{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 96,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.9785,
      "step": 1
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 1.0381,
      "step": 2
    },
    {
      "epoch": 0.19,
      "learning_rate": 2e-05,
      "loss": 0.5466,
      "step": 3
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.999429490929718e-05,
      "loss": 2.5332,
      "step": 4
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9977186146800707e-05,
      "loss": 0.4666,
      "step": 5
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.994869323391895e-05,
      "loss": 0.2064,
      "step": 6
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.990884868158239e-05,
      "loss": 0.2202,
      "step": 7
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.985769795314804e-05,
      "loss": 0.1354,
      "step": 8
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.9795299412524948e-05,
      "loss": 0.2023,
      "step": 9
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.9721724257579907e-05,
      "loss": 0.1093,
      "step": 10
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.963705643889941e-05,
      "loss": 0.126,
      "step": 11
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.954139256400049e-05,
      "loss": 0.1307,
      "step": 12
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.9434841787099804e-05,
      "loss": 0.1108,
      "step": 13
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.9317525684566686e-05,
      "loss": 0.1157,
      "step": 14
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.918957811620231e-05,
      "loss": 0.1541,
      "step": 15
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.9051145072503216e-05,
      "loss": 0.1068,
      "step": 16
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.8902384508083518e-05,
      "loss": 0.1263,
      "step": 17
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.8743466161445823e-05,
      "loss": 0.1382,
      "step": 18
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.857457136130651e-05,
      "loss": 0.1142,
      "step": 19
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.839589281969639e-05,
      "loss": 0.1077,
      "step": 20
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.8207634412072765e-05,
      "loss": 0.116,
      "step": 21
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.8010010944693846e-05,
      "loss": 0.1183,
      "step": 22
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.780324790952092e-05,
      "loss": 0.1205,
      "step": 23
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.758758122692791e-05,
      "loss": 0.1078,
      "step": 24
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.7363256976511972e-05,
      "loss": 0.1143,
      "step": 25
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.7130531116312202e-05,
      "loss": 0.1083,
      "step": 26
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.688966919075687e-05,
      "loss": 0.1096,
      "step": 27
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.6640946027672395e-05,
      "loss": 0.0986,
      "step": 28
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.6384645424699835e-05,
      "loss": 0.11,
      "step": 29
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.612105982547663e-05,
      "loss": 0.0924,
      "step": 30
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.5850489985953076e-05,
      "loss": 0.1262,
      "step": 31
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.5573244631224364e-05,
      "loss": 0.0916,
      "step": 32
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5289640103269626e-05,
      "loss": 0.0937,
      "step": 33
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.1017,
      "step": 34
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.4704654806027558e-05,
      "loss": 0.0928,
      "step": 35
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.4403941515576344e-05,
      "loss": 0.1088,
      "step": 36
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.4098203247965876e-05,
      "loss": 0.1034,
      "step": 37
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.3787788856105762e-05,
      "loss": 0.1024,
      "step": 38
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.3473052528448203e-05,
      "loss": 0.0874,
      "step": 39
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.3154353384852559e-05,
      "loss": 0.097,
      "step": 40
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.283205506682304e-05,
      "loss": 0.0948,
      "step": 41
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.2506525322587207e-05,
      "loss": 0.0788,
      "step": 42
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.2178135587488515e-05,
      "loss": 0.087,
      "step": 43
    },
    {
      "epoch": 2.75,
      "learning_rate": 1.1847260560171895e-05,
      "loss": 0.0866,
      "step": 44
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.1514277775045768e-05,
      "loss": 0.0992,
      "step": 45
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.1179567171508463e-05,
      "loss": 0.0916,
      "step": 46
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.0843510660430447e-05,
      "loss": 0.0786,
      "step": 47
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.0506491688387128e-05,
      "loss": 0.0915,
      "step": 48
    },
    {
      "epoch": 3.06,
      "learning_rate": 1.0168894800139311e-05,
      "loss": 0.0815,
      "step": 49
    },
    {
      "epoch": 3.12,
      "learning_rate": 9.83110519986069e-06,
      "loss": 0.0762,
      "step": 50
    },
    {
      "epoch": 3.19,
      "learning_rate": 9.493508311612874e-06,
      "loss": 0.0908,
