checkpoint-1000/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.24393218685205512,
  "eval_steps": 1000,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.24,
      "learning_rate": 4.30097988740375e-05,
      "logits/chosen": 0.4760148227214813,
      "logits/rejected": 0.18961936235427856,
      "logps/chosen": -138.3477783203125,
      "logps/rejected": -107.43299865722656,
      "loss": 0.0534,
      "rewards/accuracies": 0.9904999732971191,
      "rewards/chosen": -1.208303689956665,
      "rewards/margins": 6.7370734214782715,
      "rewards/rejected": -7.945377349853516,
      "step": 1000
    },
    {
      "epoch": 0.24,
      "eval_logits/chosen": 0.5734850168228149,
      "eval_logits/rejected": 0.30642059445381165,
      "eval_logps/chosen": -140.49411010742188,
      "eval_logps/rejected": -130.59207153320312,
      "eval_loss": 0.021729281172156334,
      "eval_rewards/accuracies": 0.994514524936676,
      "eval_rewards/chosen": -1.671390414237976,
      "eval_rewards/margins": 8.56452751159668,
      "eval_rewards/rejected": -10.235918045043945,
      "eval_runtime": 379.8081,
      "eval_samples_per_second": 9.597,
      "eval_steps_per_second": 4.8,
      "step": 1000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 4099,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 4.588610344556954e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}