{ "epoch": 3.0, "eval_logits/chosen": 0.1689453125, "eval_logits/rejected": 1.71875, "eval_logps/chosen": -1624.0, "eval_logps/rejected": -1808.0, "eval_loss": 1.7063828706741333, "eval_rewards/accuracies": 0.626953125, "eval_rewards/chosen": -13.0625, "eval_rewards/margins": 2.109375, "eval_rewards/rejected": -15.1875, "eval_runtime": 46.4619, "eval_samples": 2000, "eval_samples_per_second": 43.046, "eval_steps_per_second": 0.689 }