Reasons and Solutions for the Decline in Model Performance after Editing

TL;DR

This paper investigates why model performance declines after editing large language models, analyzing data and model factors, and proposes the D4S method to mitigate performance degradation by controlling the L1-norm of editing layers.

Contribution

It identifies key data and model factors affecting editing performance and introduces the D4S method to reduce performance decline during knowledge editing.

Findings

Diversity of editing targets and sequence length impact performance.

L1-norm of editing layer correlates with editing accuracy.

D4S method reduces model damage and improves multiple editing effectiveness.

Abstract

Knowledge editing technology has received widespread attention for low-cost updates of incorrect or outdated knowledge in large-scale language models. However, recent research has found that edited models often exhibit varying degrees of performance degradation. The reasons behind this phenomenon and potential solutions have not yet been provided. In order to investigate the reasons for the performance decline of the edited model and optimize the editing method, this work explores the underlying reasons from both data and model perspectives. Specifically, 1) from a data perspective, to clarify the impact of data on the performance of editing models, this paper first constructs a Multi-Question Dataset (MQD) to evaluate the impact of different types of editing data on model performance. The performance of the editing model is mainly affected by the diversity of editing targets and sequence length, as determined through experiments. 2) From a model perspective, this article explores the factors that affect the performance of editing models. The results indicate a strong correlation between the L1-norm of the editing model layer and the editing accuracy, and clarify that this is an important factor leading to the bottleneck of editing performance. Finally, in order to improve the performance of the editing model, this paper further proposes a Dump for Sequence (D4S) method, which successfully overcomes the previous editing bottleneck by reducing the L1-norm of the editing layer, allowing users to perform multiple effective edits and minimizing model damage. Our code is available at https://github.com/nlpkeg/D4S.