CoME: An Unlearning-based Approach to Conflict-free Model Editing

TL;DR

CoME introduces an unlearning-based framework for conflict-free model editing that selectively removes outdated knowledge, improving accuracy and reliability of large language models without compromising linguistic features.

Contribution

It presents a novel unlearning approach to mitigate knowledge conflicts in model editing, enhancing the effectiveness of knowledge updates in LLMs.

Findings

CoME improves editing accuracy on GPT-J and LLaMA-3.

Targeted removal of outdated knowledge enhances model reliability.

CoME maintains generative performance while updating knowledge.

Abstract

Large language models (LLMs) often retain outdated or incorrect information from pre-training, which undermines their reliability. While model editing methods have been developed to address such errors without full re-training, they frequently suffer from knowledge conflicts, where outdated information interferes with new knowledge. In this work, we propose Conflict-free Model Editing (CoME), a novel framework that enhances the accuracy of knowledge updates in LLMs by selectively removing outdated knowledge. CoME leverages unlearning to mitigate knowledge interference, allowing new information to be integrated without compromising relevant linguistic features. Through experiments on GPT-J and LLaMA-3 using Counterfact and ZsRE datasets, we demonstrate that CoME improves both editing accuracy and model reliability when applied to existing editing methods. Our results highlight that the targeted removal of outdated knowledge is crucial for enhancing model editing effectiveness and maintaining the model's generative performance.