checkpoint-4/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.25,
  "global_step": 4,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06,
      "learning_rate": 0.1,
      "loss": 10.9672,
      "theoretical_loss": 35.35283692336181,
      "tokens_seen": 8192
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.09333333333333334,
      "loss": 10.9619,
      "theoretical_loss": 29.541230243953947,
      "tokens_seen": 16384
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.09333333333333334,
      "loss": 15.0187,
      "theoretical_loss": 26.629312696422083,
      "tokens_seen": 24576
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.08666666666666667,
      "loss": 16.1625,
      "theoretical_loss": 24.7548431869014,
      "tokens_seen": 32768
    }
  ],
  "max_steps": 16,
  "num_train_epochs": 9223372036854775807,
  "total_flos": 16722690048000.0,
  "trial_name": null,
  "trial_params": null
}