改变提示中角色标签，能显著提升大模型自纠错能力。
The Self-Correction Illusion: Role Relabeling Gates Explicit Error Flagging in Large Language Models
- 通过改变错误信息的角色标签（如从<thought>改为用户消息），不修改内容即可提升纠错率。
- 纠错率最高提升93个百分点，10组实验结果均显著改善。
- 不同任务适合不同角色标签，提示设计需考虑领域差异。
近期研究表明，大型语言模型代理虽能纠正外部来源的错误，却难以识别自身推理过程中的错误。我们探究这是否源于能力缺陷，还是角色标签设计的产物。为此，提出一种无需训练的干预方法——源条件角色重标签：保持错误陈述字节完全一致，仅改变其在对话模板中的角色标签（如<thought>、用户消息、工具响应或系统<记忆>）。在涵盖闭源API与700亿参数至小型模型的12种模型-领域组合上测试，将<thought>重标为外部角色后，显式纠错率提升23至93个百分点，10组实验结果显著。表明模型无法发现自我生成错误，很大程度上是角色标签设计所致，而非认知缺陷。最有效标签具有领域依赖性：数学任务中<记忆>最优，逻辑推理中用户消息更优。这提示角色标签应作为指令微调中的关键实验变量，以更直接地弥合自纠错差距。
原文摘要 · Abstract (English)
Recent works show that LLM agents struggle to correct errors in their own reasoning traces, despite their ability to correct errors from external sources. We ask whether this reflects a capability deficit or an artifact of the role labeling. To test this, we design a training-free intervention, source-conditioned role relabeling, that keeps the erroneous claim byte-identical and varies only its message role. The claim is presented inside the agent's "<thought>", a user message, a tool response, or a system "<memory>" block. We test 12 model-domain combinations spanning closed-weight APIs and open-weight models from 70B-class down to smaller families. Relabeling "<thought>" to an external role increases the explicit-correction rate by 23 to 93 percentage points, significant in 10 of 12 experimental settings. This suggests that these models' failure to detect a self-generated error is largely an artifact of how the claim is role-labeled in the chat template, rather than a pure cognitive deficit. The most effective role label is domain-dependent: "<memory>" dominates in most math experiments, while a user message dominates in logical deduction. Recognizing role-label handling as a key experimental variable in instruction tuning presents a more direct path to closing the self-correction gap.d
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。