Addressing the Reasoning Gap: Mechanistic Circuit-Based Knowledge Editing in Large Language Models

The paper introduces MCircKE, a mechanistic circuit-based knowledge editing framework for large language models. MCircKE identifies the causal circuits involved in a specific reasoning task and surgically updates parameters only within those circuits, thereby addressing the reasoning gap where edited facts are not used in multi-step reasoning. Experiments on the MQuAKE-series benchmarks show that this approach improves multi-hop reasoning performance after knowledge editing.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computation and Language.

• llms
• benchmarks