EvoEdit: Evolving Null-space Alignment for Robust and Efficient Knowledge Editing

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Lyu, Sicheng, Gu, Yu, Wang, Xinyu, Huang, Jerry, Luan, Sitao, Cui, Yufei, Chang, Xiao-Wen, Lu, Peng
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911568131784704
author Lyu, Sicheng
Gu, Yu
Wang, Xinyu
Huang, Jerry
Luan, Sitao
Cui, Yufei
Chang, Xiao-Wen
Lu, Peng
author_facet Lyu, Sicheng
Gu, Yu
Wang, Xinyu
Huang, Jerry
Luan, Sitao
Cui, Yufei
Chang, Xiao-Wen
Lu, Peng
contents Large language models (LLMs) require continual updates to rectify outdated or erroneous knowledge. Model editing has emerged as a compelling paradigm for introducing targeted modifications without the computational burden of full retraining. Existing approaches are mainly based on a locate-then-edit framework. However, in sequential editing contexts, where multiple updates are applied over time, they exhibit significant limitations and suffer from catastrophic interference, i.e., new edits compromise previously integrated updates and degrade preserved knowledge. To address these challenges, we introduce EvoEdit, a novel editing strategy that mitigates catastrophic interference through sequential null-space alignment, enabling stable and efficient model editing. By performing sequential null-space alignment for each incoming edit, EvoEdit preserves both original and previously modified knowledge representations and maintains output invariance on preserved knowledge even across long edit sequences, effectively mitigating interference. Evaluations on real-world sequential knowledge-editing benchmarks show that EvoEdit achieves better or comparable performance than prior state-of-the-art locate-then-edit techniques, with up to 3.53 times speedup. Overall, these results underscore the necessity of developing more principled approaches for designing LLMs in dynamically evolving information settings, while providing a simple yet effective solution with strong theoretical guarantees.
format Preprint
id arxiv_https___arxiv_org_abs_2510_13851
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle EvoEdit: Evolving Null-space Alignment for Robust and Efficient Knowledge Editing
Lyu, Sicheng
Gu, Yu
Wang, Xinyu
Huang, Jerry
Luan, Sitao
Cui, Yufei
Chang, Xiao-Wen
Lu, Peng
Computation and Language
Machine Learning
Large language models (LLMs) require continual updates to rectify outdated or erroneous knowledge. Model editing has emerged as a compelling paradigm for introducing targeted modifications without the computational burden of full retraining. Existing approaches are mainly based on a locate-then-edit framework. However, in sequential editing contexts, where multiple updates are applied over time, they exhibit significant limitations and suffer from catastrophic interference, i.e., new edits compromise previously integrated updates and degrade preserved knowledge. To address these challenges, we introduce EvoEdit, a novel editing strategy that mitigates catastrophic interference through sequential null-space alignment, enabling stable and efficient model editing. By performing sequential null-space alignment for each incoming edit, EvoEdit preserves both original and previously modified knowledge representations and maintains output invariance on preserved knowledge even across long edit sequences, effectively mitigating interference. Evaluations on real-world sequential knowledge-editing benchmarks show that EvoEdit achieves better or comparable performance than prior state-of-the-art locate-then-edit techniques, with up to 3.53 times speedup. Overall, these results underscore the necessity of developing more principled approaches for designing LLMs in dynamically evolving information settings, while providing a simple yet effective solution with strong theoretical guarantees.
title EvoEdit: Evolving Null-space Alignment for Robust and Efficient Knowledge Editing
topic Computation and Language
Machine Learning
url https://arxiv.org/abs/2510.13851