eval_results.json

{
    "epoch": 0.9885057471264368,
    "eval_logits/chosen": -0.779742419719696,
    "eval_logits/rejected": -0.755063533782959,
    "eval_logps/chosen": -183.93116760253906,
    "eval_logps/rejected": -185.45208740234375,
    "eval_loss": 2.0544536113739014,
    "eval_nll_loss": 0.5179869532585144,
    "eval_rewards/accuracies": 0.5260869860649109,
    "eval_rewards/chosen": -18.393117904663086,
    "eval_rewards/margins": 0.15209028124809265,
    "eval_rewards/rejected": -18.54520606994629,
    "eval_runtime": 73.2539,
    "eval_samples": 1826,
    "eval_samples_per_second": 24.927,
    "eval_steps_per_second": 1.57
}