checkpoint-500/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.4,
  "eval_steps": 200,
  "global_step": 500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.16,
      "learning_rate": 0.0,
      "loss": 292.3211,
      "step": 200
    },
    {
      "epoch": 0.16,
      "eval_loss": NaN,
      "eval_runtime": 67.797,
      "eval_samples_per_second": 3.687,
      "eval_steps_per_second": 0.472,
      "step": 200
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0,
      "loss": 76.016,
      "step": 400
    },
    {
      "epoch": 0.32,
      "eval_loss": NaN,
      "eval_runtime": 67.8532,
      "eval_samples_per_second": 3.684,
      "eval_steps_per_second": 0.472,
      "step": 400
    }
  ],
  "logging_steps": 200,
  "max_steps": 1250,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 4.0608404078592e+16,
  "trial_name": null,
  "trial_params": null
}