05,
      "loss": 0.977,
      "step": 2945
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.889057334347914e-05,
      "loss": 0.9685,
      "step": 2946
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.888979939054337e-05,
      "loss": 0.8687,
      "step": 2947
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8889025173871314e-05,
      "loss": 0.874,
      "step": 2948
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8888250693471546e-05,
      "loss": 2.1459,
      "step": 2949
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.888747594935258e-05,
      "loss": 0.7649,
      "step": 2950
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8886700941523004e-05,
      "loss": 1.0229,
      "step": 2951
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8885925669991346e-05,
      "loss": 1.0047,
      "step": 2952
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8885150134766175e-05,
      "loss": 0.9281,
      "step": 2953
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8884374335856056e-05,
      "loss": 1.1979,
      "step": 2954
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.888359827326955e-05,
      "loss": 1.8173,
      "step": 2955
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8882821947015235e-05,
      "loss": 1.0909,
      "step": 2956
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.888204535710166e-05,
      "loss": 1.0165,
      "step": 2957
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.888126850353741e-05,
      "loss": 0.817,
      "step": 2958
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8880491386331066e-05,
      "loss": 1.4873,
      "step": 2959
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.88797140054912e-05,
      "loss": 1.4652,
      "step": 2960
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8878936361026403e-05,
      "loss": 1.0329,
      "step": 2961
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.887815845294525e-05,
      "loss": 0.8664,
      "step": 2962
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8877380281256334e-05,
      "loss": 0.8827,
      "step": 2963
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.887660184596824e-05,
      "loss": 0.7672,
      "step": 2964
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8875823147089575e-05,
      "loss": 0.9559,
      "step": 2965
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.887504418462891e-05,
      "loss": 0.9456,
      "step": 2966
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.887426495859487e-05,
      "loss": 0.7191,
      "step": 2967
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.887348546899605e-05,
      "loss": 1.0134,
      "step": 2968
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.887270571584106e-05,
      "loss": 0.9544,
      "step": 2969
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.887192569913849e-05,
      "loss": 1.0154,
      "step": 2970
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.887114541889697e-05,
      "loss": 0.615,
      "step": 2971
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.887036487512511e-05,
      "loss": 1.151,
      "step": 2972
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8869584067831517e-05,
      "loss": 0.9686,
      "step": 2973
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.886880299702482e-05,
      "loss": 0.6173,
      "step": 2974
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.886802166271364e-05,
      "loss": 0.8768,
      "step": 2975
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8867240064906605e-05,
      "loss": 1.0324,
      "step": 2976
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8866458203612344e-05,
      "loss": 1.134,
      "step": 2977
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8865676078839485e-05,
      "loss": 0.8039,
      "step": 2978
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8864893690596656e-05,
      "loss": 0.6354,
      "step": 2979
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.88641110388925e-05,
      "loss": 0.7907,
      "step": 2980
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.886332812373567e-05,
      "loss": 0.9553,
      "step": 2981
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8862544945134794e-05,
      "loss": 1.0297,
      "step": 2982
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.886176150309852e-05,
      "loss": 1.0379,
      "step": 2983
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.88609777976355e-05,
      "loss": 1.4562,
      "step": 2984
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.886019382875439e-05,
      "loss": 0.873,
      "step": 2985
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.885940959646382e-05,
      "loss": 0.976,
      "step": 2986
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.885862510077249e-05,
      "loss": 1.0246,
      "step": 2987
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8857840341689026e-05,
      "loss": 0.9837,
      "step": 2988
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.885705531922211e-05,
      "loss": 1.2443,
      "step": 2989
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.885627003338039e-05,
      "loss": 0.9216,
      "step": 2990
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.885548448417256e-05,
      "loss": 0.9786,
      "step": 2991
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8854698671607273e-05,
      "loss": 0.9358,
      "step": 2992
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8853912595693216e-05,
      "loss": 0.947,
      "step": 2993
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8853126256439055e-05,
      "loss": 0.9391,
      "step": 2994
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8852339653853484e-05,
      "loss": 0.9647,
      "step": 2995
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8851552787945176e-05,
      "loss": 0.963,
      "step": 2996
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.885076565872282e-05,
      "loss": 0.9398,
      "step": 2997
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8849978266195114e-05,
      "loss": 1.098,
      "step": 2998
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8849190610370735e-05,
      "loss": 2.3977,
      "step": 2999
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.88484026912584e-05,
      "loss": 0.5647,
      "step": 3000
    },
    {
      "epoch": 0.1,
      "eval_loss": 0.9596259593963623,
      "eval_runtime": 279.1952,
      "eval_samples_per_second": 7.704,
      "eval_steps_per_second": 1.927,
      "step": 3000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.884761450886679e-05,
      "loss": 0.5331,
      "step": 3001
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.884682606320461e-05,
      "loss": 1.0445,
      "step": 3002
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8846037354280564e-05,
      "loss": 1.014,
      "step": 3003
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.884524838210336e-05,
      "loss": 1.0323,
      "step": 3004
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8844459146681716e-05,
      "loss": 0.8803,
      "step": 3005
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8843669648024335e-05,
      "loss": 0.599,
      "step": 3006
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.884287988613994e-05,
      "loss": 0.7,
      "step": 3007
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8842089861037235e-05,
      "loss": 0.9111,
      "step": 3008
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.884129957272495e-05,
      "loss": 0.8716,
      "step": 3009
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8840509021211825e-05,
      "loss": 0.8915,
      "step": 3010
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.883971820650657e-05,
      "loss": 0.9289,
      "step": 3011
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.883892712861791e-05,
      "loss": 1.1084,
      "step": 3012
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.88381357875546e-05,
      "loss": 1.0249,
      "step": 3013
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8837344183325354e-05,
      "loss": 0.8415,
      "step": 3014
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.883655231593893e-05,
      "loss": 1.0192,
      "step": 3015
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8835760185404055e-05,
      "loss": 1.0892,
      "step": 3016
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.883496779172948e-05,
      "loss": 1.0686,
      "step": 3017
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.883417513492395e-05,
      "loss": 0.9694,
      "step": 3018
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.883338221499622e-05,
      "loss": 0.9282,
      "step": 3019
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8832589031955044e-05,
      "loss": 1.1761,
      "step": 3020
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.883179558580918e-05,
      "loss": 0.933,
      "step": 3021
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.883100187656738e-05,
      "loss": 0.9077,
      "step": 3022
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.883020790423841e-05,
      "loss": 1.1395,
      "step": 3023
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.882941366883104e-05,
      "loss": 1.0002,
      "step": 3024
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.882861917035402e-05,
      "loss": 0.9282,
      "step": 3025
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.882782440881615e-05,
      "loss": 0.9403,
      "step": 3026
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.882702938422618e-05,
      "loss": 0.9653,
      "step": 3027
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8826234096592896e-05,
      "loss": 1.217,
      "step": 3028
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.882543854592507e-05,
      "loss": 1.0496,
      "step": 3029
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8824642732231504e-05,
      "loss": 0.8795,
      "step": 3030
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8823846655520964e-05,
      "loss": 0.8266,
      "step": 3031
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.882305031580225e-05,
      "loss": 0.9399,
      "step": 3032
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8822253713084146e-05,
      "loss": 1.0232,
      "step": 3033
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.882145684737545e-05,
      "loss": 0.978,
      "step": 3034
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.882065971868496e-05,
      "loss": 1.0773,
      "step": 3035
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.881986232702147e-05,
      "loss": 1.0183,
      "step": 3036
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.881906467239379e-05,
      "loss": 1.0213,
      "step": 3037
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8818266754810724e-05,
      "loss": 1.1437,
      "step": 3038
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.881746857428108e-05,
      "loss": 1.0626,
      "step": 3039
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.881667013081367e-05,
      "loss": 1.2335,
      "step": 3040
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8815871424417305e-05,
      "loss": 1.2153,
      "step": 3041
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.881507245510081e-05,
      "loss": 1.0027,
      "step": 3042
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8814273222873e-05,
      "loss": 1.9011,
      "step": 3043
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.88134737277427e-05,
      "loss": 5.3018,
      "step": 3044
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.881267396971874e-05,
      "loss": 5.387,
      "step": 3045
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8811873948809925e-05,
      "loss": 5.684,
      "step": 3046
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8811073665025134e-05,
      "loss": 5.6865,
      "step": 3047
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.881027311837315e-05,
      "loss": 5.6722,
      "step": 3048
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8809472308862844e-05,
      "loss": 5.6907,
      "step": 3049
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.880867123650306e-05,
      "loss": 5.5938,
      "step": 3050
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.880786990130262e-05,
      "loss": 5.5617,
      "step": 3051
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.880706830327038e-05,
      "loss": 2.7421,
      "step": 3052
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.880626644241519e-05,
      "loss": 1.3304,
      "step": 3053
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.88054643187459e-05,
      "loss": 1.201,
      "step": 3054
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.880466193227137e-05,
      "loss": 1.0622,
      "step": 3055
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.880385928300045e-05,
      "loss": 0.6808,
      "step": 3056
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.880305637094201e-05,
      "loss": 0.748,
      "step": 3057
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.880225319610492e-05,
      "loss": 0.8301,
      "step": 3058
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8801449758498025e-05,
      "loss": 0.4514,
      "step": 3059
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8800646058130205e-05,
      "loss": 0.8556,
      "step": 3060
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.879984209501034e-05,
      "loss": 1.0026,
      "step": 3061
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.87990378691473e-05,
      "loss": 1.0478,
      "step": 3062
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.879823338054996e-05,
      "loss": 1.1326,
      "step": 3063
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.879742862922721e-05,
      "loss": 0.9685,
      "step": 3064
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.879662361518792e-05,
      "loss": 0.9911,
      "step": 3065
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.879581833844099e-05,
      "loss": 0.6458,
      "step": 3066
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.879501279899531e-05,
      "loss": 1.3757,
      "step": 3067
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.879420699685977e-05,
      "loss": 1.1768,
      "step": 3068
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8793400932043254e-05,
      "loss": 1.103,
      "step": 3069
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8792594604554683e-05,
      "loss": 0.8821,
      "step": 3070
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.879178801440294e-05,
      "loss": 0.7674,
      "step": 3071
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.879098116159694e-05,
      "loss": 1.0102,
      "step": 3072
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8790174046145584e-05,
      "loss": 1.0156,
      "step": 3073
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8789366668057783e-05,
      "loss": 1.0933,
      "step": 3074
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.878855902734246e-05,
      "loss": 0.9803,
      "step": 3075
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.878775112400852e-05,
      "loss": 1.2043,
      "step": 3076
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.878694295806488e-05,
      "loss": 0.9959,
      "step": 3077
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.878613452952047e-05,
      "loss": 0.9358,
      "step": 3078
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.878532583838421e-05,
      "loss": 0.9406,
      "step": 3079
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.878451688466504e-05,
      "loss": 1.0059,
      "step": 3080
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.878370766837187e-05,
      "loss": 1.0644,
      "step": 3081
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8782898189513645e-05,
      "loss": 0.6945,
      "step": 3082
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.87820884480993e-05,
      "loss": 0.8971,
      "step": 3083
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.878127844413779e-05,
      "loss": 1.3572,
      "step": 3084
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8780468177638026e-05,
      "loss": 1.5267,
      "step": 3085
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.877965764860898e-05,
      "loss": 1.4058,
      "step": 3086
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.877884685705958e-05,
      "loss": 1.5995,
      "step": 3087
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.877803580299879e-05,
      "loss": 1.5543,
      "step": 3088
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8777224486435555e-05,
      "loss": 1.5801,
      "step": 3089
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.877641290737884e-05,
      "loss": 1.6302,
      "step": 3090
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8775601065837605e-05,
      "loss": 1.6325,
      "step": 3091
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8774788961820806e-05,
      "loss": 1.6303,
      "step": 3092
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8773976595337404e-05,
      "loss": 1.6324,
      "step": 3093
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8773163966396384e-05,
      "loss": 1.5052,
      "step": 3094
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.87723510750067e-05,
      "loss": 1.0043,
      "step": 3095
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.877153792117733e-05,
      "loss": 0.9699,
      "step": 3096
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8770724504917264e-05,
      "loss": 1.0752,
      "step": 3097
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.876991082623548e-05,
      "loss": 0.9081,
      "step": 3098
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8769096885140935e-05,
      "loss": 0.8889,
      "step": 3099
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.876828268164264e-05,
      "loss": 0.7362,
      "step": 3100
    },
    {
      "epoch": 0.1,
      "eval_loss": 0.9801816344261169,
      "eval_runtime": 278.0036,
      "eval_samples_per_second": 7.737,
      "eval_steps_per_second": 1.935,
      "step": 3100
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.876746821574958e-05,
      "loss": 0.7684,
      "step": 3101
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8766653487470734e-05,
      "loss": 0.9924,
      "step": 3102
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.876583849681511e-05,
      "loss": 0.8606,
      "step": 3103
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.876502324379171e-05,
      "loss": 0.8404,
      "step": 3104
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.876420772840951e-05,
      "loss": 0.9104,
      "step": 3105
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.876339195067753e-05,
      "loss": 0.8051,
      "step": 3106
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.876257591060478e-05,
      "loss": 1.0423,
      "step": 3107
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.876175960820026e-05,
      "loss": 1.279,
      "step": 3108
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8760943043472984e-05,
      "loss": 1.115,
      "step": 3109
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.876012621643196e-05,
      "loss": 0.9335,
      "step": 3110
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.875930912708622e-05,
      "loss": 0.7042,
      "step": 3111
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.875849177544478e-05,
      "loss": 1.1014,
      "step": 3112
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.875767416151665e-05,
      "loss": 0.8891,
      "step": 3113
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.875685628531087e-05,
      "loss": 0.2902,
      "step": 3114
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.875603814683647e-05,
      "loss": 0.3228,
      "step": 3115
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.875521974610247e-05,
      "loss": 0.8094,
      "step": 3116
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.875440108311792e-05,
      "loss": 1.553,
      "step": 3117
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8753582157891845e-05,
      "loss": 1.0637,
      "step": 3118
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.875276297043328e-05,
      "loss": 0.8906,
      "step": 3119
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8751943520751296e-05,
      "loss": 0.3557,
      "step": 3120
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.875112380885492e-05,
      "loss": 0.983,
      "step": 3121
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.87503038347532e-05,
      "loss": 0.9097,
      "step": 3122
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8749483598455196e-05,
      "loss": 0.9836,
      "step": 3123
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.874866309996996e-05,
      "loss": 1.0989,
      "step": 3124
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.874784233930655e-05,
      "loss": 0.9592,
      "step": 3125
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.874702131647402e-05,
      "loss": 0.9129,
      "step": 3126
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.874620003148145e-05,