      "step": 51
    },
    {
      "epoch": 3.25,
      "learning_rate": 9.156489339569555e-06,
      "loss": 0.0753,
      "step": 52
    },
    {
      "epoch": 3.31,
      "learning_rate": 8.820432828491542e-06,
      "loss": 0.0576,
      "step": 53
    },
    {
      "epoch": 3.38,
      "learning_rate": 8.485722224954237e-06,
      "loss": 0.0742,
      "step": 54
    },
    {
      "epoch": 3.44,
      "learning_rate": 8.15273943982811e-06,
      "loss": 0.0976,
      "step": 55
    },
    {
      "epoch": 3.5,
      "learning_rate": 7.821864412511485e-06,
      "loss": 0.0964,
      "step": 56
    },
    {
      "epoch": 3.56,
      "learning_rate": 7.493474677412795e-06,
      "loss": 0.0894,
      "step": 57
    },
    {
      "epoch": 3.62,
      "learning_rate": 7.16794493317696e-06,
      "loss": 0.0877,
      "step": 58
    },
    {
      "epoch": 3.69,
      "learning_rate": 6.845646615147445e-06,
      "loss": 0.0871,
      "step": 59
    },
    {
      "epoch": 3.75,
      "learning_rate": 6.526947471551799e-06,
      "loss": 0.0714,
      "step": 60
    },
    {
      "epoch": 3.81,
      "learning_rate": 6.21221114389424e-06,
      "loss": 0.0806,
      "step": 61
    },
    {
      "epoch": 3.88,
      "learning_rate": 5.901796752034128e-06,
      "loss": 0.0879,
      "step": 62
    },
    {
      "epoch": 3.94,
      "learning_rate": 5.5960584844236565e-06,
      "loss": 0.0731,
      "step": 63
    },
    {
      "epoch": 4.0,
      "learning_rate": 5.295345193972445e-06,
      "loss": 0.0867,
      "step": 64
    },
    {
      "epoch": 4.06,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.0633,
      "step": 65
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.710359896730379e-06,
      "loss": 0.0632,
      "step": 66
    },
    {
      "epoch": 4.19,
      "learning_rate": 4.426755368775637e-06,
      "loss": 0.0824,
      "step": 67
    },
    {
      "epoch": 4.25,
      "learning_rate": 4.149510014046922e-06,
      "loss": 0.0883,
      "step": 68
    },
    {
      "epoch": 4.31,
      "learning_rate": 3.878940174523371e-06,
      "loss": 0.0598,
      "step": 69
    },
    {
      "epoch": 4.38,
      "learning_rate": 3.6153545753001663e-06,
      "loss": 0.0612,
      "step": 70
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.3590539723276083e-06,
      "loss": 0.0597,
      "step": 71
    },
    {
      "epoch": 4.5,
      "learning_rate": 3.110330809243134e-06,
      "loss": 0.0798,
      "step": 72
    },
    {
      "epoch": 4.56,
      "learning_rate": 2.869468883687798e-06,
      "loss": 0.0694,
      "step": 73
    },
    {
      "epoch": 4.62,
      "learning_rate": 2.6367430234880286e-06,
      "loss": 0.0791,
      "step": 74
    },
    {
      "epoch": 4.69,
      "learning_rate": 2.4124187730720916e-06,
      "loss": 0.0692,
      "step": 75
    },
    {
      "epoch": 4.75,
      "learning_rate": 2.196752090479083e-06,
      "loss": 0.0687,
      "step": 76
    },
    {
      "epoch": 4.81,
      "learning_rate": 1.9899890553061565e-06,
      "loss": 0.0582,
      "step": 77
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.7923655879272395e-06,
      "loss": 0.056,
      "step": 78
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.60410718030361e-06,
      "loss": 0.0554,
      "step": 79
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.425428638693489e-06,
      "loss": 0.0484,
      "step": 80
    },
    {
      "epoch": 5.06,
      "learning_rate": 1.2565338385541792e-06,
      "loss": 0.0637,
      "step": 81
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.097615491916485e-06,
      "loss": 0.0595,
      "step": 82
    },
    {
      "epoch": 5.19,
      "learning_rate": 9.488549274967873e-07,
      "loss": 0.0533,
      "step": 83
    },
    {
      "epoch": 5.25,
      "learning_rate": 8.10421883797694e-07,
      "loss": 0.0834,
      "step": 84
    },
    {
      "epoch": 5.31,
      "learning_rate": 6.824743154333157e-07,
      "loss": 0.0598,
      "step": 85
    },
    {
      "epoch": 5.38,
      "learning_rate": 5.651582129001987e-07,
      "loss": 0.0504,
      "step": 86
    },
    {
      "epoch": 5.44,
      "learning_rate": 4.5860743599951186e-07,
      "loss": 0.0688,
      "step": 87
    },
    {
      "epoch": 5.5,
      "learning_rate": 3.629435611005916e-07,
      "loss": 0.0621,
      "step": 88
    },
    {
      "epoch": 5.56,
      "learning_rate": 2.7827574242009434e-07,
      "loss": 0.0482,
      "step": 89
    },
    {
      "epoch": 5.62,
      "learning_rate": 2.0470058747505516e-07,
      "loss": 0.0515,
      "step": 90
    },
    {
      "epoch": 5.69,
      "learning_rate": 1.4230204685196202e-07,
      "loss": 0.0695,
      "step": 91
    },
    {
      "epoch": 5.75,
      "learning_rate": 9.11513184176116e-08,
      "loss": 0.051,
      "step": 92
    },
    {
      "epoch": 5.81,
      "learning_rate": 5.1306766081048456e-08,
      "loss": 0.0379,
      "step": 93
    },
    {
      "epoch": 5.88,
      "learning_rate": 2.2813853199292745e-08,
      "loss": 0.0565,
      "step": 94
    },
    {
      "epoch": 5.94,
      "learning_rate": 5.705090702819993e-09,
      "loss": 0.0524,
      "step": 95
    },
    {
      "epoch": 6.0,
      "learning_rate": 0.0,
      "loss": 0.0478,
      "step": 96
    },
    {
      "epoch": 6.0,
      "step": 96,
      "total_flos": 2277230247936.0,
      "train_loss": 0.14340813954671225,
      "train_runtime": 691.2623,
      "train_samples_per_second": 8.515,
      "train_steps_per_second": 0.139
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 96,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 800,
  "total_flos": 2277230247936.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}