trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 114,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.0,
      "eval_loss": 3.9040439128875732,
      "eval_runtime": 17.2228,
      "eval_samples_per_second": 27.638,
      "eval_steps_per_second": 3.484,
      "step": 38
    },
    {
      "epoch": 2.0,
      "eval_loss": 3.8737525939941406,
      "eval_runtime": 20.5198,
      "eval_samples_per_second": 23.197,
      "eval_steps_per_second": 2.924,
      "step": 76
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 1.8681073188781738,
      "learning_rate": 6.140350877192982e-06,
      "loss": 3.8733,
      "step": 100
    }
  ],
  "logging_steps": 100,
  "max_steps": 114,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 235158582067200.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}