      "loss": 1.1831,
      "step": 3127
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.874537848433789e-05,
      "loss": 1.1603,
      "step": 3128
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8744556675052415e-05,
      "loss": 1.0468,
      "step": 3129
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8743734603634105e-05,
      "loss": 1.0109,
      "step": 3130
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.874291227009203e-05,
      "loss": 1.6028,
      "step": 3131
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8742089674435265e-05,
      "loss": 1.2437,
      "step": 3132
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8741266816672895e-05,
      "loss": 0.7824,
      "step": 3133
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8740443696814e-05,
      "loss": 0.9896,
      "step": 3134
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.873962031486768e-05,
      "loss": 1.0794,
      "step": 3135
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8738796670843004e-05,
      "loss": 0.9633,
      "step": 3136
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.873797276474909e-05,
      "loss": 1.1689,
      "step": 3137
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.873714859659501e-05,
      "loss": 1.0214,
      "step": 3138
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8736324166389867e-05,
      "loss": 1.0058,
      "step": 3139
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.873549947414278e-05,
      "loss": 0.9926,
      "step": 3140
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8734674519862834e-05,
      "loss": 0.9459,
      "step": 3141
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8733849303559144e-05,
      "loss": 1.0475,
      "step": 3142
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.873302382524082e-05,
      "loss": 0.9172,
      "step": 3143
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.873219808491698e-05,
      "loss": 0.9274,
      "step": 3144
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8731372082596726e-05,
      "loss": 1.0315,
      "step": 3145
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.873054581828919e-05,
      "loss": 0.3997,
      "step": 3146
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.872971929200349e-05,
      "loss": 0.5968,
      "step": 3147
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.872889250374875e-05,
      "loss": 1.0488,
      "step": 3148
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.872806545353409e-05,
      "loss": 1.022,
      "step": 3149
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.872723814136866e-05,
      "loss": 1.0085,
      "step": 3150
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.872641056726157e-05,
      "loss": 0.9722,
      "step": 3151
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.872558273122197e-05,
      "loss": 0.858,
      "step": 3152
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8724754633258995e-05,
      "loss": 0.849,
      "step": 3153
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8723926273381795e-05,
      "loss": 0.8513,
      "step": 3154
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8723097651599495e-05,
      "loss": 1.1669,
      "step": 3155
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.872226876792126e-05,
      "loss": 1.2488,
      "step": 3156
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.872143962235625e-05,
      "loss": 1.0299,
      "step": 3157
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.872061021491359e-05,
      "loss": 0.9572,
      "step": 3158
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8719780545602455e-05,
      "loss": 0.8786,
      "step": 3159
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8718950614432e-05,
      "loss": 1.0116,
      "step": 3160
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.871812042141139e-05,
      "loss": 0.9905,
      "step": 3161
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.871728996654978e-05,
      "loss": 1.5771,
      "step": 3162
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8716459249856364e-05,
      "loss": 1.013,
      "step": 3163
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.871562827134028e-05,
      "loss": 0.8689,
      "step": 3164
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.871479703101072e-05,
      "loss": 0.9901,
      "step": 3165
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.871396552887685e-05,
      "loss": 0.9787,
      "step": 3166
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.871313376494786e-05,
      "loss": 0.9969,
      "step": 3167
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8712301739232935e-05,
      "loss": 1.0084,
      "step": 3168
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8711469451741256e-05,
      "loss": 1.121,
      "step": 3169
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8710636902482e-05,
      "loss": 1.136,
      "step": 3170
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8709804091464376e-05,
      "loss": 0.9513,
      "step": 3171
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.870897101869756e-05,
      "loss": 0.7103,
      "step": 3172
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8708137684190765e-05,
      "loss": 1.0275,
      "step": 3173
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.870730408795318e-05,
      "loss": 1.6434,
      "step": 3174
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.870647022999402e-05,
      "loss": 1.4933,
      "step": 3175
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.870563611032247e-05,
      "loss": 1.4081,
      "step": 3176
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.870480172894776e-05,
      "loss": 1.537,
      "step": 3177
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8703967085879095e-05,
      "loss": 1.5473,
      "step": 3178
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8703132181125675e-05,
      "loss": 1.6367,
      "step": 3179
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.870229701469674e-05,
      "loss": 1.6261,
      "step": 3180
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8701461586601495e-05,
      "loss": 1.537,
      "step": 3181
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.870062589684916e-05,
      "loss": 1.5504,
      "step": 3182
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.869978994544897e-05,
      "loss": 1.5253,
      "step": 3183
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.869895373241016e-05,
      "loss": 1.4842,
      "step": 3184
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8698117257741935e-05,
      "loss": 1.0255,
      "step": 3185
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8697280521453565e-05,
      "loss": 0.9562,
      "step": 3186
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.869644352355426e-05,
      "loss": 1.0119,
      "step": 3187
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.869560626405327e-05,
      "loss": 1.0071,
      "step": 3188
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.869476874295983e-05,
      "loss": 1.1964,
      "step": 3189
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8693930960283204e-05,
      "loss": 1.3891,
      "step": 3190
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.869309291603263e-05,
      "loss": 1.0448,
      "step": 3191
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.869225461021736e-05,
      "loss": 1.8154,
      "step": 3192
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8691416042846636e-05,
      "loss": 1.6862,
      "step": 3193
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8690577213929746e-05,
      "loss": 0.8987,
      "step": 3194
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.868973812347592e-05,
      "loss": 1.0384,
      "step": 3195
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.868889877149444e-05,
      "loss": 1.023,
      "step": 3196
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.868805915799456e-05,
      "loss": 1.1465,
      "step": 3197
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8687219282985565e-05,
      "loss": 0.9298,
      "step": 3198
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.868637914647671e-05,
      "loss": 1.3605,
      "step": 3199
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.868553874847728e-05,
      "loss": 1.0859,
      "step": 3200
    },
    {
      "epoch": 0.11,
      "eval_loss": 0.9583781361579895,
      "eval_runtime": 265.8008,
      "eval_samples_per_second": 8.093,
      "eval_steps_per_second": 2.024,
      "step": 3200
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8684698088996545e-05,
      "loss": 1.0454,
      "step": 3201
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.868385716804379e-05,
      "loss": 1.0288,
      "step": 3202
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.86830159856283e-05,
      "loss": 0.9525,
      "step": 3203
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.868217454175937e-05,
      "loss": 1.0724,
      "step": 3204
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.868133283644627e-05,
      "loss": 1.0062,
      "step": 3205
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8680490869698304e-05,
      "loss": 0.7568,
      "step": 3206
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.867964864152477e-05,
      "loss": 1.0986,
      "step": 3207
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.867880615193495e-05,
      "loss": 1.2812,
      "step": 3208
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.867796340093816e-05,
      "loss": 0.6049,
      "step": 3209
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.86771203885437e-05,
      "loss": 1.1644,
      "step": 3210
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.867627711476088e-05,
      "loss": 1.3696,
      "step": 3211
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.867543357959899e-05,
      "loss": 1.3888,
      "step": 3212
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8674589783067384e-05,
      "loss": 1.0141,
      "step": 3213
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.867374572517533e-05,
      "loss": 0.9243,
      "step": 3214
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8672901405932174e-05,
      "loss": 0.9684,
      "step": 3215
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.867205682534723e-05,
      "loss": 1.0169,
      "step": 3216
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.867121198342982e-05,
      "loss": 0.949,
      "step": 3217
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.867036688018928e-05,
      "loss": 1.3867,
      "step": 3218
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.866952151563493e-05,
      "loss": 1.0217,
      "step": 3219
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8668675889776095e-05,
      "loss": 0.8949,
      "step": 3220
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8667830002622127e-05,
      "loss": 0.9934,
      "step": 3221
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8666983854182366e-05,
      "loss": 0.9372,
      "step": 3222
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.866613744446614e-05,
      "loss": 1.1197,
      "step": 3223
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.866529077348279e-05,
      "loss": 3.1444,
      "step": 3224
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.866444384124168e-05,
      "loss": 3.7891,
      "step": 3225
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.866359664775215e-05,
      "loss": 1.031,
      "step": 3226
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8662749193023553e-05,
      "loss": 0.9029,
      "step": 3227
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8661901477065244e-05,
      "loss": 1.2416,
      "step": 3228
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.866105349988659e-05,
      "loss": 1.0529,
      "step": 3229
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.866020526149694e-05,
      "loss": 1.0963,
      "step": 3230
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.865935676190566e-05,
      "loss": 1.1993,
      "step": 3231
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8658508001122126e-05,
      "loss": 0.8713,
      "step": 3232
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8657658979155706e-05,
      "loss": 1.0963,
      "step": 3233
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.865680969601577e-05,
      "loss": 1.0539,
      "step": 3234
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8655960151711685e-05,
      "loss": 1.0412,
      "step": 3235
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8655110346252854e-05,
      "loss": 1.061,
      "step": 3236
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8654260279648636e-05,
      "loss": 1.0904,
      "step": 3237
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8653409951908415e-05,
      "loss": 0.9632,
      "step": 3238
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.86525593630416e-05,
      "loss": 1.0591,
      "step": 3239
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.865170851305755e-05,
      "loss": 0.8792,
      "step": 3240
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8650857401965686e-05,
      "loss": 1.1171,
      "step": 3241
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.865000602977539e-05,
      "loss": 1.013,
      "step": 3242
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8649154396496064e-05,
      "loss": 1.0938,
      "step": 3243
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.864830250213711e-05,
      "loss": 1.1457,
      "step": 3244
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.864745034670794e-05,
      "loss": 0.9674,
      "step": 3245
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8646597930217954e-05,
      "loss": 0.859,
      "step": 3246
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.864574525267655e-05,
      "loss": 1.0051,
      "step": 3247
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.864489231409316e-05,
      "loss": 0.994,
      "step": 3248
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8644039114477205e-05,
      "loss": 0.7728,
      "step": 3249
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.864318565383809e-05,
      "loss": 1.1331,
      "step": 3250
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.864233193218523e-05,
      "loss": 1.011,
      "step": 3251
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.864147794952806e-05,
      "loss": 1.2148,
      "step": 3252
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.864062370587602e-05,
      "loss": 0.9159,
      "step": 3253
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.863976920123852e-05,
      "loss": 1.0682,
      "step": 3254
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8638914435625004e-05,
      "loss": 0.8492,
      "step": 3255
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.863805940904491e-05,
      "loss": 0.9189,
      "step": 3256
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.863720412150766e-05,
      "loss": 1.3127,
      "step": 3257
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.863634857302273e-05,
      "loss": 1.3039,
      "step": 3258
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.863549276359953e-05,
      "loss": 1.2057,
      "step": 3259
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8634636693247524e-05,
      "loss": 1.2682,
      "step": 3260
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.863378036197617e-05,
      "loss": 1.2086,
      "step": 3261
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.86329237697949e-05,
      "loss": 1.0384,
      "step": 3262
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.863206691671319e-05,
      "loss": 0.9518,
      "step": 3263
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.863120980274049e-05,
      "loss": 0.8107,
      "step": 3264
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.863035242788627e-05,
      "loss": 0.559,
      "step": 3265
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8629494792159994e-05,
      "loss": 1.1085,
      "step": 3266
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.862863689557111e-05,
      "loss": 1.1207,
      "step": 3267
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8627778738129124e-05,
      "loss": 0.897,
      "step": 3268
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.862692031984348e-05,
      "loss": 0.7369,
      "step": 3269
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8626061640723665e-05,
      "loss": 1.0677,
      "step": 3270
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8625202700779165e-05,
      "loss": 1.0465,
      "step": 3271
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.862434350001945e-05,
      "loss": 0.9184,
      "step": 3272
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.862348403845402e-05,
      "loss": 1.0354,
      "step": 3273
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.862262431609235e-05,
      "loss": 1.0102,
      "step": 3274
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.862176433294394e-05,
      "loss": 1.0169,
      "step": 3275
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.862090408901828e-05,
      "loss": 1.0395,
      "step": 3276
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.862004358432486e-05,
      "loss": 0.9272,
      "step": 3277
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8619182818873196e-05,
      "loss": 0.992,
      "step": 3278
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.861832179267278e-05,
      "loss": 0.9654,
      "step": 3279
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8617460505733114e-05,
      "loss": 1.0787,
      "step": 3280
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.861659895806372e-05,
      "loss": 1.0903,
      "step": 3281
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.861573714967409e-05,
      "loss": 0.9559,
      "step": 3282
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.861487508057375e-05,
      "loss": 1.2362,
      "step": 3283
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.861401275077222e-05,
      "loss": 1.13,
      "step": 3284
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.861315016027902e-05,
      "loss": 1.0619,
      "step": 3285
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8612287309103654e-05,
      "loss": 0.99,
      "step": 3286
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.861142419725567e-05,
      "loss": 1.0971,
      "step": 3287
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8610560824744585e-05,
      "loss": 1.1359,
      "step": 3288
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.860969719157994e-05,
      "loss": 0.8592,
      "step": 3289
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.860883329777126e-05,
      "loss": 0.6851,
      "step": 3290
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.860796914332809e-05,
      "loss": 0.8666,
      "step": 3291
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.860710472825996e-05,
      "loss": 1.0965,
      "step": 3292
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.860624005257642e-05,
      "loss": 0.9315,
      "step": 3293
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.860537511628701e-05,
      "loss": 0.9689,
      "step": 3294
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8604509919401284e-05,
      "loss": 1.1512,
      "step": 3295
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.860364446192879e-05,
      "loss": 0.9346,
      "step": 3296
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.860277874387909e-05,
      "loss": 1.0494,
      "step": 3297
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8601912765261734e-05,
      "loss": 0.8551,
      "step": 3298
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.860104652608628e-05,
      "loss": 0.9851,
      "step": 3299
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8600180026362296e-05,
      "loss": 0.9427,
      "step": 3300
    },
    {
      "epoch": 0.11,
      "eval_loss": 0.9509454369544983,
      "eval_runtime": 274.0737,
      "eval_samples_per_second": 7.848,
      "eval_steps_per_second": 1.963,
      "step": 3300
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.859931326609935e-05,
      "loss": 0.9452,
      "step": 3301
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.859844624530701e-05,
      "loss": 1.0086,
      "step": 3302
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8597578963994825e-05,
      "loss": 1.0513,
      "step": 3303
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8596711422172414e-05,
      "loss": 0.9902,
      "step": 3304
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8595843619849316e-05,
      "loss": 0.9328,
      "step": 3305
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.859497555703513e-05,
      "loss": 0.9809,
      "step": 3306
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8594107233739435e-05,
      "loss": 1.1015,
      "step": 3307
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8593238649971814e-05,
      "loss": 1.045,
      "step": 3308
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.859236980574186e-05,
