{ "epoch": 1.0, "eval_epsilon_dpo/beta": 0.0005516819655895233, "eval_epsilon_dpo/beta_margin_grad_mean": -0.43731194734573364, "eval_epsilon_dpo/beta_margin_grad_std": 0.13020560145378113, "eval_epsilon_dpo/beta_margin_mean": 0.27542224526405334, "eval_epsilon_dpo/beta_margin_std": 0.5839142203330994, "eval_epsilon_dpo/loss_margin_mean": 506.5054016113281, "eval_kl/n_epsilon_steps": 0.3274828791618347, "eval_kl/p_epsilon_steps": 0.6725171208381653, "eval_logits/chosen": -4.05236291885376, "eval_logits/rejected": -4.968125343322754, "eval_logps/chosen": -881.2797241210938, "eval_logps/ref_chosen": -68.29110717773438, "eval_logps/ref_rejected": -92.08038330078125, "eval_logps/rejected": -1411.574462890625, "eval_loss": 0.6066829562187195, "eval_rewards/accuracies": 0.6922088861465454, "eval_rewards/chosen": -0.4506734609603882, "eval_rewards/margins": 0.27542227506637573, "eval_rewards/rejected": -0.7260956764221191, "eval_runtime": 38.4171, "eval_samples": 2339, "eval_samples_per_second": 60.884, "eval_steps_per_second": 1.926 }