CaKE: Circuit-aware Editing Enables Generalizable Knowledge Learners

원문 초록

Knowledge Editing (KE) enables the modification of outdated or incorrect information in large language models (LLMs). While existing KE methods can update isolated facts, they struggle to generalize these updates to multi-hop reasoning tasks that depend on the modified knowledge. Through an analysis of reasoning circuits -- the neural pathways LLMs use for knowledge-based inference, we observe that current layer-localized KE approaches, such as MEMIT and WISE, which edit only single or a few model layers, struggle to effectively incorporate updated information into these reasoning pathways. To address this limitation, we propose CaKE (Circuit-aware Knowledge Editing), a novel method that enables more effective integration of updated knowledge in LLMs. CaKE leverages strategically curated data, guided by our circuits-based analysis, that enforces the model to utilize the modified knowledge, stimulating the model to develop appropriate reasoning circuits for newly integrated knowledge. Experimental results show that CaKE enables more accurate and consistent use of updated knowledge across related reasoning tasks, leading to an average of 20% improvement in multi-hop reasoning accuracy on MQuAKE dataset compared to existing KE methods. We release the code and data in https://github.com/zjunlp/CaKE.