TY - GEN
T1 - Navigating the Dual Facets
T2 - 62nd Annual Meeting of the Association for Computational Linguistics, ACL 2024
AU - Lin, Zihao
AU - Beigi, Mohammad
AU - Li, Hongxuan
AU - Zhou, Yufan
AU - Zhang, Yuxiang
AU - Wang, Qifan
AU - Yin, Wenpeng
AU - Huang, Lifu
N1 - Publisher Copyright:
© 2024 Association for Computational Linguistics.
PY - 2024
Y1 - 2024
N2 - Memory Editing (ME) has emerged as an efficient method to modify erroneous facts or inject new facts into Large Language Models (LLMs). Two mainstream ME methods exist: parameter-modifying ME and parameter-preserving ME (integrating extra modules while preserving original parameters). Regrettably, previous studies on ME evaluation have two critical limitations: (i) evaluating LLMs with single edit only, neglecting the need for continuous editing, and (ii) evaluations focusing solely on basic factual triples, overlooking broader LLM capabilities like logical reasoning and reading understanding. This study addresses these limitations with contributions threefold: (i) We explore how ME affects a wide range of fundamental capabilities of LLMs under sequential editing. Experimental results reveal an intriguing phenomenon: Most parameter-modifying ME consistently degrade performance across all tasks after a few sequential edits. In contrast, parameter-preserving ME effectively maintains LLMs' fundamental capabilities but struggles to accurately recall edited knowledge presented in a different format. (ii) We extend our evaluation to different editing settings, such as layers to edit, model size, instruction tuning, etc. Experimental findings indicate several strategies that can potentially mitigate the adverse effects of ME. (iii) We further explain why parameter-modifying ME damages LLMs from three dimensions: parameter changes after editing, language modeling capability, and the in-context learning capability. Our in-depth study advocates more careful use of ME in real-world scenarios.
AB - Memory Editing (ME) has emerged as an efficient method to modify erroneous facts or inject new facts into Large Language Models (LLMs). Two mainstream ME methods exist: parameter-modifying ME and parameter-preserving ME (integrating extra modules while preserving original parameters). Regrettably, previous studies on ME evaluation have two critical limitations: (i) evaluating LLMs with single edit only, neglecting the need for continuous editing, and (ii) evaluations focusing solely on basic factual triples, overlooking broader LLM capabilities like logical reasoning and reading understanding. This study addresses these limitations with contributions threefold: (i) We explore how ME affects a wide range of fundamental capabilities of LLMs under sequential editing. Experimental results reveal an intriguing phenomenon: Most parameter-modifying ME consistently degrade performance across all tasks after a few sequential edits. In contrast, parameter-preserving ME effectively maintains LLMs' fundamental capabilities but struggles to accurately recall edited knowledge presented in a different format. (ii) We extend our evaluation to different editing settings, such as layers to edit, model size, instruction tuning, etc. Experimental findings indicate several strategies that can potentially mitigate the adverse effects of ME. (iii) We further explain why parameter-modifying ME damages LLMs from three dimensions: parameter changes after editing, language modeling capability, and the in-context learning capability. Our in-depth study advocates more careful use of ME in real-world scenarios.
UR - http://www.scopus.com/inward/record.url?scp=85204494332&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85204494332&partnerID=8YFLogxK
U2 - 10.18653/v1/2024.acl-long.742
DO - 10.18653/v1/2024.acl-long.742
M3 - Conference contribution
AN - SCOPUS:85204494332
T3 - Proceedings of the Annual Meeting of the Association for Computational Linguistics
SP - 13755
EP - 13772
BT - Long Papers
A2 - Ku, Lun-Wei
A2 - Martins, Andre F. T.
A2 - Srikumar, Vivek
PB - Association for Computational Linguistics (ACL)
Y2 - 11 August 2024 through 16 August 2024
ER -