      "loss": 0.7721,
      "step": 3309
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.859150070105916e-05,
      "loss": 0.8477,
      "step": 3310
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8590631335933314e-05,
      "loss": 1.1528,
      "step": 3311
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.858976171037392e-05,
      "loss": 1.0325,
      "step": 3312
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.858889182439058e-05,
      "loss": 0.9918,
      "step": 3313
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.858802167799288e-05,
      "loss": 1.1521,
      "step": 3314
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8587151271190446e-05,
      "loss": 0.898,
      "step": 3315
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.858628060399289e-05,
      "loss": 0.8774,
      "step": 3316
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.85854096764098e-05,
      "loss": 1.1038,
      "step": 3317
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8584538488450805e-05,
      "loss": 1.0191,
      "step": 3318
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8583667040125526e-05,
      "loss": 1.0496,
      "step": 3319
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.858279533144358e-05,
      "loss": 0.9814,
      "step": 3320
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.858192336241459e-05,
      "loss": 0.9643,
      "step": 3321
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.858105113304818e-05,
      "loss": 0.6335,
      "step": 3322
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.858017864335399e-05,
      "loss": 0.8095,
      "step": 3323
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.857930589334164e-05,
      "loss": 1.0069,
      "step": 3324
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.857843288302077e-05,
      "loss": 1.1028,
      "step": 3325
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8577559612401014e-05,
      "loss": 1.0166,
      "step": 3326
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.857668608149202e-05,
      "loss": 1.0468,
      "step": 3327
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8575812290303415e-05,
      "loss": 0.9738,
      "step": 3328
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.857493823884487e-05,
      "loss": 0.9301,
      "step": 3329
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.857406392712601e-05,
      "loss": 1.1454,
      "step": 3330
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.857318935515651e-05,
      "loss": 0.6491,
      "step": 3331
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.857231452294601e-05,
      "loss": 0.5582,
      "step": 3332
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.857143943050417e-05,
      "loss": 0.7892,
      "step": 3333
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8570564077840654e-05,
      "loss": 0.6441,
      "step": 3334
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.856968846496513e-05,
      "loss": 0.6892,
      "step": 3335
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8568812591887245e-05,
      "loss": 0.6189,
      "step": 3336
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8567936458616684e-05,
      "loss": 2.0794,
      "step": 3337
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.856706006516312e-05,
      "loss": 4.1197,
      "step": 3338
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8566183411536225e-05,
      "loss": 3.778,
      "step": 3339
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8565306497745674e-05,
      "loss": 0.9695,
      "step": 3340
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8564429323801165e-05,
      "loss": 1.1362,
      "step": 3341
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.856355188971236e-05,
      "loss": 0.9443,
      "step": 3342
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.856267419548895e-05,
      "loss": 0.9089,
      "step": 3343
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.856179624114062e-05,
      "loss": 0.8559,
      "step": 3344
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.856091802667708e-05,
      "loss": 1.0774,
      "step": 3345
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.856003955210802e-05,
      "loss": 1.0909,
      "step": 3346
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.855916081744313e-05,
      "loss": 0.8959,
      "step": 3347
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.855828182269211e-05,
      "loss": 0.9999,
      "step": 3348
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8557402567864666e-05,
      "loss": 0.9825,
      "step": 3349
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.855652305297052e-05,
      "loss": 0.9067,
      "step": 3350
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.855564327801936e-05,
      "loss": 0.9604,
      "step": 3351
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8554763243020905e-05,
      "loss": 0.9845,
      "step": 3352
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8553882947984875e-05,
      "loss": 1.03,
      "step": 3353
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.855300239292098e-05,
      "loss": 0.9957,
      "step": 3354
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8552121577838954e-05,
      "loss": 0.977,
      "step": 3355
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.855124050274851e-05,
      "loss": 1.1398,
      "step": 3356
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.855035916765938e-05,
      "loss": 0.9921,
      "step": 3357
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.854947757258129e-05,
      "loss": 0.9892,
      "step": 3358
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.854859571752398e-05,
      "loss": 1.0966,
      "step": 3359
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.854771360249718e-05,
      "loss": 1.0595,
      "step": 3360
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.854683122751062e-05,
      "loss": 1.029,
      "step": 3361
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.854594859257406e-05,
      "loss": 1.0653,
      "step": 3362
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8545065697697225e-05,
      "loss": 1.0606,
      "step": 3363
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8544182542889876e-05,
      "loss": 0.8505,
      "step": 3364
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8543299128161754e-05,
      "loss": 0.8481,
      "step": 3365
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8542415453522615e-05,
      "loss": 0.926,
      "step": 3366
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.854153151898222e-05,
      "loss": 1.0991,
      "step": 3367
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.854064732455031e-05,
      "loss": 1.0047,
      "step": 3368
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8539762870236664e-05,
      "loss": 0.9829,
      "step": 3369
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8538878156051045e-05,
      "loss": 0.7673,
      "step": 3370
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8537993182003205e-05,
      "loss": 1.0792,
      "step": 3371
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.853710794810293e-05,
      "loss": 1.1191,
      "step": 3372
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.853622245435999e-05,
      "loss": 0.968,
      "step": 3373
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.853533670078415e-05,
      "loss": 0.7279,
      "step": 3374
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8534450687385205e-05,
      "loss": 0.9599,
      "step": 3375
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8533564414172915e-05,
      "loss": 1.0717,
      "step": 3376
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8532677881157087e-05,
      "loss": 0.9442,
      "step": 3377
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.853179108834749e-05,
      "loss": 0.978,
      "step": 3378
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.853090403575392e-05,
      "loss": 0.9095,
      "step": 3379
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8530016723386176e-05,
      "loss": 1.064,
      "step": 3380
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.852912915125405e-05,
      "loss": 0.9602,
      "step": 3381
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8528241319367335e-05,
      "loss": 0.7786,
      "step": 3382
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8527353227735837e-05,
      "loss": 0.7363,
      "step": 3383
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.852646487636936e-05,
      "loss": 1.0032,
      "step": 3384
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.852557626527771e-05,
      "loss": 0.8927,
      "step": 3385
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.85246873944707e-05,
      "loss": 0.9374,
      "step": 3386
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8523798263958146e-05,
      "loss": 0.959,
      "step": 3387
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.852290887374985e-05,
      "loss": 0.9698,
      "step": 3388
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.852201922385564e-05,
      "loss": 1.2889,
      "step": 3389
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.852112931428534e-05,
      "loss": 0.7456,
      "step": 3390
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.852023914504877e-05,
      "loss": 0.8323,
      "step": 3391
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.851934871615575e-05,
      "loss": 0.9704,
      "step": 3392
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.851845802761613e-05,
      "loss": 2.0011,
      "step": 3393
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.851756707943973e-05,
      "loss": 1.0758,
      "step": 3394
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.851667587163638e-05,
      "loss": 1.0186,
      "step": 3395
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.851578440421593e-05,
      "loss": 0.9002,
      "step": 3396
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8514892677188215e-05,
      "loss": 0.9841,
      "step": 3397
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.851400069056309e-05,
      "loss": 0.9438,
      "step": 3398
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8513108444350384e-05,
      "loss": 0.9876,
      "step": 3399
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.851221593855996e-05,
      "loss": 1.1892,
      "step": 3400
    },
    {
      "epoch": 0.11,
      "eval_loss": 0.9470433592796326,
      "eval_runtime": 229.0415,
      "eval_samples_per_second": 9.391,
      "eval_steps_per_second": 2.349,
      "step": 3400
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.851132317320167e-05,
      "loss": 1.2565,
      "step": 3401
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.851043014828536e-05,
      "loss": 0.9182,
      "step": 3402
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.85095368638209e-05,
      "loss": 1.1694,
      "step": 3403
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8508643319818156e-05,
      "loss": 0.9256,
      "step": 3404
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.850774951628698e-05,
      "loss": 0.9096,
      "step": 3405
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8506855453237245e-05,
      "loss": 0.8894,
      "step": 3406
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.850596113067881e-05,
      "loss": 0.1742,
      "step": 3407
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.850506654862157e-05,
      "loss": 1.0349,
      "step": 3408
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8504171707075386e-05,
      "loss": 1.1596,
      "step": 3409
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8503276606050144e-05,
      "loss": 0.8305,
      "step": 3410
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.850238124555572e-05,
      "loss": 1.112,
      "step": 3411
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8501485625601996e-05,
      "loss": 0.9957,
      "step": 3412
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.850058974619886e-05,
      "loss": 0.8463,
      "step": 3413
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.849969360735621e-05,
      "loss": 0.9502,
      "step": 3414
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.849879720908393e-05,
      "loss": 1.0431,
      "step": 3415
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.849790055139193e-05,
      "loss": 1.0593,
      "step": 3416
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8497003634290094e-05,
      "loss": 0.4856,
      "step": 3417
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8496106457788336e-05,
      "loss": 0.8961,
      "step": 3418
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.849520902189655e-05,
      "loss": 0.9407,
      "step": 3419
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8494311326624643e-05,
      "loss": 1.0706,
      "step": 3420
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.849341337198253e-05,
      "loss": 1.1105,
      "step": 3421
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8492515157980127e-05,
      "loss": 1.0191,
      "step": 3422
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.849161668462735e-05,
      "loss": 0.8581,
      "step": 3423
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.849071795193411e-05,
      "loss": 1.2714,
      "step": 3424
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8489818959910334e-05,
      "loss": 1.0491,
      "step": 3425
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8488919708565945e-05,
      "loss": 0.9186,
      "step": 3426
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.848802019791086e-05,
      "loss": 1.0759,
      "step": 3427
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.848712042795505e-05,
      "loss": 1.1646,
      "step": 3428
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8486220398708395e-05,
      "loss": 0.9165,
      "step": 3429
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.848532011018086e-05,
      "loss": 0.9046,
      "step": 3430
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.848441956238238e-05,
      "loss": 1.4128,
      "step": 3431
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.84835187553229e-05,
      "loss": 1.2767,
      "step": 3432
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.848261768901235e-05,
      "loss": 0.9835,
      "step": 3433
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.84817163634607e-05,
      "loss": 0.9459,
      "step": 3434
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8480814778677886e-05,
      "loss": 0.8663,
      "step": 3435
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8479912934673857e-05,
      "loss": 0.8829,
      "step": 3436
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.847901083145858e-05,
      "loss": 0.1791,
      "step": 3437
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8478108469042e-05,
      "loss": 0.1131,
      "step": 3438
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8477205847434106e-05,
      "loss": 0.7722,
      "step": 3439
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8476302966644835e-05,
      "loss": 0.9478,
      "step": 3440
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8475399826684164e-05,
      "loss": 0.7614,
      "step": 3441
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8474496427562066e-05,
      "loss": 0.977,
      "step": 3442
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.847359276928851e-05,
      "loss": 0.7798,
      "step": 3443
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.847268885187348e-05,
      "loss": 1.0922,
      "step": 3444
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8471784675326936e-05,
      "loss": 1.039,
      "step": 3445
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.847088023965889e-05,
      "loss": 0.9714,
      "step": 3446
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.84699755448793e-05,
      "loss": 0.8898,
      "step": 3447
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.846907059099816e-05,
      "loss": 1.5807,
      "step": 3448
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8468165378025474e-05,
      "loss": 1.0337,
      "step": 3449
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.846725990597122e-05,
      "loss": 0.8667,
      "step": 3450
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8466354174845396e-05,
      "loss": 1.0246,
      "step": 3451
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.846544818465801e-05,
      "loss": 0.9743,
      "step": 3452
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.846454193541905e-05,
      "loss": 0.9313,
      "step": 3453
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.846363542713854e-05,
      "loss": 0.7949,
      "step": 3454
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8462728659826465e-05,
      "loss": 1.1035,
      "step": 3455
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.846182163349285e-05,
      "loss": 0.9662,
      "step": 3456
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.846091434814771e-05,
      "loss": 0.9916,
      "step": 3457
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.846000680380105e-05,
      "loss": 1.002,
      "step": 3458
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.84590990004629e-05,
      "loss": 1.0584,
      "step": 3459
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8458190938143276e-05,
      "loss": 0.2321,
      "step": 3460
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.845728261685221e-05,
      "loss": 0.7616,
      "step": 3461
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.845637403659972e-05,
      "loss": 1.0638,
      "step": 3462
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.845546519739583e-05,
      "loss": 1.0197,
      "step": 3463
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.845455609925059e-05,
      "loss": 0.7699,
      "step": 3464
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.845364674217403e-05,
      "loss": 1.0143,
      "step": 3465
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8452737126176195e-05,
      "loss": 1.0588,
      "step": 3466
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.845182725126711e-05,
      "loss": 1.0258,
      "step": 3467
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.845091711745683e-05,
      "loss": 0.9628,
      "step": 3468
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8450006724755415e-05,
      "loss": 1.0043,
      "step": 3469
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.844909607317289e-05,
      "loss": 0.7805,
      "step": 3470
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8448185162719326e-05,
      "loss": 1.015,
      "step": 3471
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.844727399340477e-05,
      "loss": 0.9668,
      "step": 3472
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.84463625652393e-05,
      "loss": 1.38,
      "step": 3473
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.844545087823295e-05,
      "loss": 0.9982,
      "step": 3474
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.844453893239581e-05,
      "loss": 0.9651,
      "step": 3475
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.844362672773792e-05,
      "loss": 0.9474,
      "step": 3476
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.844271426426938e-05,
      "loss": 1.0322,
      "step": 3477
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8441801542000246e-05,
      "loss": 0.9856,
      "step": 3478
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8440888560940596e-05,
      "loss": 1.2881,
      "step": 3479
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.843997532110051e-05,
      "loss": 1.0003,
      "step": 3480
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8439061822490075e-05,
      "loss": 0.52,
      "step": 3481
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.843814806511937e-05,
      "loss": 0.485,
      "step": 3482
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.843723404899848e-05,
      "loss": 0.7026,
      "step": 3483
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.84363197741375e-05,
      "loss": 0.9324,
      "step": 3484
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8435405240546536e-05,
      "loss": 1.0029,
      "step": 3485
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.843449044823566e-05,
      "loss": 0.9312,
      "step": 3486
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8433575397214984e-05,
      "loss": 0.8632,
      "step": 3487
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8432660087494606e-05,
      "loss": 0.7516,
      "step": 3488
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.843174451908463e-05,
      "loss": 0.7628,
      "step": 3489
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.843082869199518e-05,
      "loss": 0.9394,
      "step": 3490
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8429912606236335e-05,
      "loss": 0.8754,
      "step": 3491
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8428996261818246e-05,
      "loss": 0.9727,
      "step": 3492
