@inproceedings{xiao-etal-2026-representation,
title = "Representation-Guided Parameter-Efficient {LLM} Unlearning",
author = "Xiao, Zeguan and
Mo, Lang and
Chen, Yun and
Yang, Lei and
Zhao, Jiehui and
Yang, Lili and
Chen, Guanhua",
editor = "Liakata, Maria and
Moreira, Viviane P. and
Zhang, Jiajun and
Jurgens, David",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-acl.717/",
pages = "14602--14616",
ISBN = "979-8-89176-395-1",
abstract = "Large Language Models (LLMs) often memorize sensitive or harmful information, necessitating effective machine unlearning techniques. While existing parameter-efficient unlearning methods have shown promise, they still struggle with the forget-retain trade-off. This can be attributed to their reliance on parameter importance metrics to identify parameters that are important exclusively for forget set, which is fundamentally limited by the superposition phenomenon. Due to the polysemantic nature of LLMs parameters, such an importance metric may struggle to disentangle parameters associated with forget and retain sets. In this work, we propose Representation-Guided Low-rank Unlearning (ReGLU), a novel approach that leverages the geometric properties of representation spaces to achieve robust and precise unlearning. First, we develop a representation-guided initialization for LoRA that identifies the optimal subspace for selective forgetting. Second, we introduce a regularization loss that constrains the outputs of the LoRA update to lie in the orthogonal complement of the retain set{'}s representation subspace, thereby minimizing interference with the model{'}s performance on the retain set. We evaluate ReGLU on the TOFU and WMDP benchmarks across multiple models. Our results demonstrate that ReGLU consistently outperforms state-of-the-art baselines, achieving superior unlearning quality while maintaining higher model utility."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="xiao-etal-2026-representation">
<titleInfo>
<title>Representation-Guided Parameter-Efficient LLM Unlearning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zeguan</namePart>
<namePart type="family">Xiao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lang</namePart>
<namePart type="family">Mo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yun</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lei</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiehui</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lili</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Guanhua</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Maria</namePart>
<namePart type="family">Liakata</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Viviane</namePart>
<namePart type="given">P</namePart>
<namePart type="family">Moreira</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiajun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">David</namePart>
<namePart type="family">Jurgens</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">San Diego, California, United States</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-395-1</identifier>
</relatedItem>
<abstract>Large Language Models (LLMs) often memorize sensitive or harmful information, necessitating effective machine unlearning techniques. While existing parameter-efficient unlearning methods have shown promise, they still struggle with the forget-retain trade-off. This can be attributed to their reliance on parameter importance metrics to identify parameters that are important exclusively for forget set, which is fundamentally limited by the superposition phenomenon. Due to the polysemantic nature of LLMs parameters, such an importance metric may struggle to disentangle parameters associated with forget and retain sets. In this work, we propose Representation-Guided Low-rank Unlearning (ReGLU), a novel approach that leverages the geometric properties of representation spaces to achieve robust and precise unlearning. First, we develop a representation-guided initialization for LoRA that identifies the optimal subspace for selective forgetting. Second, we introduce a regularization loss that constrains the outputs of the LoRA update to lie in the orthogonal complement of the retain set’s representation subspace, thereby minimizing interference with the model’s performance on the retain set. We evaluate ReGLU on the TOFU and WMDP benchmarks across multiple models. Our results demonstrate that ReGLU consistently outperforms state-of-the-art baselines, achieving superior unlearning quality while maintaining higher model utility.</abstract>
<identifier type="citekey">xiao-etal-2026-representation</identifier>
<location>
<url>https://aclanthology.org/2026.findings-acl.717/</url>
</location>
<part>
<date>2026-07</date>
<extent unit="page">
<start>14602</start>
<end>14616</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Representation-Guided Parameter-Efficient LLM Unlearning
%A Xiao, Zeguan
%A Mo, Lang
%A Chen, Yun
%A Yang, Lei
%A Zhao, Jiehui
%A Yang, Lili
%A Chen, Guanhua
%Y Liakata, Maria
%Y Moreira, Viviane P.
%Y Zhang, Jiajun
%Y Jurgens, David
%S Findings of the Association for Computational Linguistics: ACL 2026
%D 2026
%8 July
%I Association for Computational Linguistics
%C San Diego, California, United States
%@ 979-8-89176-395-1
%F xiao-etal-2026-representation
%X Large Language Models (LLMs) often memorize sensitive or harmful information, necessitating effective machine unlearning techniques. While existing parameter-efficient unlearning methods have shown promise, they still struggle with the forget-retain trade-off. This can be attributed to their reliance on parameter importance metrics to identify parameters that are important exclusively for forget set, which is fundamentally limited by the superposition phenomenon. Due to the polysemantic nature of LLMs parameters, such an importance metric may struggle to disentangle parameters associated with forget and retain sets. In this work, we propose Representation-Guided Low-rank Unlearning (ReGLU), a novel approach that leverages the geometric properties of representation spaces to achieve robust and precise unlearning. First, we develop a representation-guided initialization for LoRA that identifies the optimal subspace for selective forgetting. Second, we introduce a regularization loss that constrains the outputs of the LoRA update to lie in the orthogonal complement of the retain set’s representation subspace, thereby minimizing interference with the model’s performance on the retain set. We evaluate ReGLU on the TOFU and WMDP benchmarks across multiple models. Our results demonstrate that ReGLU consistently outperforms state-of-the-art baselines, achieving superior unlearning quality while maintaining higher model utility.
%U https://aclanthology.org/2026.findings-acl.717/
%P 14602-14616
Markdown (Informal)
[Representation-Guided Parameter-Efficient LLM Unlearning](https://aclanthology.org/2026.findings-acl.717/) (Xiao et al., Findings 2026)
ACL
- Zeguan Xiao, Lang Mo, Yun Chen, Lei Yang, Jiehui Zhao, Lili Yang, and Guanhua Chen. 2026. Representation-Guided Parameter-Efficient LLM Unlearning. In Findings of the Association for Computational Linguistics: ACL 2026, pages 14602–14616, San Diego, California, United States. Association for Computational Linguistics.