all_results.json

{
    "epoch": 2.997999555456768,
    "eval_logits/chosen": -0.3320940136909485,
    "eval_logits/rejected": -0.29884636402130127,
    "eval_logps/chosen": -0.9399133324623108,
    "eval_logps/rejected": -1.080655574798584,
    "eval_loss": 1.0073015689849854,
    "eval_odds_ratio_loss": 0.6738813519477844,
    "eval_rewards/accuracies": 0.515999972820282,
    "eval_rewards/chosen": -0.09399133920669556,
    "eval_rewards/margins": 0.01407422125339508,
    "eval_rewards/rejected": -0.10806556046009064,
    "eval_runtime": 185.8105,
    "eval_samples_per_second": 5.382,
    "eval_sft_loss": 0.9399133324623108,
    "eval_steps_per_second": 2.691,
    "total_flos": 1.8817568285770383e+18,
    "train_loss": 1.0353579054523618,
    "train_runtime": 16950.0138,
    "train_samples_per_second": 1.593,
    "train_steps_per_second": 0.099
}