eval_results.json

{
    "epoch": 0.9987943737441393,
    "eval_logits/chosen": -3.4579527378082275,
    "eval_logits/rejected": -3.4530882835388184,
    "eval_logps/chosen": -0.5745237469673157,
    "eval_logps/rejected": -0.6532463431358337,
    "eval_loss": 1.509045958518982,
    "eval_rewards/accuracies": 0.5890957713127136,
    "eval_rewards/chosen": -0.5745237469673157,
    "eval_rewards/margins": 0.07872254401445389,
    "eval_rewards/rejected": -0.6532463431358337,
    "eval_runtime": 434.8486,
    "eval_samples": 2994,
    "eval_samples_per_second": 6.885,
    "eval_steps_per_second": 0.432
}