Edit Less, Achieve More: Dynamic Sparse Neuron Masking for Lifelong Knowledge Editing in LLMs

TL;DR

This paper introduces NMKE, a fine-grained, neuron-level knowledge editing framework for LLMs that uses dynamic sparse masking to improve lifelong editing accuracy and preserve generalization.

Contribution

The paper proposes a novel neuron-specific, entropy-guided dynamic masking approach for precise knowledge editing in LLMs, reducing errors and preserving capabilities.

Findings

NMKE outperforms existing methods in success rates.

It maintains model generalization after multiple edits.

Fewer parameter modifications are needed for effective editing.

Abstract

Lifelong knowledge editing enables continuous, precise updates to outdated knowledge in large language models (LLMs) without computationally expensive full retraining. However, existing methods often accumulate errors throughout the editing process, causing a gradual decline in both editing accuracy and generalization. To tackle this problem, we propose Neuron-Specific Masked Knowledge Editing (NMKE), a novel fine-grained editing framework that combines neuron-level attribution with dynamic sparse masking. Leveraging neuron functional attribution, we identify two key types of knowledge neurons, with knowledge-general neurons activating consistently across prompts and knowledge-specific neurons activating to specific prompts. NMKE further introduces an entropy-guided dynamic sparse mask, locating relevant neurons to the target knowledge. This strategy enables precise neuron-level knowledge editing with fewer parameter modifications. Experimental results from thousands of sequential edits demonstrate that NMKE outperforms existing methods in maintaining high editing success rates and preserving model general capabilities in lifelong editing.