eval_results.json

{
    "epoch": 0.9995419147961521,
    "eval_logits/chosen": -0.7738187313079834,
    "eval_logits/rejected": -0.5984290242195129,
    "eval_logps/chosen": -129.0717315673828,
    "eval_logps/rejected": -278.73272705078125,
    "eval_loss": 0.33219456672668457,
    "eval_rewards/accuracies": 0.9017341136932373,
    "eval_rewards/chosen": 0.552308976650238,
    "eval_rewards/margins": 1.252788782119751,
    "eval_rewards/rejected": -0.7004798650741577,
    "eval_runtime": 251.2644,
    "eval_samples": 2763,
    "eval_samples_per_second": 10.996,
    "eval_steps_per_second": 1.377
}