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8428079658750995e-05,
      "loss": 1.0667,
      "step": 3493
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.842716279704472e-05,
      "loss": 1.1734,
      "step": 3494
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.842624567670953e-05,
      "loss": 1.0429,
      "step": 3495
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8425328297755576e-05,
      "loss": 0.9971,
      "step": 3496
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8424410660192953e-05,
      "loss": 0.9504,
      "step": 3497
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8423492764031816e-05,
      "loss": 1.0064,
      "step": 3498
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.842257460928228e-05,
      "loss": 1.2018,
      "step": 3499
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8421656195954497e-05,
      "loss": 1.0606,
      "step": 3500
    },
    {
      "epoch": 0.12,
      "eval_loss": 0.9460594654083252,
      "eval_runtime": 237.5255,
      "eval_samples_per_second": 9.056,
      "eval_steps_per_second": 2.265,
      "step": 3500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.842073752405859e-05,
      "loss": 1.0198,
      "step": 3501
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.841981859360472e-05,
      "loss": 1.0857,
      "step": 3502
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8418899404603016e-05,
      "loss": 0.9487,
      "step": 3503
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8417979957063624e-05,
      "loss": 1.0821,
      "step": 3504
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.841706025099671e-05,
      "loss": 0.9834,
      "step": 3505
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8416140286412416e-05,
      "loss": 0.9691,
      "step": 3506
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.84152200633209e-05,
      "loss": 0.9466,
      "step": 3507
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.841429958173232e-05,
      "loss": 0.932,
      "step": 3508
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.841337884165684e-05,
      "loss": 1.0865,
      "step": 3509
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.841245784310463e-05,
      "loss": 0.8965,
      "step": 3510
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.841153658608585e-05,
      "loss": 1.0003,
      "step": 3511
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.841061507061067e-05,
      "loss": 0.6519,
      "step": 3512
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8409693296689254e-05,
      "loss": 0.732,
      "step": 3513
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.84087712643318e-05,
      "loss": 1.0212,
      "step": 3514
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8407848973548484e-05,
      "loss": 1.0848,
      "step": 3515
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.840692642434947e-05,
      "loss": 0.786,
      "step": 3516
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8406003616744955e-05,
      "loss": 0.9338,
      "step": 3517
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.840508055074512e-05,
      "loss": 0.9129,
      "step": 3518
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.840415722636017e-05,
      "loss": 1.0258,
      "step": 3519
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.840323364360027e-05,
      "loss": 1.8074,
      "step": 3520
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8402309802475654e-05,
      "loss": 2.5289,
      "step": 3521
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.840138570299649e-05,
      "loss": 2.6161,
      "step": 3522
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8400461345173e-05,
      "loss": 2.4656,
      "step": 3523
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8399536729015374e-05,
      "loss": 2.713,
      "step": 3524
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.839861185453383e-05,
      "loss": 1.5587,
      "step": 3525
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.839768672173857e-05,
      "loss": 2.0524,
      "step": 3526
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.83967613306398e-05,
      "loss": 2.5404,
      "step": 3527
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.839583568124776e-05,
      "loss": 2.1555,
      "step": 3528
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.839490977357265e-05,
      "loss": 2.1264,
      "step": 3529
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8393983607624693e-05,
      "loss": 2.6096,
      "step": 3530
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.839305718341413e-05,
      "loss": 2.6894,
      "step": 3531
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.839213050095116e-05,
      "loss": 1.5013,
      "step": 3532
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.839120356024604e-05,
      "loss": 0.9934,
      "step": 3533
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.839027636130899e-05,
      "loss": 1.0131,
      "step": 3534
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.838934890415025e-05,
      "loss": 1.0374,
      "step": 3535
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.838842118878005e-05,
      "loss": 0.9444,
      "step": 3536
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8387493215208654e-05,
      "loss": 0.7206,
      "step": 3537
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.838656498344628e-05,
      "loss": 0.2674,
      "step": 3538
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.838563649350319e-05,
      "loss": 0.145,
      "step": 3539
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8384707745389634e-05,
      "loss": 0.7852,
      "step": 3540
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8383778739115856e-05,
      "loss": 1.2348,
      "step": 3541
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8382849474692125e-05,
      "loss": 0.9412,
      "step": 3542
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.838191995212869e-05,
      "loss": 1.0372,
      "step": 3543
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.838099017143582e-05,
      "loss": 0.9998,
      "step": 3544
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8380060132623776e-05,
      "loss": 1.025,
      "step": 3545
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8379129835702816e-05,
      "loss": 1.0836,
      "step": 3546
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.837819928068322e-05,
      "loss": 0.9888,
      "step": 3547
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.837726846757527e-05,
      "loss": 0.8353,
      "step": 3548
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.837633739638923e-05,
      "loss": 0.9596,
      "step": 3549
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.837540606713538e-05,
      "loss": 0.8768,
      "step": 3550
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8374474479823996e-05,
      "loss": 1.0235,
      "step": 3551
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.837354263446537e-05,
      "loss": 1.0099,
      "step": 3552
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.837261053106978e-05,
      "loss": 0.9948,
      "step": 3553
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8371678169647536e-05,
      "loss": 1.0606,
      "step": 3554
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8370745550208924e-05,
      "loss": 1.0881,
      "step": 3555
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8369812672764224e-05,
      "loss": 1.0622,
      "step": 3556
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.836887953732375e-05,
      "loss": 1.0128,
      "step": 3557
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.83679461438978e-05,
      "loss": 1.0957,
      "step": 3558
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8367012492496674e-05,
      "loss": 1.0352,
      "step": 3559
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.836607858313069e-05,
      "loss": 1.0523,
      "step": 3560
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.836514441581014e-05,
      "loss": 1.0297,
      "step": 3561
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8364209990545365e-05,
      "loss": 0.9831,
      "step": 3562
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.836327530734665e-05,
      "loss": 1.0358,
      "step": 3563
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.836234036622432e-05,
      "loss": 0.9911,
      "step": 3564
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.836140516718872e-05,
      "loss": 0.9208,
      "step": 3565
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.836046971025015e-05,
      "loss": 0.9738,
      "step": 3566
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.835953399541895e-05,
      "loss": 1.1469,
      "step": 3567
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.835859802270544e-05,
      "loss": 1.0355,
      "step": 3568
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.835766179211996e-05,
      "loss": 0.9291,
      "step": 3569
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8356725303672844e-05,
      "loss": 1.1244,
      "step": 3570
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.835578855737443e-05,
      "loss": 1.017,
      "step": 3571
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8354851553235056e-05,
      "loss": 0.9557,
      "step": 3572
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8353914291265075e-05,
      "loss": 0.7634,
      "step": 3573
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.835297677147483e-05,
      "loss": 1.0004,
      "step": 3574
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8352038993874665e-05,
      "loss": 1.103,
      "step": 3575
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8351100958474936e-05,
      "loss": 0.9688,
      "step": 3576
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.835016266528601e-05,
      "loss": 1.1822,
      "step": 3577
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.834922411431823e-05,
      "loss": 1.3954,
      "step": 3578
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.834828530558197e-05,
      "loss": 1.207,
      "step": 3579
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8347346239087575e-05,
      "loss": 1.0718,
      "step": 3580
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8346406914845434e-05,
      "loss": 1.015,
      "step": 3581
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.83454673328659e-05,
      "loss": 1.5225,
      "step": 3582
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.834452749315936e-05,
      "loss": 1.5011,
      "step": 3583
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.834358739573618e-05,
      "loss": 1.1088,
      "step": 3584
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8342647040606735e-05,
      "loss": 1.0252,
      "step": 3585
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.834170642778142e-05,
      "loss": 1.048,
      "step": 3586
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.834076555727061e-05,
      "loss": 0.992,
      "step": 3587
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.833982442908469e-05,
      "loss": 0.9791,
      "step": 3588
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.833888304323406e-05,
      "loss": 1.0134,
      "step": 3589
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.83379413997291e-05,
      "loss": 0.9594,
      "step": 3590
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.833699949858021e-05,
      "loss": 0.9037,
      "step": 3591
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.833605733979779e-05,
      "loss": 0.872,
      "step": 3592
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8335114923392245e-05,
      "loss": 1.1054,
      "step": 3593
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8334172249373966e-05,
      "loss": 1.1696,
      "step": 3594
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.833322931775337e-05,
      "loss": 0.9833,
      "step": 3595
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.833228612854087e-05,
      "loss": 1.3253,
      "step": 3596
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.833134268174687e-05,
      "loss": 0.5272,
      "step": 3597
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.83303989773818e-05,
      "loss": 0.4161,
      "step": 3598
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.832945501545605e-05,
      "loss": 0.292,
      "step": 3599
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8328510795980066e-05,
      "loss": 0.6117,
      "step": 3600
    },
    {
      "epoch": 0.12,
      "eval_loss": 0.9627674221992493,
      "eval_runtime": 238.8633,
      "eval_samples_per_second": 9.005,
      "eval_steps_per_second": 2.252,
      "step": 3600
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.832756631896427e-05,
      "loss": 1.316,
      "step": 3601
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.832662158441907e-05,
      "loss": 1.253,
      "step": 3602
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.832567659235492e-05,
      "loss": 1.2211,
      "step": 3603
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.832473134278224e-05,
      "loss": 1.1665,
      "step": 3604
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8323785835711456e-05,
      "loss": 0.9622,
      "step": 3605
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.832284007115302e-05,
      "loss": 1.2196,
      "step": 3606
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.832189404911738e-05,
      "loss": 1.0371,
      "step": 3607
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.832094776961496e-05,
      "loss": 1.0627,
      "step": 3608
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.832000123265621e-05,
      "loss": 0.8786,
      "step": 3609
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.831905443825159e-05,
      "loss": 0.9554,
      "step": 3610
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.831810738641155e-05,
      "loss": 1.1567,
      "step": 3611
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8317160077146544e-05,
      "loss": 1.099,
      "step": 3612
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.831621251046703e-05,
      "loss": 0.9366,
      "step": 3613
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.831526468638346e-05,
      "loss": 0.9588,
      "step": 3614
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8314316604906305e-05,
      "loss": 1.0244,
      "step": 3615
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.831336826604604e-05,
      "loss": 1.0374,
      "step": 3616
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.831241966981312e-05,
      "loss": 0.9554,
      "step": 3617
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.831147081621803e-05,
      "loss": 0.2645,
      "step": 3618
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8310521705271225e-05,
      "loss": 0.219,
      "step": 3619
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8309572336983214e-05,
      "loss": 0.2111,
      "step": 3620
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8308622711364446e-05,
      "loss": 0.3208,
      "step": 3621
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8307672828425424e-05,
      "loss": 0.2581,
      "step": 3622
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.830672268817663e-05,
      "loss": 0.2231,
      "step": 3623
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.830577229062855e-05,
      "loss": 0.2089,
      "step": 3624
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8304821635791674e-05,
      "loss": 0.2125,
      "step": 3625
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.830387072367651e-05,
      "loss": 0.256,
      "step": 3626
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.830291955429355e-05,
      "loss": 0.232,
      "step": 3627
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.830196812765328e-05,
      "loss": 0.2745,
      "step": 3628
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.830101644376622e-05,
      "loss": 0.9413,
      "step": 3629
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.830006450264287e-05,
      "loss": 1.0022,
      "step": 3630
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8299112304293746e-05,
      "loss": 0.6867,
      "step": 3631
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.829815984872935e-05,
      "loss": 0.981,
      "step": 3632
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.829720713596021e-05,
      "loss": 0.8072,
      "step": 3633
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.829625416599682e-05,
      "loss": 0.9274,
      "step": 3634
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.829530093884973e-05,
      "loss": 0.9769,
      "step": 3635
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8294347454529445e-05,
      "loss": 0.8922,
      "step": 3636
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.829339371304649e-05,
      "loss": 0.8771,
      "step": 3637
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8292439714411405e-05,
      "loss": 0.9944,
      "step": 3638
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.829148545863471e-05,
      "loss": 1.1894,
      "step": 3639
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.829053094572695e-05,
      "loss": 0.5505,
      "step": 3640
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8289576175698655e-05,
      "loss": 1.057,
      "step": 3641
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.828862114856038e-05,
      "loss": 0.6506,
      "step": 3642
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8287665864322644e-05,
      "loss": 0.5271,
      "step": 3643
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.828671032299601e-05,
      "loss": 0.505,
      "step": 3644
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.828575452459102e-05,
      "loss": 0.3842,
      "step": 3645
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.828479846911824e-05,
      "loss": 0.3448,
      "step": 3646
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.82838421565882e-05,
      "loss": 0.9416,
      "step": 3647
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8282885587011475e-05,
      "loss": 0.6967,
      "step": 3648
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.828192876039862e-05,
      "loss": 0.8725,
      "step": 3649
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8280971676760204e-05,
      "loss": 0.5671,
      "step": 3650
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8280014336106783e-05,
      "loss": 0.6336,
      "step": 3651
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.827905673844894e-05,
      "loss": 0.9971,
      "step": 3652
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8278098883797215e-05,
      "loss": 0.9584,
      "step": 3653
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.827714077216223e-05,
      "loss": 1.0292,
      "step": 3654
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8276182403554524e-05,
      "loss": 0.968,
      "step": 3655
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.827522377798469e-05,
      "loss": 0.9014,
      "step": 3656
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.827426489546331e-05,
      "loss": 0.9877,
      "step": 3657
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.827330575600097e-05,
      "loss": 0.9678,
      "step": 3658
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8272346359608266e-05,
      "loss": 0.7767,
      "step": 3659
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.827138670629578e-05,
      "loss": 0.8251,
      "step": 3660
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8270426796074106e-05,
      "loss": 0.9291,
      "step": 3661
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8269466628953855e-05,
      "loss": 1.0907,
      "step": 3662
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.82685062049456e-05,
      "loss": 1.2839,
      "step": 3663
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8267545524059963e-05,
      "loss": 1.0062,
      "step": 3664
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.826658458630756e-05,
      "loss": 1.1567,
      "step": 3665
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.826562339169897e-05,
      "loss": 0.9351,
      "step": 3666
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.826466194024483e-05,
      "loss": 0.4496,
      "step": 3667
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.826370023195574e-05,
      "loss": 1.6849,
      "step": 3668
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.826273826684232e-05,
      "loss": 1.6615,
      "step": 3669
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8261776044915196e-05,
      "loss": 1.6396,
      "step": 3670
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.826081356618499e-05,
      "loss": 1.6275,
      "step": 3671
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.825985083066231e-05,
      "loss": 1.5948,
      "step": 3672
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8258887838357816e-05,
      "loss": 1.6179,
      "step": 3673
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8257924589282105e-05,
      "loss": 1.5968,
