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

Jinzhe Liu, Junshu Sun, Shufan Shen, Chenxue Yang, Shuhui Wang

Keywords: Lifelong model editing, fine-grained editing, knowledge editing, large language models

TL;DR: NMKE is a fine-grained framework that uses neuron-level attribution and sparse masking for precise, efficient lifelong knowledge 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Lifelong knowledge editing enables continuous, precise updates to outdated knowledge in large language models (LLMs) without computationally expensive full retraining.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.

Primary Area: Deep learning (e.g., architectures, generative models, optimization for deep networks, foundation models, LLMs)

Submission Number: 2884