Equitable Alignment of LLMs Across Languages with Regularized Direct Preference Optimization

Thibaud Leteno; Antoine Gourru; Charlotte Laclau; Christophe Gravier

Equitable Alignment of LLMs Across Languages with Regularized Direct Preference Optimization

Thibaud Leteno, Antoine Gourru, Charlotte Laclau, Christophe Gravier

Correct Metadata for

Use this form to create a GitHub issue with structured data describing the correction. You will need a GitHub account. Once you create that issue, the correction will be reviewed by a staff member.

⚠️ Mobile Users: Submitting this form to create a new issue will only work with github.com, not the GitHub Mobile app.

Important: The Anthology treat PDFs as authoritative. Please use this form only to correct data that is out of line with the PDF. See our corrections guidelines if you need to change the PDF.

Title Adjust the title. Retain tags such as <fixed-case>.

Authors Adjust author names and order to match the PDF.

Abstract Correct abstract if needed. Retain XML formatting tags such as <tex-math>. You may use ... for bold, ... for italic, ... for underline, <sc>...</sc> for small-caps, <tt>...<tt> for typewriter text, <url>...</url> for URLs, <a href=...> for hyperlinks, and <par/> for paragraph breaks.

Verification against PDF Ensure that the new title/authors match the snapshot below. (If there is no snapshot or it is too small, consult the PDF.)

Authors concatenated from the text boxes above:

ALL author names match the snapshot above—including middle initials, hyphens, and accents.

Abstract

Aligning large language models with human values is a critical challenge for ensuring public safety. This alignment is commonly achieved through methods such as Direct Preference Optimization and RLAIF/RLHF. Despite extensive research, prior studies show that models often fail to maintain consistent alignment performance across languages, resulting in unequal treatment of users. To address this issue, we introduce a regularization term in the Direct Preference Optimization fine-tuning process that penalizes discrepancies in model behavior between languages using a bilingual corpus. Our results demonstrate that this approach effectively reduces the alignment gap across languages following fine-tuning.

Anthology ID:: 2026.findings-emnlp.121
Volume:: Findings of the Association for Computational Linguistics: EMNLP 2026
Month:: October
Year:: 2026
Address:: Budapest, Hungary
Editors:: Hung-Yi Lee, Sunipa Dev, Jessy Li, Desmond Elliott
Venue:: Findings
SIG:: SIGDAT
Publisher:: Association for Computational Linguistics
Note:
Pages:: 2486–2495
Language:
URL:: https://aclanthology.org/2026.findings-emnlp.121/
DOI:
Bibkey:
Cite (ACL):: Thibaud Leteno, Antoine Gourru, Charlotte Laclau, and Christophe Gravier. 2026. Equitable Alignment of LLMs Across Languages with Regularized Direct Preference Optimization. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 2486–2495, Budapest, Hungary. Association for Computational Linguistics.
Cite (Informal):: Equitable Alignment of LLMs Across Languages with Regularized Direct Preference Optimization (Leteno et al., Findings 2026)
Copy Citation:
PDF:: https://aclanthology.org/2026.findings-emnlp.121.pdf

PDF Cite Search Fix data

Export citation

BibTeX
MODS XML
Endnote
Preformatted

@inproceedings{leteno-etal-2026-equitable,
    title = "Equitable Alignment of {LLM}s Across Languages with Regularized Direct Preference Optimization",
    author = "Leteno, Thibaud  and
      Gourru, Antoine  and
      Laclau, Charlotte  and
      Gravier, Christophe",
    editor = "Lee, Hung-Yi  and
      Dev, Sunipa  and
      Li, Jessy  and
      Elliott, Desmond",
    booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
    month = oct,
    year = "2026",
    address = "Budapest, Hungary",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2026.findings-emnlp.121/",
    pages = "2486--2495",
    ISBN = "979-8-89176-462-0",
    abstract = "Aligning large language models with human values is a critical challenge for ensuring public safety. This alignment is commonly achieved through methods such as Direct Preference Optimization and RLAIF/RLHF. Despite extensive research, prior studies show that models often fail to maintain consistent alignment performance across languages, resulting in unequal treatment of users. To address this issue, we introduce a regularization term in the Direct Preference Optimization fine-tuning process that penalizes discrepancies in model behavior between languages using a bilingual corpus. Our results demonstrate that this approach effectively reduces the alignment gap across languages following fine-tuning."
}

Download as File

<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="leteno-etal-2026-equitable">
    <titleInfo>
        <title>Equitable Alignment of LLMs Across Languages with Regularized Direct Preference Optimization</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Thibaud</namePart>
        <namePart type="family">Leteno</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Antoine</namePart>
        <namePart type="family">Gourru</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Charlotte</namePart>
        <namePart type="family">Laclau</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Christophe</namePart>
        <namePart type="family">Gravier</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
    </relatedItem>
    <abstract>Aligning large language models with human values is a critical challenge for ensuring public safety. This alignment is commonly achieved through methods such as Direct Preference Optimization and RLAIF/RLHF. Despite extensive research, prior studies show that models often fail to maintain consistent alignment performance across languages, resulting in unequal treatment of users. To address this issue, we introduce a regularization term in the Direct Preference Optimization fine-tuning process that penalizes discrepancies in model behavior between languages using a bilingual corpus. Our results demonstrate that this approach effectively reduces the alignment gap across languages following fine-tuning.</abstract>
    <identifier type="citekey">leteno-etal-2026-equitable</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-emnlp.121/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>2486</start>
            <end>2495</end>
        </extent>
    </part>
</mods>
</modsCollection>

Download as File

%0 Conference Proceedings
%T Equitable Alignment of LLMs Across Languages with Regularized Direct Preference Optimization
%A Leteno, Thibaud
%A Gourru, Antoine
%A Laclau, Charlotte
%A Gravier, Christophe
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F leteno-etal-2026-equitable
%X Aligning large language models with human values is a critical challenge for ensuring public safety. This alignment is commonly achieved through methods such as Direct Preference Optimization and RLAIF/RLHF. Despite extensive research, prior studies show that models often fail to maintain consistent alignment performance across languages, resulting in unequal treatment of users. To address this issue, we introduce a regularization term in the Direct Preference Optimization fine-tuning process that penalizes discrepancies in model behavior between languages using a bilingual corpus. Our results demonstrate that this approach effectively reduces the alignment gap across languages following fine-tuning.
%U https://aclanthology.org/2026.findings-emnlp.121/
%P 2486-2495

Download as File

Markdown (Informal)

[Equitable Alignment of LLMs Across Languages with Regularized Direct Preference Optimization](https://aclanthology.org/2026.findings-emnlp.121/) (Leteno et al., Findings 2026)

Equitable Alignment of LLMs Across Languages with Regularized Direct Preference Optimization (Leteno et al., Findings 2026)

ACL

Thibaud Leteno, Antoine Gourru, Charlotte Laclau, and Christophe Gravier. 2026. Equitable Alignment of LLMs Across Languages with Regularized Direct Preference Optimization. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 2486–2495, Budapest, Hungary. Association for Computational Linguistics.