eval_results.json

{
    "epoch": 1.0,
    "eval_error": 0.23849666118621826,
    "eval_logits/chosen": -2.8272712230682373,
    "eval_logits/rejected": -2.8083930015563965,
    "eval_logps/chosen": -477.6735534667969,
    "eval_logps/rejected": -481.281494140625,
    "eval_loss": 0.8262702226638794,
    "eval_preference/chosen": 0.5308043956756592,
    "eval_preference/rejected": 0.46919554471969604,
    "eval_rewards/accuracies": 0.5099999904632568,
    "eval_rewards/chosen": 2.068169116973877,
    "eval_rewards/margins": 0.0749686062335968,
    "eval_rewards/rejected": 1.993200421333313,
    "eval_runtime": 41.6193,
    "eval_samples_per_second": 24.027,
    "eval_steps_per_second": 3.003
}