<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="saeidi-etal-2025-insights">
    <titleInfo>
        <title>Insights into Alignment: Evaluating DPO and its Variants Across Multiple Tasks</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Amir</namePart>
        <namePart type="family">Saeidi</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Shivanshu</namePart>
        <namePart type="family">Verma</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Md</namePart>
        <namePart type="given">Nayem</namePart>
        <namePart type="family">Uddin</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Chitta</namePart>
        <namePart type="family">Baral</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025-07</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop)</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Jin</namePart>
            <namePart type="family">Zhao</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Mingyang</namePart>
            <namePart type="family">Wang</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Zhu</namePart>
            <namePart type="family">Liu</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Vienna, Austria</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-254-1</identifier>
    </relatedItem>
    <abstract>This study evaluates Direct Preference Optimization (DPO) and its variants for aligning Large Language Models (LLMs) with human preferences, testing three configurations: (1) with Supervised Fine-Tuning (SFT), (2) without SFT, and (3) without SFT but using an instruction-tuned model. We further investigate how training set size influences model performance. Our evaluation spans 13 benchmarks—covering dialogue, reasoning, mathematical problem-solving, question answering, truthfulness, MT-Bench, Big Bench, and the Open LLM Leaderboard. We find that: (1) alignment methods often achieve near-optimal performance even with smaller subsets of training data; (2) although they offer limited improvements on complex reasoning tasks, they enhance mathematical problem-solving; and (3) using an instruction-tuned model improves truthfulness. These insights highlight the conditions under which alignment methods excel, as well as their limitations.</abstract>
    <identifier type="citekey">saeidi-etal-2025-insights</identifier>
    <identifier type="doi">10.18653/v1/2025.acl-srw.26</identifier>
    <location>
        <url>https://aclanthology.org/2025.acl-srw.26/</url>
    </location>
    <part>
        <date>2025-07</date>
        <extent unit="page">
            <start>409</start>
            <end>421</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Insights into Alignment: Evaluating DPO and its Variants Across Multiple Tasks
%A Saeidi, Amir
%A Verma, Shivanshu
%A Uddin, Md Nayem
%A Baral, Chitta
%Y Zhao, Jin
%Y Wang, Mingyang
%Y Liu, Zhu
%S Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop)
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-254-1
%F saeidi-etal-2025-insights
%X This study evaluates Direct Preference Optimization (DPO) and its variants for aligning Large Language Models (LLMs) with human preferences, testing three configurations: (1) with Supervised Fine-Tuning (SFT), (2) without SFT, and (3) without SFT but using an instruction-tuned model. We further investigate how training set size influences model performance. Our evaluation spans 13 benchmarks—covering dialogue, reasoning, mathematical problem-solving, question answering, truthfulness, MT-Bench, Big Bench, and the Open LLM Leaderboard. We find that: (1) alignment methods often achieve near-optimal performance even with smaller subsets of training data; (2) although they offer limited improvements on complex reasoning tasks, they enhance mathematical problem-solving; and (3) using an instruction-tuned model improves truthfulness. These insights highlight the conditions under which alignment methods excel, as well as their limitations.
%R 10.18653/v1/2025.acl-srw.26
%U https://aclanthology.org/2025.acl-srw.26/
%U https://doi.org/10.18653/v1/2025.acl-srw.26
%P 409-421

Markdown (Informal)

[Insights into Alignment: Evaluating DPO and its Variants Across Multiple Tasks](https://aclanthology.org/2025.acl-srw.26/) (Saeidi et al., ACL 2025)

• Insights into Alignment: Evaluating DPO and its Variants Across Multiple Tasks (Saeidi et al., ACL 2025)

ACL

• Amir Saeidi, Shivanshu Verma, Md Nayem Uddin, and Chitta Baral. 2025. Insights into Alignment: Evaluating DPO and its Variants Across Multiple Tasks. In Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 4: Student Research Workshop), pages 409–421, Vienna, Austria. Association for Computational Linguistics.