eval_results.json

{
    "epoch": 0.9993060374739764,
    "eval_logits/chosen": -0.9284578561782837,
    "eval_logits/rejected": -0.923488974571228,
    "eval_logps/chosen": -483.4286193847656,
    "eval_logps/rejected": -527.1544189453125,
    "eval_loss": 0.6072572469711304,
    "eval_rewards/accuracies": 0.7875939607620239,
    "eval_rewards/chosen": -0.22044964134693146,
    "eval_rewards/margins": 0.19842886924743652,
    "eval_rewards/rejected": -0.4188785254955292,
    "eval_runtime": 379.0532,
    "eval_samples": 8491,
    "eval_samples_per_second": 22.401,
    "eval_steps_per_second": 0.351
}