trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.951871657754011,
  "eval_steps": 500,
  "global_step": 138,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.79,
      "learning_rate": 0.00016290322580645163,
      "loss": 0.7756,
      "step": 37
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0001032258064516129,
      "loss": 0.6459,
      "step": 74
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.3548387096774194e-05,
      "loss": 0.5823,
      "step": 111
    }
  ],
  "logging_steps": 37,
  "max_steps": 138,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 4.487438068088832e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}