      "step": 3674
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8256961083445826e-05,
      "loss": 1.6019,
      "step": 3675
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.825599732085962e-05,
      "loss": 1.6417,
      "step": 3676
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8255033301534136e-05,
      "loss": 1.6281,
      "step": 3677
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.825406902547999e-05,
      "loss": 1.6211,
      "step": 3678
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.825310449270785e-05,
      "loss": 1.081,
      "step": 3679
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.825213970322835e-05,
      "loss": 1.0112,
      "step": 3680
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.825117465705214e-05,
      "loss": 1.0558,
      "step": 3681
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.825020935418989e-05,
      "loss": 1.014,
      "step": 3682
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.824924379465224e-05,
      "loss": 0.7625,
      "step": 3683
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8248277978449866e-05,
      "loss": 0.397,
      "step": 3684
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.824731190559341e-05,
      "loss": 0.341,
      "step": 3685
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.824634557609355e-05,
      "loss": 0.3826,
      "step": 3686
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.824537898996095e-05,
      "loss": 0.3533,
      "step": 3687
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8244412147206284e-05,
      "loss": 0.3829,
      "step": 3688
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.824344504784023e-05,
      "loss": 1.0694,
      "step": 3689
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8242477691873453e-05,
      "loss": 0.892,
      "step": 3690
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.824151007931664e-05,
      "loss": 1.0349,
      "step": 3691
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8240542210180474e-05,
      "loss": 1.0454,
      "step": 3692
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.823957408447563e-05,
      "loss": 0.8706,
      "step": 3693
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.82386057022128e-05,
      "loss": 0.855,
      "step": 3694
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.823763706340269e-05,
      "loss": 0.9068,
      "step": 3695
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8236668168055976e-05,
      "loss": 1.0196,
      "step": 3696
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.823569901618336e-05,
      "loss": 1.0771,
      "step": 3697
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.823472960779554e-05,
      "loss": 0.4277,
      "step": 3698
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.823375994290321e-05,
      "loss": 1.0841,
      "step": 3699
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.823279002151709e-05,
      "loss": 0.9513,
      "step": 3700
    },
    {
      "epoch": 0.12,
      "eval_loss": 0.9867281913757324,
      "eval_runtime": 227.9659,
      "eval_samples_per_second": 9.436,
      "eval_steps_per_second": 2.36,
      "step": 3700
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8231819843647885e-05,
      "loss": 0.4547,
      "step": 3701
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.82308494093063e-05,
      "loss": 0.8252,
      "step": 3702
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.822987871850305e-05,
      "loss": 0.361,
      "step": 3703
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.822890777124884e-05,
      "loss": 0.4935,
      "step": 3704
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.822793656755441e-05,
      "loss": 1.0548,
      "step": 3705
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.822696510743048e-05,
      "loss": 1.0771,
      "step": 3706
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8225993390887744e-05,
      "loss": 1.0703,
      "step": 3707
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.822502141793696e-05,
      "loss": 1.006,
      "step": 3708
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.822404918858887e-05,
      "loss": 1.1809,
      "step": 3709
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.822307670285417e-05,
      "loss": 0.7931,
      "step": 3710
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.822210396074361e-05,
      "loss": 1.3327,
      "step": 3711
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.822113096226794e-05,
      "loss": 1.0195,
      "step": 3712
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8220157707437884e-05,
      "loss": 0.8243,
      "step": 3713
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8219184196264205e-05,
      "loss": 1.2611,
      "step": 3714
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8218210428757635e-05,
      "loss": 0.8519,
      "step": 3715
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8217236404928933e-05,
      "loss": 0.5849,
      "step": 3716
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8216262124788855e-05,
      "loss": 0.9906,
      "step": 3717
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.821528758834815e-05,
      "loss": 0.9443,
      "step": 3718
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.821431279561757e-05,
      "loss": 1.001,
      "step": 3719
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.821333774660788e-05,
      "loss": 0.9234,
      "step": 3720
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.821236244132986e-05,
      "loss": 0.4616,
      "step": 3721
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.821138687979426e-05,
      "loss": 1.7084,
      "step": 3722
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8210411062011857e-05,
      "loss": 3.1561,
      "step": 3723
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.820943498799342e-05,
      "loss": 1.6923,
      "step": 3724
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.820845865774973e-05,
      "loss": 1.1505,
      "step": 3725
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.820748207129156e-05,
      "loss": 0.8623,
      "step": 3726
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.82065052286297e-05,
      "loss": 0.9715,
      "step": 3727
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8205528129774915e-05,
      "loss": 0.9803,
      "step": 3728
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.820455077473801e-05,
      "loss": 2.5531,
      "step": 3729
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8203573163529766e-05,
      "loss": 0.964,
      "step": 3730
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.820259529616098e-05,
      "loss": 0.8876,
      "step": 3731
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.820161717264244e-05,
      "loss": 1.0317,
      "step": 3732
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8200638792984965e-05,
      "loss": 1.0293,
      "step": 3733
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.819966015719933e-05,
      "loss": 0.8742,
      "step": 3734
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8198681265296355e-05,
      "loss": 1.0033,
      "step": 3735
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8197702117286836e-05,
      "loss": 1.0359,
      "step": 3736
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.819672271318159e-05,
      "loss": 0.9913,
      "step": 3737
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8195743052991426e-05,
      "loss": 0.9912,
      "step": 3738
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8194763136727164e-05,
      "loss": 1.0434,
      "step": 3739
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.819378296439961e-05,
      "loss": 0.9363,
      "step": 3740
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.81928025360196e-05,
      "loss": 1.0034,
      "step": 3741
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.819182185159795e-05,
      "loss": 0.9144,
      "step": 3742
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.81908409111455e-05,
      "loss": 0.9203,
      "step": 3743
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.818985971467305e-05,
      "loss": 0.4378,
      "step": 3744
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.818887826219145e-05,
      "loss": 0.935,
      "step": 3745
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8187896553711545e-05,
      "loss": 0.9634,
      "step": 3746
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8186914589244144e-05,
      "loss": 1.059,
      "step": 3747
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.818593236880012e-05,
      "loss": 0.8354,
      "step": 3748
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8184949892390295e-05,
      "loss": 0.9481,
      "step": 3749
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.8183967160025525e-05,
      "loss": 3.5828,
      "step": 3750
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.818298417171665e-05,
      "loss": 0.9184,
      "step": 3751
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8182000927474535e-05,
      "loss": 1.0243,
      "step": 3752
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.818101742731003e-05,
      "loss": 0.9953,
      "step": 3753
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.818003367123398e-05,
      "loss": 0.8895,
      "step": 3754
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.817904965925726e-05,
      "loss": 1.1199,
      "step": 3755
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8178065391390726e-05,
      "loss": 0.9888,
      "step": 3756
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8177080867645245e-05,
      "loss": 0.8845,
      "step": 3757
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.81760960880317e-05,
      "loss": 0.4384,
      "step": 3758
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.817511105256093e-05,
      "loss": 1.0222,
      "step": 3759
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.817412576124385e-05,
      "loss": 1.0259,
      "step": 3760
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.817314021409131e-05,
      "loss": 1.0276,
      "step": 3761
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.817215441111419e-05,
      "loss": 2.3866,
      "step": 3762
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8171168352323385e-05,
      "loss": 1.024,
      "step": 3763
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8170182037729774e-05,
      "loss": 1.1503,
      "step": 3764
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8169195467344255e-05,
      "loss": 1.0925,
      "step": 3765
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.81682086411777e-05,
      "loss": 0.8332,
      "step": 3766
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.816722155924102e-05,
      "loss": 0.8898,
      "step": 3767
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.816623422154511e-05,
      "loss": 0.9853,
      "step": 3768
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.816524662810087e-05,
      "loss": 0.9973,
      "step": 3769
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.816425877891919e-05,
      "loss": 0.9524,
      "step": 3770
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8163270674010996e-05,
      "loss": 1.0498,
      "step": 3771
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8162282313387176e-05,
      "loss": 0.9867,
      "step": 3772
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.816129369705866e-05,
      "loss": 0.641,
      "step": 3773
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.816030482503635e-05,
      "loss": 0.8821,
      "step": 3774
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.815931569733116e-05,
      "loss": 1.0129,
      "step": 3775
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.815832631395402e-05,
      "loss": 0.8432,
      "step": 3776
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.815733667491585e-05,
      "loss": 0.836,
      "step": 3777
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.815634678022757e-05,
      "loss": 1.0303,
      "step": 3778
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.815535662990012e-05,
      "loss": 0.9331,
      "step": 3779
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.815436622394441e-05,
      "loss": 1.6612,
      "step": 3780
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8153375562371395e-05,
      "loss": 1.7098,
      "step": 3781
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.815238464519201e-05,
      "loss": 1.6661,
      "step": 3782
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.815139347241717e-05,
      "loss": 1.6711,
      "step": 3783
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.815040204405785e-05,
      "loss": 1.6683,
      "step": 3784
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8149410360124966e-05,
      "loss": 1.5932,
      "step": 3785
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8148418420629484e-05,
      "loss": 1.5779,
      "step": 3786
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.814742622558236e-05,
      "loss": 1.5717,
      "step": 3787
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8146433774994524e-05,
      "loss": 1.5814,
      "step": 3788
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.814544106887696e-05,
      "loss": 1.572,
      "step": 3789
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.81444481072406e-05,
      "loss": 1.5542,
      "step": 3790
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8143454890096426e-05,
      "loss": 0.9323,
      "step": 3791
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.81424614174554e-05,
      "loss": 1.0114,
      "step": 3792
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.814146768932848e-05,
      "loss": 0.9886,
      "step": 3793
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.814047370572664e-05,
      "loss": 0.9602,
      "step": 3794
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.813947946666087e-05,
      "loss": 1.4426,
      "step": 3795
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.813848497214211e-05,
      "loss": 1.3249,
      "step": 3796
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8137490222181374e-05,
      "loss": 0.997,
      "step": 3797
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.813649521678963e-05,
      "loss": 0.9295,
      "step": 3798
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.813549995597786e-05,
      "loss": 1.0226,
      "step": 3799
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.813450443975706e-05,
      "loss": 1.3779,
      "step": 3800
    },
    {
      "epoch": 0.13,
      "eval_loss": 0.9605572819709778,
      "eval_runtime": 227.0217,
      "eval_samples_per_second": 9.475,
      "eval_steps_per_second": 2.37,
      "step": 3800
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8133508668138204e-05,
      "loss": 0.9718,
      "step": 3801
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.81325126411323e-05,
      "loss": 1.0353,
      "step": 3802
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8131516358750344e-05,
      "loss": 0.9589,
      "step": 3803
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.813051982100333e-05,
      "loss": 0.9047,
      "step": 3804
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.812952302790226e-05,
      "loss": 0.6771,
      "step": 3805
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8128525979458126e-05,
      "loss": 1.0259,
      "step": 3806
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8127528675681956e-05,
      "loss": 0.8835,
      "step": 3807
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.812653111658475e-05,
      "loss": 1.0052,
      "step": 3808
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.812553330217753e-05,
      "loss": 0.9379,
      "step": 3809
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.812453523247129e-05,
      "loss": 1.0078,
      "step": 3810
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.812353690747706e-05,
      "loss": 2.1435,
      "step": 3811
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.812253832720588e-05,
      "loss": 0.9533,
      "step": 3812
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.812153949166874e-05,
      "loss": 0.9208,
      "step": 3813
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.812054040087669e-05,
      "loss": 0.8361,
      "step": 3814
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.811954105484075e-05,
      "loss": 1.0892,
      "step": 3815
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.811854145357196e-05,
      "loss": 0.9798,
      "step": 3816
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.811754159708136e-05,
      "loss": 1.0588,
      "step": 3817
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.811654148537996e-05,
      "loss": 1.0486,
      "step": 3818
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.811554111847884e-05,
      "loss": 0.9691,
      "step": 3819
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.811454049638901e-05,
      "loss": 0.9177,
      "step": 3820
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8113539619121534e-05,
      "loss": 0.5276,
      "step": 3821
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.811253848668746e-05,
      "loss": 0.9383,
      "step": 3822
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8111537099097837e-05,
      "loss": 1.0567,
      "step": 3823
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.811053545636372e-05,
      "loss": 1.0324,
      "step": 3824
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.810953355849617e-05,
      "loss": 0.944,
      "step": 3825
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.810853140550624e-05,
      "loss": 0.8132,
      "step": 3826
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8107528997405e-05,
      "loss": 0.9642,
      "step": 3827
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8106526334203526e-05,
      "loss": 0.8706,
      "step": 3828
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.810552341591287e-05,
      "loss": 0.9902,
      "step": 3829
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8104520242544105e-05,
      "loss": 0.8412,
      "step": 3830
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.810351681410832e-05,
      "loss": 1.1999,
      "step": 3831
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.810251313061658e-05,
      "loss": 0.9312,
      "step": 3832
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8101509192079965e-05,
      "loss": 1.0445,
      "step": 3833
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8100504998509566e-05,
      "loss": 1.1754,
      "step": 3834
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8099500549916465e-05,
      "loss": 1.1185,
      "step": 3835
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8098495846311756e-05,
      "loss": 0.9855,
      "step": 3836
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.809749088770652e-05,
      "loss": 1.0794,
      "step": 3837
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.809648567411186e-05,
      "loss": 1.1788,
      "step": 3838
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8095480205538866e-05,
      "loss": 1.1738,
      "step": 3839
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8094474481998644e-05,
      "loss": 1.126,
      "step": 3840
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.809346850350229e-05,
      "loss": 1.1511,
      "step": 3841
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.809246227006092e-05,
      "loss": 1.0153,
      "step": 3842
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8091455781685645e-05,
      "loss": 0.968,
      "step": 3843
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8090449038387564e-05,
      "loss": 0.9564,
      "step": 3844
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.808944204017779e-05,
      "loss": 1.0774,
      "step": 3845
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.808843478706745e-05,
      "loss": 0.9615,
      "step": 3846
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8087427279067666e-05,
      "loss": 0.9788,
      "step": 3847
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.808641951618955e-05,
      "loss": 0.9778,
      "step": 3848
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.808541149844423e-05,
      "loss": 1.0507,
      "step": 3849
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8084403225842834e-05,
      "loss": 0.9681,
      "step": 3850
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.80833946983965e-05,
      "loss": 1.0419,
      "step": 3851
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8082385916116356e-05,
      "loss": 1.1149,
      "step": 3852
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.808137687901354e-05,
      "loss": 1.3813,
      "step": 3853
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.808036758709919e-05,
      "loss": 1.0159,
      "step": 3854
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.807935804038445e-05,
      "loss": 0.9242,
      "step": 3855
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.807834823888046e-05,
      "loss": 0.9409,
