{ "epoch": 1.0, "eval_logits/chosen": -3.1484375, "eval_logits/rejected": -3.20703125, "eval_logps/chosen": -885.0, "eval_logps/rejected": -913.0, "eval_loss": 0.5779687762260437, "eval_rewards/accuracies": 0.5625, "eval_rewards/chosen": -0.176513671875, "eval_rewards/margins": 0.3870849609375, "eval_rewards/rejected": -0.5640869140625, "eval_runtime": 4.7753, "eval_samples_per_second": 20.941, "eval_steps_per_second": 0.838 }