eval_results.json

{
    "epoch": 0.9880609304240429,
    "eval_logits/chosen": 9.119073867797852,
    "eval_logits/rejected": 9.919014930725098,
    "eval_logps/chosen": -109.64173889160156,
    "eval_logps/ref_chosen": -106.16099548339844,
    "eval_logps/ref_rejected": -119.78473663330078,
    "eval_logps/rejected": -123.23009490966797,
    "eval_loss": 0.6934764981269836,
    "eval_rewards/accuracies": 0.5018518567085266,
    "eval_rewards/chosen": -0.034807514399290085,
    "eval_rewards/margins": -0.0003539324679877609,
    "eval_rewards/rejected": -0.034453582018613815,
    "eval_runtime": 161.8044,
    "eval_samples": 1080,
    "eval_samples_per_second": 6.675,
    "eval_steps_per_second": 1.669
}