trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 18,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.17,
      "grad_norm": 13.434505281320261,
      "learning_rate": 1e-05,
      "loss": 1.3367,
      "step": 1
    },
    {
      "epoch": 0.83,
      "grad_norm": 14.471944136941827,
      "learning_rate": 1.8314696123025456e-05,
      "loss": 1.3711,
      "step": 5
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.732906699180603,
      "eval_runtime": 4.1761,
      "eval_samples_per_second": 10.536,
      "eval_steps_per_second": 0.718,
      "step": 6
    },
    {
      "epoch": 1.67,
      "grad_norm": 4.3684518967471755,
      "learning_rate": 1e-05,
      "loss": 1.0734,
      "step": 10
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.6323952674865723,
      "eval_runtime": 3.9686,
      "eval_samples_per_second": 11.087,
      "eval_steps_per_second": 0.756,
      "step": 12
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.6854955998641517,
      "learning_rate": 1.6853038769745466e-06,
      "loss": 0.8291,
      "step": 15
    },
    {
      "epoch": 3.0,
      "eval_loss": 1.640893816947937,
      "eval_runtime": 3.9721,
      "eval_samples_per_second": 11.077,
      "eval_steps_per_second": 0.755,
      "step": 18
    },
    {
      "epoch": 3.0,
      "step": 18,
      "total_flos": 3716488888320.0,
      "train_loss": 1.0292662117216322,
      "train_runtime": 320.9926,
      "train_samples_per_second": 3.551,
      "train_steps_per_second": 0.056
    }
  ],
  "logging_steps": 5,
  "max_steps": 18,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 3716488888320.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}