Is Bigger Edit Batch Size Always Better? -- An Empirical Study on Model Editing with Llama-3

TL;DR

This paper empirically investigates the impact of edit batch size on large language model editing, revealing that larger batch sizes can impair performance more than smaller, sequential edits, and advocates for combined editing strategies.

Contribution

It provides a comprehensive analysis of batch versus sequential editing strategies on Llama-3, highlighting the limitations of increasing batch sizes and proposing a hybrid approach for better model editing.

Findings

Larger edit batch sizes may degrade model performance.

Sequential editing can be more effective than large batch editing.

Hybrid editing strategies outperform pure batch or sequential methods.

Abstract

This study presents a targeted model editing analysis focused on the latest large language model, Llama-3. We explore the efficacy of popular model editing techniques - ROME, MEMIT, and EMMET, which are designed for precise layer interventions. We identify the most effective layers for targeted edits through an evaluation that encompasses up to 4096 edits across three distinct strategies: sequential editing, batch editing, and a hybrid approach we call as sequential-batch editing. Our findings indicate that increasing edit batch-sizes may degrade model performance more significantly than using smaller edit batches sequentially for equal number of edits. With this, we argue that sequential model editing is an important component for scaling model editing methods and future research should focus on methods that combine both batched and sequential editing. This observation suggests a potential limitation in current model editing methods which push towards bigger edit batch sizes, and we hope it paves way for future investigations into optimizing batch sizes and model editing performance.