eval_results.json

{
    "epoch": 2.999297541394882,
    "eval_logits/chosen": 0.023057356476783752,
    "eval_logits/rejected": 0.08547364920377731,
    "eval_logps/chosen": -1.2622427940368652,
    "eval_logps/rejected": -1.3865184783935547,
    "eval_loss": 1.2624977827072144,
    "eval_rewards/accuracies": 0.5467358827590942,
    "eval_rewards/chosen": -1.2622427940368652,
    "eval_rewards/margins": 0.12427545338869095,
    "eval_rewards/rejected": -1.3865184783935547,
    "eval_runtime": 39.7519,
    "eval_samples": 1345,
    "eval_samples_per_second": 33.835,
    "eval_steps_per_second": 8.478
}