Model Merging for Knowledge Editing

TL;DR

This paper introduces a two-stage model merging framework combining supervised fine-tuning and merging techniques to improve knowledge editing in large language models, especially in sequential editing scenarios.

Contribution

The proposed method effectively enhances knowledge updating in LLMs while preserving their original capabilities without architectural modifications.

Findings

Outperforms existing knowledge editing methods in sequential scenarios

Better preserves original model performance after editing

Does not require architectural changes

Abstract

Large Language Models (LLMs) require continuous updates to maintain accurate and current knowledge as the world evolves. While existing knowledge editing approaches offer various solutions for knowledge updating, they often struggle with sequential editing scenarios and harm the general capabilities of the model, thereby significantly hampering their practical applicability. This paper proposes a two-stage framework combining robust supervised fine-tuning (R-SFT) with model merging for knowledge editing. Our method first fine-tunes the LLM to internalize new knowledge fully, then merges the fine-tuned model with the original foundation model to preserve newly acquired knowledge and general capabilities. Experimental results demonstrate that our approach significantly outperforms existing methods in sequential editing while better preserving the original performance of the model, all without requiring any architectural changes. Code is available at: https://github.com/Applied-Machine-Learning-Lab/MM4KE.