      "step": 3856
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8077338182598374e-05,
      "loss": 1.0274,
      "step": 3857
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.807632787154934e-05,
      "loss": 1.0015,
      "step": 3858
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8075317305744515e-05,
      "loss": 1.0305,
      "step": 3859
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.807430648519505e-05,
      "loss": 0.9924,
      "step": 3860
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8073295409912114e-05,
      "loss": 0.7823,
      "step": 3861
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.807228407990686e-05,
      "loss": 0.9988,
      "step": 3862
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.807127249519045e-05,
      "loss": 1.0178,
      "step": 3863
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8070260655774056e-05,
      "loss": 1.0434,
      "step": 3864
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.806924856166886e-05,
      "loss": 1.0265,
      "step": 3865
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.806823621288601e-05,
      "loss": 1.0661,
      "step": 3866
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.80672236094367e-05,
      "loss": 4.5584,
      "step": 3867
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.806621075133211e-05,
      "loss": 3.4208,
      "step": 3868
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.806519763858342e-05,
      "loss": 0.899,
      "step": 3869
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.806418427120179e-05,
      "loss": 1.1536,
      "step": 3870
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.806317064919845e-05,
      "loss": 0.9787,
      "step": 3871
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.806215677258456e-05,
      "loss": 0.9083,
      "step": 3872
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.806114264137132e-05,
      "loss": 0.9779,
      "step": 3873
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.806012825556993e-05,
      "loss": 0.9731,
      "step": 3874
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8059113615191584e-05,
      "loss": 1.0624,
      "step": 3875
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8058098720247485e-05,
      "loss": 1.1141,
      "step": 3876
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.805708357074884e-05,
      "loss": 1.5233,
      "step": 3877
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8056068166706846e-05,
      "loss": 1.461,
      "step": 3878
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.805505250813272e-05,
      "loss": 4.3672,
      "step": 3879
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.805403659503769e-05,
      "loss": 1.0157,
      "step": 3880
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.805302042743294e-05,
      "loss": 1.06,
      "step": 3881
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.805200400532971e-05,
      "loss": 0.8398,
      "step": 3882
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.805098732873921e-05,
      "loss": 0.9508,
      "step": 3883
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.804997039767267e-05,
      "loss": 2.8804,
      "step": 3884
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.804895321214131e-05,
      "loss": 0.9111,
      "step": 3885
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8047935772156373e-05,
      "loss": 0.5426,
      "step": 3886
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.804691807772909e-05,
      "loss": 1.0617,
      "step": 3887
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.804590012887067e-05,
      "loss": 1.1247,
      "step": 3888
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.804488192559238e-05,
      "loss": 1.0235,
      "step": 3889
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.804386346790545e-05,
      "loss": 0.9281,
      "step": 3890
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8042844755821126e-05,
      "loss": 0.8518,
      "step": 3891
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.804182578935065e-05,
      "loss": 0.2929,
      "step": 3892
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.804080656850527e-05,
      "loss": 0.1737,
      "step": 3893
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8039787093296246e-05,
      "loss": 0.3392,
      "step": 3894
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.803876736373483e-05,
      "loss": 1.1373,
      "step": 3895
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8037747379832265e-05,
      "loss": 0.8962,
      "step": 3896
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8036727141599834e-05,
      "loss": 0.9061,
      "step": 3897
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.803570664904879e-05,
      "loss": 1.0198,
      "step": 3898
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.803468590219039e-05,
      "loss": 0.6197,
      "step": 3899
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.803366490103592e-05,
      "loss": 0.9037,
      "step": 3900
    },
    {
      "epoch": 0.13,
      "eval_loss": 0.9738965034484863,
      "eval_runtime": 227.1975,
      "eval_samples_per_second": 9.468,
      "eval_steps_per_second": 2.368,
      "step": 3900
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.803264364559664e-05,
      "loss": 0.9634,
      "step": 3901
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8031622135883834e-05,
      "loss": 1.4138,
      "step": 3902
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.803060037190876e-05,
      "loss": 1.0142,
      "step": 3903
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8029578353682725e-05,
      "loss": 1.0009,
      "step": 3904
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.802855608121699e-05,
      "loss": 0.995,
      "step": 3905
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.802753355452285e-05,
      "loss": 0.982,
      "step": 3906
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.802651077361159e-05,
      "loss": 0.9831,
      "step": 3907
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.80254877384945e-05,
      "loss": 1.0192,
      "step": 3908
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8024464449182875e-05,
      "loss": 1.0277,
      "step": 3909
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8023440905688015e-05,
      "loss": 1.0422,
      "step": 3910
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.802241710802122e-05,
      "loss": 0.8345,
      "step": 3911
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8021393056193794e-05,
      "loss": 1.0257,
      "step": 3912
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.802036875021703e-05,
      "loss": 0.9884,
      "step": 3913
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.801934419010225e-05,
      "loss": 0.9493,
      "step": 3914
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.801831937586075e-05,
      "loss": 0.8077,
      "step": 3915
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.801729430750387e-05,
      "loss": 1.0128,
      "step": 3916
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.80162689850429e-05,
      "loss": 1.0519,
      "step": 3917
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.801524340848917e-05,
      "loss": 1.1978,
      "step": 3918
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8014217577854e-05,
      "loss": 1.0311,
      "step": 3919
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.801319149314871e-05,
      "loss": 1.2768,
      "step": 3920
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.801216515438465e-05,
      "loss": 0.9306,
      "step": 3921
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.801113856157312e-05,
      "loss": 0.9536,
      "step": 3922
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.801011171472547e-05,
      "loss": 1.0078,
      "step": 3923
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.800908461385303e-05,
      "loss": 0.9087,
      "step": 3924
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.800805725896715e-05,
      "loss": 1.0833,
      "step": 3925
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.800702965007916e-05,
      "loss": 1.0491,
      "step": 3926
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.800600178720041e-05,
      "loss": 1.056,
      "step": 3927
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.800497367034223e-05,
      "loss": 0.9691,
      "step": 3928
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8003945299516006e-05,
      "loss": 1.0786,
      "step": 3929
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.800291667473307e-05,
      "loss": 1.1099,
      "step": 3930
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.8001887796004765e-05,
      "loss": 1.3822,
      "step": 3931
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.800085866334248e-05,
      "loss": 1.0605,
      "step": 3932
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.799982927675753e-05,
      "loss": 1.0153,
      "step": 3933
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.799879963626134e-05,
      "loss": 0.8382,
      "step": 3934
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7997769741865226e-05,
      "loss": 0.8366,
      "step": 3935
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7996739593580585e-05,
      "loss": 0.8046,
      "step": 3936
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7995709191418783e-05,
      "loss": 1.0109,
      "step": 3937
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.799467853539119e-05,
      "loss": 0.9738,
      "step": 3938
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.799364762550919e-05,
      "loss": 1.087,
      "step": 3939
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.799261646178416e-05,
      "loss": 1.0433,
      "step": 3940
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.799158504422748e-05,
      "loss": 1.1256,
      "step": 3941
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7990553372850546e-05,
      "loss": 1.1082,
      "step": 3942
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7989521447664755e-05,
      "loss": 1.0059,
      "step": 3943
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7988489268681466e-05,
      "loss": 0.9177,
      "step": 3944
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.798745683591212e-05,
      "loss": 0.9838,
      "step": 3945
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.798642414936807e-05,
      "loss": 0.8638,
      "step": 3946
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.798539120906075e-05,
      "loss": 0.8225,
      "step": 3947
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.798435801500154e-05,
      "loss": 0.3068,
      "step": 3948
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.798332456720187e-05,
      "loss": 0.891,
      "step": 3949
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7982290865673116e-05,
      "loss": 0.9919,
      "step": 3950
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.798125691042672e-05,
      "loss": 1.0379,
      "step": 3951
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.798022270147409e-05,
      "loss": 0.9629,
      "step": 3952
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7979188238826635e-05,
      "loss": 0.9237,
      "step": 3953
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7978153522495776e-05,
      "loss": 1.0442,
      "step": 3954
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.797711855249294e-05,
      "loss": 0.9667,
      "step": 3955
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.797608332882956e-05,
      "loss": 0.8545,
      "step": 3956
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7975047851517054e-05,
      "loss": 1.0297,
      "step": 3957
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7974012120566844e-05,
      "loss": 1.115,
      "step": 3958
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.797297613599039e-05,
      "loss": 2.4379,
      "step": 3959
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.797193989779911e-05,
      "loss": 0.9546,
      "step": 3960
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.797090340600444e-05,
      "loss": 0.8931,
      "step": 3961
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7969866660617845e-05,
      "loss": 1.029,
      "step": 3962
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.796882966165075e-05,
      "loss": 1.0445,
      "step": 3963
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7967792409114606e-05,
      "loss": 0.967,
      "step": 3964
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7966754903020873e-05,
      "loss": 1.0248,
      "step": 3965
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7965717143381e-05,
      "loss": 1.3638,
      "step": 3966
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.796467913020644e-05,
      "loss": 1.0071,
      "step": 3967
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7963640863508665e-05,
      "loss": 1.0414,
      "step": 3968
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.796260234329911e-05,
      "loss": 0.5251,
      "step": 3969
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7961563569589274e-05,
      "loss": 0.2454,
      "step": 3970
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.79605245423906e-05,
      "loss": 0.5799,
      "step": 3971
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7959485261714566e-05,
      "loss": 0.3365,
      "step": 3972
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.795844572757265e-05,
      "loss": 0.7491,
      "step": 3973
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7957405939976324e-05,
      "loss": 1.064,
      "step": 3974
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7956365898937064e-05,
      "loss": 0.7947,
      "step": 3975
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.795532560446636e-05,
      "loss": 1.0782,
      "step": 3976
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.795428505657569e-05,
      "loss": 0.9289,
      "step": 3977
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.795324425527654e-05,
      "loss": 0.9644,
      "step": 3978
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7952203200580403e-05,
      "loss": 0.9086,
      "step": 3979
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.795116189249878e-05,
      "loss": 0.7772,
      "step": 3980
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7950120331043144e-05,
      "loss": 1.0453,
      "step": 3981
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.794907851622502e-05,
      "loss": 0.9148,
      "step": 3982
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.794803644805589e-05,
      "loss": 1.0881,
      "step": 3983
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7946994126547275e-05,
      "loss": 4.9338,
      "step": 3984
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7945951551710666e-05,
      "loss": 5.2832,
      "step": 3985
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.794490872355758e-05,
      "loss": 2.3279,
      "step": 3986
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.794386564209953e-05,
      "loss": 0.9308,
      "step": 3987
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.794282230734804e-05,
      "loss": 0.6797,
      "step": 3988
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.79417787193146e-05,
      "loss": 0.9283,
      "step": 3989
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.794073487801076e-05,
      "loss": 1.1037,
      "step": 3990
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7939690783448024e-05,
      "loss": 1.0988,
      "step": 3991
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7938646435637936e-05,
      "loss": 1.1579,
      "step": 3992
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7937601834592015e-05,
      "loss": 1.2461,
      "step": 3993
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.793655698032179e-05,
      "loss": 1.0908,
      "step": 3994
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.793551187283881e-05,
      "loss": 1.0822,
      "step": 3995
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7934466512154596e-05,
      "loss": 0.9621,
      "step": 3996
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7933420898280696e-05,
      "loss": 1.1105,
      "step": 3997
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.793237503122865e-05,
      "loss": 1.2796,
      "step": 3998
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.793132891101001e-05,
      "loss": 0.9271,
      "step": 3999
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.793028253763633e-05,
      "loss": 1.1565,
      "step": 4000
    },
    {
      "epoch": 0.13,
      "eval_loss": 0.9760063886642456,
      "eval_runtime": 228.4989,
      "eval_samples_per_second": 9.414,
      "eval_steps_per_second": 2.354,
      "step": 4000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.792923591111914e-05,
      "loss": 0.9916,
      "step": 4001
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7928189031470015e-05,
      "loss": 1.0583,
      "step": 4002
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.792714189870051e-05,
      "loss": 0.9428,
      "step": 4003
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.792609451282217e-05,
      "loss": 0.8969,
      "step": 4004
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.792504687384658e-05,
      "loss": 0.944,
      "step": 4005
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.792399898178529e-05,
      "loss": 1.0538,
      "step": 4006
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7922950836649867e-05,
      "loss": 1.0836,
      "step": 4007
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7921902438451885e-05,
      "loss": 1.0795,
      "step": 4008
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7920853787202926e-05,
      "loss": 0.7942,
      "step": 4009
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.791980488291456e-05,
      "loss": 1.0328,
      "step": 4010
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.791875572559838e-05,
      "loss": 0.7375,
      "step": 4011
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7917706315265934e-05,
      "loss": 1.1514,
      "step": 4012
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7916656651928836e-05,
      "loss": 1.0192,
      "step": 4013
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.791560673559867e-05,
      "loss": 0.9533,
      "step": 4014
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7914556566287024e-05,
      "loss": 0.9367,
      "step": 4015
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.79135061440055e-05,
      "loss": 0.9978,
      "step": 4016
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7912455468765674e-05,
      "loss": 1.0301,
      "step": 4017
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7911404540579166e-05,
      "loss": 1.0082,
      "step": 4018
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.791035335945756e-05,
      "loss": 1.0886,
      "step": 4019
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.790930192541248e-05,
      "loss": 0.9438,
      "step": 4020
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7908250238455516e-05,
      "loss": 0.9064,
      "step": 4021
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7907198298598286e-05,
      "loss": 1.0507,
      "step": 4022
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7906146105852404e-05,
      "loss": 0.888,
      "step": 4023
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7905093660229484e-05,
      "loss": 1.0431,
      "step": 4024
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.790404096174115e-05,
      "loss": 1.1044,
      "step": 4025
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.790298801039901e-05,
      "loss": 1.1739,
      "step": 4026
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.79019348062147e-05,
      "loss": 1.1504,
      "step": 4027
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.790088134919985e-05,
      "loss": 0.7145,
      "step": 4028
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.789982763936608e-05,
      "loss": 1.0694,
      "step": 4029
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.789877367672503e-05,
      "loss": 0.9788,
      "step": 4030
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.789771946128833e-05,
      "loss": 1.1644,
      "step": 4031
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.789666499306761e-05,
      "loss": 0.7515,
      "step": 4032
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7895610272074535e-05,
      "loss": 0.8754,
      "step": 4033
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7894555298320726e-05,
      "loss": 0.4332,
      "step": 4034
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.789350007181784e-05,
      "loss": 0.9311,
      "step": 4035
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.789244459257753e-05,
      "loss": 1.0669,
      "step": 4036
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.789138886061144e-05,
      "loss": 0.934,
      "step": 4037
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.789033287593123e-05,
