eval_results.json

{
    "epoch": 1.9940933254577673,
    "eval_logits/chosen": -2.721290349960327,
    "eval_logits/rejected": -2.727541208267212,
    "eval_logps/chosen": -379.29864501953125,
    "eval_logps/rejected": -519.5195922851562,
    "eval_loss": 0.4716758131980896,
    "eval_rewards/accuracies": 0.6344696879386902,
    "eval_rewards/chosen": -1.6806795597076416,
    "eval_rewards/margins": 1.5150176286697388,
    "eval_rewards/rejected": -3.19569730758667,
    "eval_runtime": 226.304,
    "eval_samples": 3656,
    "eval_samples_per_second": 16.155,
    "eval_steps_per_second": 0.292
}