eval_results.json

{
    "epoch": 0.9982631930527722,
    "eval_logits/chosen": -1.3676064014434814,
    "eval_logits/rejected": -1.3771528005599976,
    "eval_logps/chosen": -1.543013095855713,
    "eval_logps/rejected": -2.0700173377990723,
    "eval_loss": 1.3837947845458984,
    "eval_rewards/accuracies": 0.827235758304596,
    "eval_rewards/chosen": -15.430130958557129,
    "eval_rewards/margins": 5.270040988922119,
    "eval_rewards/rejected": -20.700172424316406,
    "eval_runtime": 87.8145,
    "eval_samples": 1961,
    "eval_samples_per_second": 22.331,
    "eval_steps_per_second": 1.401
}