all_results.json

{
    "epoch": 1.0,
    "eval_dpo_losses": 0.664600670337677,
    "eval_logits/chosen": -2.755434513092041,
    "eval_logits/rejected": -2.7203128337860107,
    "eval_logps/chosen": -267.8385314941406,
    "eval_logps/rejected": -255.50228881835938,
    "eval_loss": 0.6824241876602173,
    "eval_positive_losses": 0.1476137936115265,
    "eval_rewards/accuracies": 0.6815000176429749,
    "eval_rewards/chosen": 0.1661677062511444,
    "eval_rewards/margins": 0.06265779584646225,
    "eval_rewards/margins_max": 0.27175262570381165,
    "eval_rewards/margins_min": -0.11723098903894424,
    "eval_rewards/margins_std": 0.1305234432220459,
    "eval_rewards/rejected": 0.10350991785526276,
    "eval_runtime": 858.7209,
    "eval_samples": 4000,
    "eval_samples_per_second": 4.658,
    "eval_steps_per_second": 0.291,
    "train_loss": 0.6776896565581647,
    "train_runtime": 67917.559,
    "train_samples": 66812,
    "train_samples_per_second": 0.984,
    "train_steps_per_second": 0.061
}