{ "epoch": 3.0, "eval_logits/chosen": -0.1736111342906952, "eval_logits/rejected": -0.31976231932640076, "eval_logps/chosen": -359.9693298339844, "eval_logps/rejected": -288.6960144042969, "eval_loss": 0.6869024038314819, "eval_rewards/accuracies": 0.567460298538208, "eval_rewards/chosen": 0.02891545742750168, "eval_rewards/margins": 0.01524704322218895, "eval_rewards/rejected": 0.013668415136635303, "eval_runtime": 150.91, "eval_samples": 2000, "eval_samples_per_second": 13.253, "eval_steps_per_second": 0.417 }