Long-form evaluation of model editing

Domenic Rosati, Robie Gonzales, Jinkun Chen, Xuemin Yu, Yahya Kayani, Frank Rudzicz, Hassan Sajjad


Abstract
Evaluations of model editing, a technique for changing the factual knowledge held by Large Language Models (LLMs), currently only use the ‘next few token’ completions after a prompt. As a result, the impact of these methods on longer natural language generation is largely unknown. We introduce long-form evaluation of model editing (LEME) a novel evaluation protocol that measures the efficacy and impact of model editing in long-form generative settings. Our protocol consists of a machine-rated survey and a classifier which correlates well with human ratings. Importantly, we find that our protocol has very little relationship with previous short-form metrics (despite being designed to extend efficacy, generalization, locality, and portability into a long-form setting), indicating that our method introduces a novel set of dimensions for understanding model editing methods. Using this protocol, we benchmark a number of model editing techniques and present several findings including that, while some methods (ROME and MEMIT) perform well in making consistent edits within a limited scope, they suffer much more from factual drift than other methods. Finally, we present a qualitative analysis that illustrates common failure modes in long-form generative settings including internal consistency, lexical cohesion, and locality issues.
Anthology ID:
2024.naacl-long.208
Volume:
Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers)
Month:
June
Year:
2024
Address:
Mexico City, Mexico
Editors:
Kevin Duh, Helena Gomez, Steven Bethard
Venue:
NAACL
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
3749–3780
Language:
URL:
https://aclanthology.org/2024.naacl-long.208
DOI:
Bibkey:
Cite (ACL):
Domenic Rosati, Robie Gonzales, Jinkun Chen, Xuemin Yu, Yahya Kayani, Frank Rudzicz, and Hassan Sajjad. 2024. Long-form evaluation of model editing. In Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers), pages 3749–3780, Mexico City, Mexico. Association for Computational Linguistics.
Cite (Informal):
Long-form evaluation of model editing (Rosati et al., NAACL 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.naacl-long.208.pdf
Copyright:
 2024.naacl-long.208.copyright.pdf