all_results.json

{
    "epoch": 3.0,
    "eval_logits/chosen": -2.6046547889709473,
    "eval_logits/rejected": -2.605973482131958,
    "eval_logps/chosen": -194.06776428222656,
    "eval_logps/rejected": -211.20343017578125,
    "eval_loss": 0.5909408926963806,
    "eval_rewards/accuracies": 0.7265625,
    "eval_rewards/chosen": -2.798844337463379,
    "eval_rewards/margins": 1.4639229774475098,
    "eval_rewards/rejected": -4.2627668380737305,
    "eval_runtime": 126.0832,
    "eval_samples_per_second": 16.204,
    "eval_steps_per_second": 0.254,
    "total_flos": 2546799910846464.0,
    "train_loss": 0.29657687264046184,
    "train_runtime": 7656.725,
    "train_samples_per_second": 7.203,
    "train_steps_per_second": 0.028
}