trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.96,
  "global_step": 15,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06,
      "learning_rate": 0,
      "loss": 3.1332,
      "step": 1
    },
    {
      "epoch": 0.96,
      "learning_rate": 5e-05,
      "loss": 3.2134,
      "step": 15
    },
    {
      "epoch": 0.96,
      "eval_accuracy": 0.040426829268292684,
      "eval_loss": 2.716796875,
      "eval_runtime": 9.2533,
      "eval_samples_per_second": 10.807,
      "eval_steps_per_second": 0.216,
      "step": 15
    },
    {
      "epoch": 0.96,
      "step": 15,
      "total_flos": 7938382233600.0,
      "train_loss": 3.2080670674641927,
      "train_runtime": 1485.1272,
      "train_samples_per_second": 0.673,
      "train_steps_per_second": 0.01
    }
  ],
  "max_steps": 15,
  "num_train_epochs": 1,
  "total_flos": 7938382233600.0,
  "trial_name": null,
  "trial_params": null
}