all_results.json

{
    "epoch": 1.0,
    "eval_logits/chosen": -2.0430071353912354,
    "eval_logits/rejected": -2.006614923477173,
    "eval_logps/chosen": -150.90753173828125,
    "eval_logps/rejected": -158.87191772460938,
    "eval_loss": 0.2096947431564331,
    "eval_rewards/accuracies": 0.753000020980835,
    "eval_rewards/chosen": -3.184338331222534,
    "eval_rewards/margins": 2.138751268386841,
    "eval_rewards/rejected": -5.323090076446533,
    "eval_runtime": 58.9022,
    "eval_samples_per_second": 16.977,
    "eval_steps_per_second": 2.122
}