<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="shi-etal-2026-modular">
    <titleInfo>
        <title>Modular Representation Steering for Multi-Attribute LLM Alignment</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Chaojian</namePart>
        <namePart type="family">Shi</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Renhong</namePart>
        <namePart type="family">Huang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Xulan</namePart>
        <namePart type="family">Qiu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Jiawei</namePart>
        <namePart type="family">Zhou</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yunsong</namePart>
        <namePart type="family">Zhou</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2026-10</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Hung-Yi</namePart>
            <namePart type="family">Lee</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Sunipa</namePart>
            <namePart type="family">Dev</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jessy</namePart>
            <namePart type="family">Li</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Desmond</namePart>
            <namePart type="family">Elliott</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Budapest, Hungary</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-459-0</identifier>
    </relatedItem>
    <abstract>Aligning large language models (LLMs) with multiple behavioral objectives from a certain corpus remains challenging, as joint optimization of attributes often leads to interference and trade-offs. While recent representation steering methods enable lightweight control over individual behaviors through hidden-state interventions, their composability for multi-attribute alignment remains underexplored. We present POLAR, Poly-attribute Alignment via Modular Representation Steering, a modular framework that composes independently trained alignment interventions within a frozen LLM. To mitigate conflicts between steering modules, we introduce calibrated and compatibility-aware intervention composition in hidden-state space. Experiments across heterogeneous alignment attributes show that POLAR achieves the best overall trade-off, with a 37.4% improvement in average rank over the strongest multi-attribute baseline while preserving response quality and controllability. The code and data will be publicly available.</abstract>
    <identifier type="citekey">shi-etal-2026-modular</identifier>
    <location>
        <url>https://aclanthology.org/2026.emnlp-main.1469/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>33619</start>
            <end>33641</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Modular Representation Steering for Multi-Attribute LLM Alignment
%A Shi, Chaojian
%A Huang, Renhong
%A Qiu, Xulan
%A Zhou, Jiawei
%A Zhou, Yunsong
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F shi-etal-2026-modular
%X Aligning large language models (LLMs) with multiple behavioral objectives from a certain corpus remains challenging, as joint optimization of attributes often leads to interference and trade-offs. While recent representation steering methods enable lightweight control over individual behaviors through hidden-state interventions, their composability for multi-attribute alignment remains underexplored. We present POLAR, Poly-attribute Alignment via Modular Representation Steering, a modular framework that composes independently trained alignment interventions within a frozen LLM. To mitigate conflicts between steering modules, we introduce calibrated and compatibility-aware intervention composition in hidden-state space. Experiments across heterogeneous alignment attributes show that POLAR achieves the best overall trade-off, with a 37.4% improvement in average rank over the strongest multi-attribute baseline while preserving response quality and controllability. The code and data will be publicly available.
%U https://aclanthology.org/2026.emnlp-main.1469/
%P 33619-33641

Markdown (Informal)

[Modular Representation Steering for Multi-Attribute LLM Alignment](https://aclanthology.org/2026.emnlp-main.1469/) (Shi et al., EMNLP 2026)

• Modular Representation Steering for Multi-Attribute LLM Alignment (Shi et al., EMNLP 2026)

ACL

• Chaojian Shi, Renhong Huang, Xulan Qiu, Jiawei Zhou, and Yunsong Zhou. 2026. Modular Representation Steering for Multi-Attribute LLM Alignment. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 33619–33641, Budapest, Hungary. Association for Computational Linguistics.