all_results.json

{
    "epoch": 2.983957219251337,
    "eval_logits/chosen": -2.267369270324707,
    "eval_logits/rejected": -2.240177869796753,
    "eval_logps/chosen": -246.88827514648438,
    "eval_logps/rejected": -233.59410095214844,
    "eval_loss": 0.5221481323242188,
    "eval_rewards/accuracies": 0.7827380895614624,
    "eval_rewards/chosen": 0.8510617017745972,
    "eval_rewards/margins": 2.6143014430999756,
    "eval_rewards/rejected": -1.763239860534668,
    "eval_runtime": 178.6162,
    "eval_samples_per_second": 14.892,
    "eval_steps_per_second": 0.235,
    "total_flos": 3289753017384960.0,
    "train_loss": 0.3371215338775334,
    "train_runtime": 10383.8324,
    "train_samples_per_second": 6.914,
    "train_steps_per_second": 0.027
}