trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.6052631578947368,
  "eval_steps": 500,
  "global_step": 92,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.16,
      "grad_norm": 0.7791975140571594,
      "learning_rate": 0.0002,
      "loss": 2.1532,
      "step": 25
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.18488548696041107,
      "learning_rate": 0.0002,
      "loss": 0.3532,
      "step": 50
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.22702211141586304,
      "learning_rate": 0.0002,
      "loss": 0.1791,
      "step": 75
    },
    {
      "epoch": 0.61,
      "step": 92,
      "total_flos": 5756148033159168.0,
      "train_loss": 0.7615472518879435,
      "train_runtime": 1644.6511,
      "train_samples_per_second": 0.443,
      "train_steps_per_second": 0.056
    }
  ],
  "logging_steps": 25,
  "max_steps": 92,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 25,
  "total_flos": 5756148033159168.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}