all_results.json

{
    "epoch": 1.971563981042654,
    "eval_logits/chosen": 100.848876953125,
    "eval_logits/rejected": 95.52344512939453,
    "eval_logps/chosen": -350.6426086425781,
    "eval_logps/rejected": -366.2401123046875,
    "eval_loss": 0.6211811304092407,
    "eval_rewards/accuracies": 0.7083333134651184,
    "eval_rewards/chosen": 0.651279628276825,
    "eval_rewards/margins": 0.8597357869148254,
    "eval_rewards/rejected": -0.2084561139345169,
    "eval_runtime": 119.3843,
    "eval_samples": 750,
    "eval_samples_per_second": 6.282,
    "eval_steps_per_second": 0.201,
    "total_flos": 0.0,
    "train_loss": 528876699520.6585,
    "train_runtime": 2168.869,
    "train_samples": 6750,
    "train_samples_per_second": 6.224,
    "train_steps_per_second": 0.048
}