Skip to yearly menu bar Skip to main content


Gaining Wisdom from Setbacks: Aligning Large Language Models via Mistake Analysis

Kai Chen · Chunwei Wang · Kuo Yang · Jianhua Han · Lanqing HONG · Fei Mi · Hang Xu · Zhengying Liu · Wenyong Huang · Zhenguo Li · Dit-Yan Yeung · Lifeng Shang

Halle B #118
[ ]
Wed 8 May 1:45 a.m. PDT — 3:45 a.m. PDT


The rapid development of large language models (LLMs) has not only provided numerous opportunities but also presented significant challenges. This becomes particularly evident when LLMs inadvertently generate harmful or toxic content, either unintentionally or because of intentional inducement. Existing alignment methods usually direct LLMs toward the favorable outcomes by utilizing human-annotated, flawless instruction-response pairs. Conversely, this study proposes a novel alignment technique based on mistake analysis, which deliberately exposes LLMs to erroneous content to learn the reasons for mistakes and how to avoid them. In this case, mistakes are repurposed into valuable data for alignment, effectively helping to avoid the production of erroneous responses. Without external models or human annotations, our method leverages a model's intrinsic ability to discern undesirable mistakes and improves the safety of its generated responses. Experimental results reveal that our method outperforms existing alignment approaches in enhancing model safety while maintaining the overall utility.

Chat is not available.