eval_results.json

{
    "epoch": 0.9999333733093477,
    "eval_logits/chosen": -0.9731917977333069,
    "eval_logits/rejected": -1.166278600692749,
    "eval_logps/chosen": -306.48046875,
    "eval_logps/rejected": -356.2023010253906,
    "eval_loss": 0.503829836845398,
    "eval_rewards/accuracies": 0.7933906316757202,
    "eval_rewards/chosen": -1.4382753372192383,
    "eval_rewards/margins": 0.6804661750793457,
    "eval_rewards/rejected": -2.118741512298584,
    "eval_runtime": 11557.6821,
    "eval_samples": 60035,
    "eval_samples_per_second": 5.194,
    "eval_steps_per_second": 1.299
}