all_results.json

{
    "epoch": 2.986666666666667,
    "eval_logits/chosen": 0.13058120012283325,
    "eval_logits/rejected": 0.3316400647163391,
    "eval_logps/chosen": -1.0981853008270264,
    "eval_logps/rejected": -1.93387770652771,
    "eval_loss": 1.138285517692566,
    "eval_odds_ratio_loss": 10.197811126708984,
    "eval_rewards/accuracies": 0.8399999737739563,
    "eval_rewards/chosen": -0.10981854796409607,
    "eval_rewards/margins": 0.0835692435503006,
    "eval_rewards/rejected": -0.19338779151439667,
    "eval_runtime": 3.5214,
    "eval_samples_per_second": 14.199,
    "eval_sft_loss": 0.11850442737340927,
    "eval_steps_per_second": 7.099,
    "total_flos": 1.79183583387648e+16,
    "train_loss": 1.8150039003008889,
    "train_runtime": 321.7396,
    "train_samples_per_second": 4.196,
    "train_steps_per_second": 0.261
}