{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 28.0,
  "global_step": 14252,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "learning_rate": 1.9801178781925346e-05,
      "loss": 2.0175,
      "step": 509
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.9601178781925344e-05,
      "loss": 1.0916,
      "step": 1018
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.9401571709233793e-05,
      "loss": 0.9,
      "step": 1527
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.920196463654224e-05,
      "loss": 0.7942,
      "step": 2036
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.900235756385069e-05,
      "loss": 0.7212,
      "step": 2545
    },
    {
      "epoch": 6.0,
      "learning_rate": 1.8803536345776033e-05,
      "loss": 0.6697,
      "step": 3054
    },
    {
      "epoch": 7.0,
      "learning_rate": 1.860432220039293e-05,
      "loss": 0.6265,
      "step": 3563
    },
    {
      "epoch": 8.0,
      "learning_rate": 1.8407072691552063e-05,
      "loss": 0.6136,
      "step": 4072
    },
    {
      "epoch": 9.0,
      "learning_rate": 1.8207465618860512e-05,
      "loss": 0.8441,
      "step": 4581
    },
    {
      "epoch": 10.0,
      "learning_rate": 1.800943025540275e-05,
      "loss": 1.0052,
      "step": 5090
    },
    {
      "epoch": 11.0,
      "learning_rate": 1.7810216110019647e-05,
      "loss": 0.9929,
      "step": 5599
    },
    {
      "epoch": 12.0,
      "learning_rate": 1.7610609037328093e-05,
      "loss": 0.983,
      "step": 6108
    },
    {
      "epoch": 13.0,
      "learning_rate": 1.7412180746561888e-05,
      "loss": 0.9804,
      "step": 6617
    },
    {
      "epoch": 14.0,
      "learning_rate": 1.7213359528487232e-05,
      "loss": 0.9818,
      "step": 7126
    },
    {
      "epoch": 15.0,
      "learning_rate": 1.701375245579568e-05,
      "loss": 0.9822,
      "step": 7635
    },
    {
      "epoch": 16.0,
      "learning_rate": 1.6814931237721025e-05,
      "loss": 0.9819,
      "step": 8144
    },
    {
      "epoch": 17.0,
      "learning_rate": 1.6614931237721024e-05,
      "loss": 0.9826,
      "step": 8653
    },
    {
      "epoch": 18.0,
      "learning_rate": 1.6416895874263262e-05,
      "loss": 0.9822,
      "step": 9162
    },
    {
      "epoch": 19.0,
      "learning_rate": 1.6218074656188606e-05,
      "loss": 0.9827,
      "step": 9671
    },
    {
      "epoch": 20.0,
      "learning_rate": 1.6019646365422398e-05,
      "loss": 0.9843,
      "step": 10180
    },
    {
      "epoch": 21.0,
      "learning_rate": 1.5820039292730847e-05,
      "loss": 0.9808,
      "step": 10689
    },
    {
      "epoch": 22.0,
      "learning_rate": 1.5620432220039293e-05,
      "loss": 0.9823,
      "step": 11198
    },
    {
      "epoch": 23.0,
      "learning_rate": 1.5420825147347742e-05,
      "loss": 0.9801,
      "step": 11707
    },
    {
      "epoch": 24.0,
      "learning_rate": 1.5222396856581534e-05,
      "loss": 0.9824,
      "step": 12216
    },
    {
      "epoch": 25.0,
      "learning_rate": 1.5022789783889981e-05,
      "loss": 0.9813,
      "step": 12725
    },
    {
      "epoch": 26.0,
      "learning_rate": 1.4823968565815325e-05,
      "loss": 0.9836,
      "step": 13234
    },
    {
      "epoch": 27.0,
      "learning_rate": 1.4624361493123773e-05,
      "loss": 0.9838,
      "step": 13743
    },
    {
      "epoch": 28.0,
      "learning_rate": 1.4425933202357564e-05,
      "loss": 0.9803,
      "step": 14252
    }
  ],
  "max_steps": 50900,
  "num_train_epochs": 100,
  "total_flos": 3675621063214080.0,
  "trial_name": null,
  "trial_params": null
}