Latent Knowledge Scalpel: Precise and Massive Knowledge Editing for Large Language Models

Xin Liu, Qiyang Song, Shaowen Xu, Kerou Zhou, Wenbo Jiang, Xiaoqi Jia, Weijuan Zhang, Heqing Huang, Yakai Li

Keywords: model editing, knowledge editing, representation, LLM

TL;DR: An LLM editor which can perform large-scale and simultaneous knowledge editing while preserving the general abilities of the edited LLMs.

Abstract: Large Language Models (LLMs) often retain inaccurate or outdated information from pre-training, leading to incorrect predictions or biased outputs during inference. While existing model editing methods can address this challenge, they struggle with editing large amounts of factual information simultaneously and may compromise the general capabilities of the models. In this paper, our empirical study demonstrates that it is feasible to edit the internal representations of LLMs and replace the entities in a manner similar to editing natural language inputs. Based on this insight, we introduce the Latent Knowledge Scalpel (LKS), an LLM editor that manipulates the latent knowledge of specific entities via a hypernetwork to enable precise and large-scale editing. Experiments conducted on Llama-2 and Mistral show that even with the number of simultaneous edits reaching 10,000, LKS effectively performs knowledge editing while preserving the general abilities of the edited LLMs.

Archival Status: Non-archival (not included in proceedings)

Submission Number: 24