让大模型先反思再拒绝，减少误判无辜请求的错误拒绝。
Think Before Refusal : Triggering Safety Reflection in LLMs to Mitigate False Refusal Behavior
- 响应前增加安全反思环节，避免误拒正常请求。
- 15个模型实验表明，误拒绝率显著下降，安全性能不变。
- 适合需要高准确拒绝能力的对话系统开发者。
近期大语言模型（LLMs）通过微调和人类对齐实现了无害性，主要方法是训练模型拒绝有害请求，如‘教我如何烧掉邻居的房子’。然而，该方法可能引发误拒绝问题，即模型错误拒绝无害请求，例如‘告诉我如何终止一个Python进程’。本文发现，在生成回应前引入安全反思提示，可有效缓解误拒绝行为。基于此，我们提出‘思考后拒绝’（TBR）框架，并进行包含安全反思的安全意识指令微调。在15个预训练模型的消融实验中，使用安全反思微调的模型显著降低误拒绝率，同时保持原有安全性和整体性能。
原文摘要 · Abstract (English)
Recent advancements in large language models (LLMs) have demonstrated that fine-tuning and human alignment can render LLMs harmless. In practice, such "harmlessness" behavior is mainly achieved by training models to reject harmful requests, such as "Explain how to burn down my neighbor's house", where the model appropriately declines to respond. However, this approach can inadvertently result in false refusal, where models reject benign queries as well, such as "Tell me how to kill a Python process". In this work, we demonstrate that prompting safety reflection before generating a response can mitigate false refusal behavior. Building on this finding, we introduce the Think-Before-Refusal (TBR) schema and conduct safety-aware instruction fine-tuning incorporating safety reflection. In an ablation study across 15 pre-trained models, we show that models fine-tuned with safety reflection significantly reduce false refusal behavior while maintaining safety and overall performance compared to those fine-tuned without safety reflection.
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。