      "loss": 1.149,
      "step": 4038
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.788927663854855e-05,
      "loss": 0.677,
      "step": 4039
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.788822014847507e-05,
      "loss": 0.9741,
      "step": 4040
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7887163405722456e-05,
      "loss": 0.7107,
      "step": 4041
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7886106410302365e-05,
      "loss": 1.0003,
      "step": 4042
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.788504916222647e-05,
      "loss": 1.5188,
      "step": 4043
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.788399166150643e-05,
      "loss": 1.0303,
      "step": 4044
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.788293390815395e-05,
      "loss": 1.2159,
      "step": 4045
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.788187590218067e-05,
      "loss": 1.0781,
      "step": 4046
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.78808176435983e-05,
      "loss": 0.5686,
      "step": 4047
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7879759132418514e-05,
      "loss": 0.9306,
      "step": 4048
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.7878700368652984e-05,
      "loss": 1.0297,
      "step": 4049
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.787764135231341e-05,
      "loss": 0.9796,
      "step": 4050
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7876582083411494e-05,
      "loss": 1.2485,
      "step": 4051
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7875522561958906e-05,
      "loss": 1.0921,
      "step": 4052
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.787446278796736e-05,
      "loss": 0.9851,
      "step": 4053
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.787340276144855e-05,
      "loss": 1.0287,
      "step": 4054
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7872342482414185e-05,
      "loss": 1.0078,
      "step": 4055
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.787128195087596e-05,
      "loss": 0.9778,
      "step": 4056
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.787022116684559e-05,
      "loss": 0.8745,
      "step": 4057
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.786916013033478e-05,
      "loss": 0.8586,
      "step": 4058
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.786809884135525e-05,
      "loss": 1.0972,
      "step": 4059
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.78670372999187e-05,
      "loss": 2.2158,
      "step": 4060
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.786597550603688e-05,
      "loss": 1.8067,
      "step": 4061
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7864913459721486e-05,
      "loss": 0.9451,
      "step": 4062
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.786385116098425e-05,
      "loss": 1.0001,
      "step": 4063
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.786278860983691e-05,
      "loss": 1.0117,
      "step": 4064
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7861725806291175e-05,
      "loss": 1.0039,
      "step": 4065
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.786066275035879e-05,
      "loss": 0.9742,
      "step": 4066
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7859599442051495e-05,
      "loss": 1.0377,
      "step": 4067
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7858535881381026e-05,
      "loss": 0.9964,
      "step": 4068
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7857472068359114e-05,
      "loss": 1.1515,
      "step": 4069
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.785640800299752e-05,
      "loss": 0.329,
      "step": 4070
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.785534368530798e-05,
      "loss": 0.4492,
      "step": 4071
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.785427911530224e-05,
      "loss": 0.8976,
      "step": 4072
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.785321429299207e-05,
      "loss": 0.902,
      "step": 4073
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.785214921838921e-05,
      "loss": 0.9505,
      "step": 4074
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7851083891505416e-05,
      "loss": 0.6563,
      "step": 4075
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.785001831235246e-05,
      "loss": 0.559,
      "step": 4076
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.78489524809421e-05,
      "loss": 0.4797,
      "step": 4077
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.78478863972861e-05,
      "loss": 0.5997,
      "step": 4078
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.784682006139623e-05,
      "loss": 0.4739,
      "step": 4079
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7845753473284274e-05,
      "loss": 0.6401,
      "step": 4080
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.784468663296199e-05,
      "loss": 0.5223,
      "step": 4081
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.784361954044117e-05,
      "loss": 0.5895,
      "step": 4082
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.784255219573358e-05,
      "loss": 0.537,
      "step": 4083
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7841484598851015e-05,
      "loss": 0.7441,
      "step": 4084
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7840416749805255e-05,
      "loss": 1.5308,
      "step": 4085
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.783934864860809e-05,
      "loss": 0.958,
      "step": 4086
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.783828029527131e-05,
      "loss": 1.1249,
      "step": 4087
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.783721168980672e-05,
      "loss": 1.0151,
      "step": 4088
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7836142832226096e-05,
      "loss": 1.0324,
      "step": 4089
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7835073722541254e-05,
      "loss": 1.0731,
      "step": 4090
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7834004360764e-05,
      "loss": 1.099,
      "step": 4091
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7832934746906125e-05,
      "loss": 0.8061,
      "step": 4092
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.783186488097944e-05,
      "loss": 1.0675,
      "step": 4093
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7830794762995766e-05,
      "loss": 0.8062,
      "step": 4094
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7829724392966915e-05,
      "loss": 0.9969,
      "step": 4095
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7828653770904685e-05,
      "loss": 1.0506,
      "step": 4096
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.782758289682092e-05,
      "loss": 0.9059,
      "step": 4097
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.782651177072743e-05,
      "loss": 0.8321,
      "step": 4098
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7825440392636045e-05,
      "loss": 0.8703,
      "step": 4099
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.782436876255859e-05,
      "loss": 0.968,
      "step": 4100
    },
    {
      "epoch": 0.14,
      "eval_loss": 0.9610686898231506,
      "eval_runtime": 227.6846,
      "eval_samples_per_second": 9.447,
      "eval_steps_per_second": 2.363,
      "step": 4100
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.78232968805069e-05,
      "loss": 1.334,
      "step": 4101
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.782222474649279e-05,
      "loss": 1.1681,
      "step": 4102
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.782115236052811e-05,
      "loss": 1.0522,
      "step": 4103
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.782007972262471e-05,
      "loss": 0.716,
      "step": 4104
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.781900683279441e-05,
      "loss": 1.0087,
      "step": 4105
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.781793369104907e-05,
      "loss": 1.6599,
      "step": 4106
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7816860297400535e-05,
      "loss": 1.3219,
      "step": 4107
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.781578665186064e-05,
      "loss": 0.5998,
      "step": 4108
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.781471275444126e-05,
      "loss": 1.0832,
      "step": 4109
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.781363860515423e-05,
      "loss": 1.1188,
      "step": 4110
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.781256420401142e-05,
      "loss": 0.9454,
      "step": 4111
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.78114895510247e-05,
      "loss": 0.98,
      "step": 4112
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.781041464620591e-05,
      "loss": 0.9491,
      "step": 4113
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7809339489566934e-05,
      "loss": 1.0348,
      "step": 4114
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7808264081119636e-05,
      "loss": 0.9407,
      "step": 4115
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.780718842087589e-05,
      "loss": 1.1868,
      "step": 4116
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7806112508847565e-05,
      "loss": 0.9829,
      "step": 4117
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.780503634504655e-05,
      "loss": 0.8664,
      "step": 4118
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.780395992948472e-05,
      "loss": 0.8791,
      "step": 4119
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.780288326217396e-05,
      "loss": 1.1756,
      "step": 4120
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7801806343126145e-05,
      "loss": 0.9827,
      "step": 4121
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.780072917235318e-05,
      "loss": 0.966,
      "step": 4122
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.779965174986694e-05,
      "loss": 1.0123,
      "step": 4123
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.779857407567933e-05,
      "loss": 1.0559,
      "step": 4124
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7797496149802256e-05,
      "loss": 0.9675,
      "step": 4125
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.77964179722476e-05,
      "loss": 1.0491,
      "step": 4126
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.779533954302727e-05,
      "loss": 0.9764,
      "step": 4127
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7794260862153185e-05,
      "loss": 0.8894,
      "step": 4128
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.779318192963723e-05,
      "loss": 1.0676,
      "step": 4129
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.779210274549134e-05,
      "loss": 0.874,
      "step": 4130
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.779102330972741e-05,
      "loss": 0.9702,
      "step": 4131
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.778994362235736e-05,
      "loss": 0.9582,
      "step": 4132
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7788863683393125e-05,
      "loss": 1.019,
      "step": 4133
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7787783492846605e-05,
      "loss": 0.8474,
      "step": 4134
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.778670305072974e-05,
      "loss": 1.1053,
      "step": 4135
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.778562235705446e-05,
      "loss": 1.0272,
      "step": 4136
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.778454141183268e-05,
      "loss": 1.0,
      "step": 4137
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.778346021507634e-05,
      "loss": 1.0282,
      "step": 4138
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7782378766797384e-05,
      "loss": 1.0197,
      "step": 4139
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.778129706700775e-05,
      "loss": 0.3319,
      "step": 4140
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7780215115719366e-05,
      "loss": 0.3518,
      "step": 4141
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.777913291294419e-05,
      "loss": 0.8714,
      "step": 4142
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7778050458694164e-05,
      "loss": 0.901,
      "step": 4143
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.777696775298124e-05,
      "loss": 1.0394,
      "step": 4144
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.777588479581736e-05,
      "loss": 1.0694,
      "step": 4145
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.777480158721449e-05,
      "loss": 1.0759,
      "step": 4146
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7773718127184595e-05,
      "loss": 1.0515,
      "step": 4147
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7772634415739624e-05,
      "loss": 0.9812,
      "step": 4148
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.777155045289155e-05,
      "loss": 0.6048,
      "step": 4149
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.777046623865233e-05,
      "loss": 1.093,
      "step": 4150
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.776938177303394e-05,
      "loss": 1.0769,
      "step": 4151
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.776829705604835e-05,
      "loss": 0.9802,
      "step": 4152
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7767212087707536e-05,
      "loss": 0.8252,
      "step": 4153
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.776612686802347e-05,
      "loss": 1.0985,
      "step": 4154
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.776504139700814e-05,
      "loss": 0.9236,
      "step": 4155
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.776395567467353e-05,
      "loss": 1.0111,
      "step": 4156
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.776286970103162e-05,
      "loss": 0.9985,
      "step": 4157
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.77617834760944e-05,
      "loss": 0.9961,
      "step": 4158
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.776069699987386e-05,
      "loss": 0.9182,
      "step": 4159
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.775961027238201e-05,
      "loss": 0.7704,
      "step": 4160
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.775852329363082e-05,
      "loss": 0.9009,
      "step": 4161
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.775743606363231e-05,
      "loss": 1.043,
      "step": 4162
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7756348582398476e-05,
      "loss": 1.0638,
      "step": 4163
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7755260849941324e-05,
      "loss": 0.9212,
      "step": 4164
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.775417286627287e-05,
      "loss": 0.9913,
      "step": 4165
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.775308463140511e-05,
      "loss": 0.9522,
      "step": 4166
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.775199614535007e-05,
      "loss": 0.8828,
      "step": 4167
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.775090740811976e-05,
      "loss": 0.94,
      "step": 4168
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.77498184197262e-05,
      "loss": 0.9346,
      "step": 4169
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7748729180181416e-05,
      "loss": 0.914,
      "step": 4170
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.774763968949743e-05,
      "loss": 0.8698,
      "step": 4171
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7746549947686274e-05,
      "loss": 1.3942,
      "step": 4172
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.774545995475997e-05,
      "loss": 0.9829,
      "step": 4173
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7744369710730565e-05,
      "loss": 0.9003,
      "step": 4174
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.774327921561008e-05,
      "loss": 1.0642,
      "step": 4175
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7742188469410554e-05,
      "loss": 0.8619,
      "step": 4176
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.774109747214403e-05,
      "loss": 0.8739,
      "step": 4177
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.774000622382258e-05,
      "loss": 0.9422,
      "step": 4178
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.77389147244582e-05,
      "loss": 0.9758,
      "step": 4179
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.773782297406298e-05,
      "loss": 0.8474,
      "step": 4180
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7736730972648965e-05,
      "loss": 1.0169,
      "step": 4181
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7735638720228207e-05,
      "loss": 5.5461,
      "step": 4182
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.773454621681275e-05,
      "loss": 5.8679,
      "step": 4183
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.773345346241467e-05,
      "loss": 5.8129,
      "step": 4184
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.773236045704603e-05,
      "loss": 4.387,
      "step": 4185
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.77312672007189e-05,
      "loss": 0.9989,
      "step": 4186
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7730173693445334e-05,
      "loss": 0.9887,
      "step": 4187
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.772907993523742e-05,
      "loss": 0.9112,
      "step": 4188
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.772798592610722e-05,
      "loss": 0.667,
      "step": 4189
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7726891666066825e-05,
      "loss": 0.8889,
      "step": 4190
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7725797155128306e-05,
      "loss": 1.0545,
      "step": 4191
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.772470239330375e-05,
      "loss": 0.7622,
      "step": 4192
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.772360738060524e-05,
      "loss": 0.9034,
      "step": 4193
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7722512117044865e-05,
      "loss": 0.8327,
      "step": 4194
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.772141660263471e-05,
      "loss": 1.104,
      "step": 4195
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.772032083738689e-05,
      "loss": 0.9917,
      "step": 4196
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.771922482131348e-05,
      "loss": 0.9108,
      "step": 4197
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.77181285544266e-05,
      "loss": 0.6669,
      "step": 4198
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7717032036738324e-05,
      "loss": 0.921,
      "step": 4199
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.771593526826078e-05,
      "loss": 1.0121,
      "step": 4200
    },
    {
      "epoch": 0.14,
      "eval_loss": 0.9520307779312134,
      "eval_runtime": 233.5444,
      "eval_samples_per_second": 9.21,
      "eval_steps_per_second": 2.304,
      "step": 4200
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7714838249006073e-05,
      "loss": 1.1156,
      "step": 4201
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.771374097898631e-05,
      "loss": 1.1067,
      "step": 4202
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.77126434582136e-05,
      "loss": 0.9671,
      "step": 4203
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.771154568670007e-05,
      "loss": 0.9353,
      "step": 4204
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7710447664457826e-05,
      "loss": 1.081,
      "step": 4205
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7709349391498996e-05,
      "loss": 0.9819,
      "step": 4206
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.770825086783571e-05,
      "loss": 1.0133,
      "step": 4207
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.770715209348009e-05,
      "loss": 1.0835,
      "step": 4208
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.770605306844427e-05,
      "loss": 0.9317,
      "step": 4209
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.770495379274038e-05,
      "loss": 1.0324,
      "step": 4210
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.770385426638056e-05,
      "loss": 0.9782,
      "step": 4211
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.770275448937694e-05,
      "loss": 1.0791,
      "step": 4212
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.770165446174166e-05,
      "loss": 0.9878,
      "step": 4213
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7700554183486875e-05,
      "loss": 1.2175,
      "step": 4214
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.769945365462473e-05,
      "loss": 1.0105,
      "step": 4215
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.769835287516736e-05,
      "loss": 0.7862,
      "step": 4216
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.769725184512694e-05,
      "loss": 0.8381,
      "step": 4217
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7696150564515607e-05,
      "loss": 0.9187,
      "step": 4218
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.769504903334553e-05,
      "loss": 0.9704,
      "step": 4219
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.769394725162886e-05,
