eval_results.json

{
    "epoch": 0.9982631930527722,
    "eval_logits/chosen": -1.4785971641540527,
    "eval_logits/rejected": -1.4859126806259155,
    "eval_logps/chosen": -1.310508370399475,
    "eval_logps/rejected": -1.7847204208374023,
    "eval_loss": 1.3999961614608765,
    "eval_rewards/accuracies": 0.8109756112098694,
    "eval_rewards/chosen": -13.105083465576172,
    "eval_rewards/margins": 4.742120742797852,
    "eval_rewards/rejected": -17.847204208374023,
    "eval_runtime": 87.7263,
    "eval_samples": 1961,
    "eval_samples_per_second": 22.354,
    "eval_steps_per_second": 1.402
}