trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.7777777777777777,
  "eval_steps": 500,
  "global_step": 8,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.11,
      "learning_rate": 6.173165676349103e-05,
      "loss": 2.1888,
      "step": 5
    },
    {
      "epoch": 1.78,
      "step": 8,
      "total_flos": 2856099095248896.0,
      "train_loss": 2.1232523322105408,
      "train_runtime": 170.4879,
      "train_samples_per_second": 3.261,
      "train_steps_per_second": 0.047
    }
  ],
  "logging_steps": 5,
  "max_steps": 8,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "total_flos": 2856099095248896.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}