Safety at One Shot: Patching Fine-Tuned LLMs with A Single Instance

📝 Original Info

• Title: Safety at One Shot: Patching Fine-Tuned LLMs with A Single Instance
• ArXiv ID: 2601.01887
• Date: 2026-01-05
• Authors: Jiawen Zhang, Lipeng He, Kejia Chen, Jian Lou, Jian Liu, Xiaohu Yang, Ruoxi Jia

📝 Abstract

Fine-tuning safety-aligned large language models (LLMs) can substantially compromise their safety. Previous approaches require many safety samples or calibration sets, which not only incur significant computational overhead during realignment but also lead to noticeable degradation in model utility. Contrary to this belief, we show that safety alignment can be fully recovered with only a single safety example, without sacrificing utility and at minimal cost. Remarkably, this recovery is effective regardless of the number of harmful examples used in finetuning or the size of the underlying model, and convergence is achieved within just a few epochs. Furthermore, we uncover the low-rank structure of the safety gradient, which explains why such efficient correction is possible. We validate our findings across five safety-aligned LLMs and multiple datasets, demonstrating the generality of our approach.

📄 Full Content

...(본문 내용이 길어 생략되었습니다. 사이트에서 전문을 확인해 주세요.)