      "loss": 1.0346,
      "step": 4220
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.769284521937776e-05,
      "loss": 0.949,
      "step": 4221
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.76917429366044e-05,
      "loss": 1.0701,
      "step": 4222
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.769064040332096e-05,
      "loss": 0.9202,
      "step": 4223
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7689537619539595e-05,
      "loss": 1.0761,
      "step": 4224
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.768843458527249e-05,
      "loss": 1.1261,
      "step": 4225
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.768733130053182e-05,
      "loss": 0.872,
      "step": 4226
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.768622776532975e-05,
      "loss": 0.9883,
      "step": 4227
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7685123979678495e-05,
      "loss": 0.975,
      "step": 4228
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.768401994359021e-05,
      "loss": 1.1325,
      "step": 4229
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7682915657077086e-05,
      "loss": 0.8377,
      "step": 4230
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.768181112015133e-05,
      "loss": 0.9961,
      "step": 4231
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.768070633282513e-05,
      "loss": 1.0731,
      "step": 4232
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.767960129511069e-05,
      "loss": 4.3866,
      "step": 4233
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7678496007020185e-05,
      "loss": 1.0277,
      "step": 4234
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.767739046856584e-05,
      "loss": 1.1801,
      "step": 4235
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.767628467975984e-05,
      "loss": 1.0555,
      "step": 4236
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.767517864061441e-05,
      "loss": 0.9532,
      "step": 4237
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7674072351141765e-05,
      "loss": 0.9414,
      "step": 4238
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.76729658113541e-05,
      "loss": 0.8824,
      "step": 4239
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.767185902126364e-05,
      "loss": 0.7845,
      "step": 4240
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.76707519808826e-05,
      "loss": 0.8281,
      "step": 4241
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7669644690223204e-05,
      "loss": 1.0733,
      "step": 4242
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7668537149297676e-05,
      "loss": 0.983,
      "step": 4243
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.766742935811824e-05,
      "loss": 0.9007,
      "step": 4244
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.766632131669714e-05,
      "loss": 0.9816,
      "step": 4245
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.766521302504658e-05,
      "loss": 0.9827,
      "step": 4246
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7664104483178826e-05,
      "loss": 1.0988,
      "step": 4247
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.76629956911061e-05,
      "loss": 0.9337,
      "step": 4248
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7661886648840634e-05,
      "loss": 1.174,
      "step": 4249
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7660777356394696e-05,
      "loss": 1.3472,
      "step": 4250
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.765966781378051e-05,
      "loss": 1.2801,
      "step": 4251
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7658558021010344e-05,
      "loss": 1.0831,
      "step": 4252
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7657447978096426e-05,
      "loss": 1.1148,
      "step": 4253
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7656337685051036e-05,
      "loss": 1.0907,
      "step": 4254
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7655227141886406e-05,
      "loss": 0.9117,
      "step": 4255
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.765411634861482e-05,
      "loss": 1.0211,
      "step": 4256
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.765300530524852e-05,
      "loss": 1.0513,
      "step": 4257
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.765189401179979e-05,
      "loss": 0.9528,
      "step": 4258
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.765078246828089e-05,
      "loss": 4.025,
      "step": 4259
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.76496706747041e-05,
      "loss": 3.4176,
      "step": 4260
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.764855863108166e-05,
      "loss": 0.9416,
      "step": 4261
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7647446337425895e-05,
      "loss": 1.0091,
      "step": 4262
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.764633379374905e-05,
      "loss": 1.9762,
      "step": 4263
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7645221000063426e-05,
      "loss": 1.2266,
      "step": 4264
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.764410795638129e-05,
      "loss": 0.9607,
      "step": 4265
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.764299466271494e-05,
      "loss": 1.019,
      "step": 4266
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.764188111907667e-05,
      "loss": 1.8794,
      "step": 4267
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7640767325478776e-05,
      "loss": 0.9929,
      "step": 4268
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7639653281933536e-05,
      "loss": 1.0878,
      "step": 4269
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.763853898845326e-05,
      "loss": 0.9166,
      "step": 4270
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.763742444505026e-05,
      "loss": 0.7612,
      "step": 4271
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7636309651736824e-05,
      "loss": 0.924,
      "step": 4272
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7635194608525265e-05,
      "loss": 1.0077,
      "step": 4273
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7634079315427895e-05,
      "loss": 0.9866,
      "step": 4274
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7632963772457024e-05,
      "loss": 0.9305,
      "step": 4275
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7631847979624966e-05,
      "loss": 1.066,
      "step": 4276
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.763073193694404e-05,
      "loss": 0.9665,
      "step": 4277
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.762961564442656e-05,
      "loss": 0.8376,
      "step": 4278
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.762849910208487e-05,
      "loss": 0.9064,
      "step": 4279
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7627382309931275e-05,
      "loss": 0.8914,
      "step": 4280
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.762626526797811e-05,
      "loss": 0.9518,
      "step": 4281
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7625147976237714e-05,
      "loss": 0.969,
      "step": 4282
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.762403043472241e-05,
      "loss": 1.2362,
      "step": 4283
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7622912643444554e-05,
      "loss": 0.7914,
      "step": 4284
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.762179460241646e-05,
      "loss": 1.0008,
      "step": 4285
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.762067631165049e-05,
      "loss": 1.1253,
      "step": 4286
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7619557771158985e-05,
      "loss": 1.0116,
      "step": 4287
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.761843898095429e-05,
      "loss": 1.0548,
      "step": 4288
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.761731994104877e-05,
      "loss": 1.1798,
      "step": 4289
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7616200651454754e-05,
      "loss": 3.3808,
      "step": 4290
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7615081112184614e-05,
      "loss": 5.7432,
      "step": 4291
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.761396132325071e-05,
      "loss": 1.1044,
      "step": 4292
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7612841284665396e-05,
      "loss": 0.5005,
      "step": 4293
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.761172099644105e-05,
      "loss": 1.0412,
      "step": 4294
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7610600458590036e-05,
      "loss": 1.0279,
      "step": 4295
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.760947967112471e-05,
      "loss": 0.9855,
      "step": 4296
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.760835863405746e-05,
      "loss": 0.9051,
      "step": 4297
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.760723734740066e-05,
      "loss": 0.9362,
      "step": 4298
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7606115811166685e-05,
      "loss": 1.2061,
      "step": 4299
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.760499402536792e-05,
      "loss": 0.9887,
      "step": 4300
    },
    {
      "epoch": 0.14,
      "eval_loss": 0.9461297392845154,
      "eval_runtime": 226.6588,
      "eval_samples_per_second": 9.49,
      "eval_steps_per_second": 2.374,
      "step": 4300
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.760387199001674e-05,
      "loss": 0.9071,
      "step": 4301
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7602749705125545e-05,
      "loss": 0.9697,
      "step": 4302
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.760162717070671e-05,
      "loss": 0.9687,
      "step": 4303
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.760050438677265e-05,
      "loss": 0.8963,
      "step": 4304
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.759938135333574e-05,
      "loss": 0.8368,
      "step": 4305
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.759825807040838e-05,
      "loss": 1.0011,
      "step": 4306
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7597134538002977e-05,
      "loss": 1.0415,
      "step": 4307
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7596010756131937e-05,
      "loss": 0.9815,
      "step": 4308
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7594886724807654e-05,
      "loss": 0.9182,
      "step": 4309
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.759376244404255e-05,
      "loss": 0.9525,
      "step": 4310
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.759263791384903e-05,
      "loss": 0.9585,
      "step": 4311
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7591513134239505e-05,
      "loss": 0.9604,
      "step": 4312
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.75903881052264e-05,
      "loss": 0.9944,
      "step": 4313
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.758926282682214e-05,
      "loss": 1.109,
      "step": 4314
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7588137299039135e-05,
      "loss": 0.9547,
      "step": 4315
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.758701152188981e-05,
      "loss": 0.8075,
      "step": 4316
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.758588549538661e-05,
      "loss": 0.9984,
      "step": 4317
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.758475921954194e-05,
      "loss": 1.0055,
      "step": 4318
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.758363269436825e-05,
      "loss": 0.9597,
      "step": 4319
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7582505919877985e-05,
      "loss": 0.9741,
      "step": 4320
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7581378896083566e-05,
      "loss": 0.9723,
      "step": 4321
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7580251622997444e-05,
      "loss": 0.8634,
      "step": 4322
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.757912410063207e-05,
      "loss": 0.9821,
      "step": 4323
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7577996328999875e-05,
      "loss": 0.9698,
      "step": 4324
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.757686830811332e-05,
      "loss": 0.9627,
      "step": 4325
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.757574003798486e-05,
      "loss": 1.0891,
      "step": 4326
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7574611518626944e-05,
      "loss": 0.9839,
      "step": 4327
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.757348275005204e-05,
      "loss": 0.9643,
      "step": 4328
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.757235373227259e-05,
      "loss": 1.0198,
      "step": 4329
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.757122446530108e-05,
      "loss": 0.9534,
      "step": 4330
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.757009494914997e-05,
      "loss": 0.9458,
      "step": 4331
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7568965183831726e-05,
      "loss": 0.9433,
      "step": 4332
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.756783516935882e-05,
      "loss": 0.9032,
      "step": 4333
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.756670490574373e-05,
      "loss": 1.0267,
      "step": 4334
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7565574392998926e-05,
      "loss": 0.9774,
      "step": 4335
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.75644436311369e-05,
      "loss": 0.9119,
      "step": 4336
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.756331262017013e-05,
      "loss": 0.8529,
      "step": 4337
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.75621813601111e-05,
      "loss": 0.9942,
      "step": 4338
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.75610498509723e-05,
      "loss": 0.693,
      "step": 4339
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.755991809276623e-05,
      "loss": 0.9156,
      "step": 4340
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.755878608550538e-05,
      "loss": 0.6735,
      "step": 4341
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.755765382920224e-05,
      "loss": 0.7465,
      "step": 4342
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.755652132386931e-05,
      "loss": 0.9577,
      "step": 4343
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.755538856951909e-05,
      "loss": 0.97,
      "step": 4344
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.75542555661641e-05,
      "loss": 0.8671,
      "step": 4345
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.755312231381684e-05,
      "loss": 0.9456,
      "step": 4346
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7551988812489814e-05,
      "loss": 0.903,
      "step": 4347
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.755085506219555e-05,
      "loss": 0.8685,
      "step": 4348
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.754972106294655e-05,
      "loss": 0.9075,
      "step": 4349
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.754858681475534e-05,
      "loss": 1.0502,
      "step": 4350
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.754745231763444e-05,
      "loss": 0.9324,
      "step": 4351
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7546317571596375e-05,
      "loss": 0.8664,
      "step": 4352
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.754518257665368e-05,
      "loss": 0.9946,
      "step": 4353
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.754404733281887e-05,
      "loss": 0.881,
      "step": 4354
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.754291184010449e-05,
      "loss": 0.8331,
      "step": 4355
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7541776098523065e-05,
      "loss": 0.531,
      "step": 4356
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.754064010808714e-05,
      "loss": 0.3812,
      "step": 4357
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.753950386880926e-05,
      "loss": 0.6121,
      "step": 4358
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.753836738070196e-05,
      "loss": 0.9671,
      "step": 4359
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7537230643777795e-05,
      "loss": 0.7983,
      "step": 4360
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.75360936580493e-05,
      "loss": 0.7038,
      "step": 4361
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.753495642352904e-05,
      "loss": 0.9051,
      "step": 4362
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7533818940229575e-05,
      "loss": 0.8742,
      "step": 4363
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7532681208163444e-05,
      "loss": 0.9837,
      "step": 4364
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7531543227343224e-05,
      "loss": 1.0764,
      "step": 4365
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.753040499778147e-05,
      "loss": 0.9412,
      "step": 4366
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.752926651949075e-05,
      "loss": 0.991,
      "step": 4367
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7528127792483616e-05,
      "loss": 0.9322,
      "step": 4368
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.752698881677267e-05,
      "loss": 0.3394,
      "step": 4369
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7525849592370464e-05,
      "loss": 0.3587,
      "step": 4370
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7524710119289584e-05,
      "loss": 0.9842,
      "step": 4371
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7523570397542606e-05,
      "loss": 0.9927,
      "step": 4372
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7522430427142106e-05,
      "loss": 1.1401,
      "step": 4373
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.752129020810068e-05,
      "loss": 1.0205,
      "step": 4374
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.752014974043092e-05,
      "loss": 0.8356,
      "step": 4375
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7519009024145394e-05,
      "loss": 0.8787,
      "step": 4376
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.751786805925672e-05,
      "loss": 0.7131,
      "step": 4377
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.751672684577747e-05,
      "loss": 1.1074,
      "step": 4378
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7515585383720266e-05,
      "loss": 1.0479,
      "step": 4379
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.751444367309769e-05,
      "loss": 0.9372,
      "step": 4380
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7513301713922364e-05,
      "loss": 0.71,
      "step": 4381
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7512159506206876e-05,
      "loss": 1.0002,
      "step": 4382
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.751101704996386e-05,
      "loss": 1.0938,
      "step": 4383
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.75098743452059e-05,
      "loss": 1.03,
      "step": 4384
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.750873139194563e-05,
      "loss": 0.8178,
      "step": 4385
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7507588190195664e-05,
      "loss": 0.79,
      "step": 4386
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7506444739968616e-05,
      "loss": 1.1195,
      "step": 4387
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7505301041277125e-05,
      "loss": 1.6832,
      "step": 4388
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.75041570941338e-05,
      "loss": 0.8637,
      "step": 4389
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.750301289855128e-05,
      "loss": 0.9509,
      "step": 4390
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7501868454542185e-05,
      "loss": 0.9874,
      "step": 4391
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.750072376211916e-05,
      "loss": 1.1699,
      "step": 4392
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.749957882129485e-05,
      "loss": 1.3268,
      "step": 4393
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.749843363208187e-05,
      "loss": 0.8381,
      "step": 4394
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.749728819449289e-05,
      "loss": 0.9545,
      "step": 4395
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.749614250854054e-05,
      "loss": 0.9663,
      "step": 4396
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.749499657423746e-05,
      "loss": 0.9484,
      "step": 4397
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.749385039159632e-05,
      "loss": 0.8268,
      "step": 4398
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7492703960629764e-05,
      "loss": 0.9759,
      "step": 4399
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.749155728135045e-05,
      "loss": 0.8732,
      "step": 4400
    },
    {
      "epoch": 0.15,
      "eval_loss": 0.9317440986633301,
      "eval_runtime": 174.9429,
      "eval_samples_per_second": 12.295,
      "eval_steps_per_second": 2.052,
      "step": 4400
    }
  ],
  "max_steps": 30000,
  "num_train_epochs": 9223372036854775807,
  "total_flos": 1.7843993400311808e+18,
  "trial_name": null,
  "trial_params": null
}