trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.768,
  "eval_steps": 100,
  "global_step": 3,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.26,
      "learning_rate": 5e-06,
      "logits/chosen": -2.7575411796569824,
      "logits/rejected": -2.648534059524536,
      "logps/chosen": -262.0839538574219,
      "logps/rejected": -262.1346435546875,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.77,
      "step": 3,
      "total_flos": 0.0,
      "train_loss": 0.69305553038915,
      "train_runtime": 253.0881,
      "train_samples_per_second": 3.951,
      "train_steps_per_second": 0.012
    }
  ],
  "logging_steps": 10,
  "max_steps": 3,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}