<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="vajda-etal-2025-improving">
    <titleInfo>
        <title>Improving LLMs for Machine Translation Using Synthetic Preference Data</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Dario</namePart>
        <namePart type="family">Vajda</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Domen</namePart>
        <namePart type="family">Vreš</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Marko</namePart>
        <namePart type="given">Robnik</namePart>
        <namePart type="family">Šikonja</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025-10</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the 2nd LUHME Workshop</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Henrique</namePart>
            <namePart type="given">Lopes</namePart>
            <namePart type="family">Cardoso</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Rui</namePart>
            <namePart type="family">Sousa-Silva</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Maarit</namePart>
            <namePart type="family">Koponen</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Antonio</namePart>
            <namePart type="family">Pareja-Lora</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>LUHME</publisher>
            <place>
                <placeTerm type="text">Bologna, Italy</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
    </relatedItem>
    <abstract>Large language models have emerged as effective machine translation systems. In this paper, we explore how a general instruction-tuned large language model can be improved for machine translation using relatively few easily produced data resources. Using Slovene as a use case, we improve the GaMS-9B-Instruct model using Direct Preference Optimization (DPO) training on a programmatically curated and enhanced subset of a public dataset. As DPO requires pairs of quality-ranked instances, we generated its training dataset by translating English Wikipedia articles using two LLMs, GaMS-9B-Instruct and EuroLLM-9B-Instruct. We ranked the resulting translations based on heuristics coupled with automatic evaluation metrics such as COMET. The evaluation shows that our fine-tuned model outperforms both models involved in the dataset generation. In comparison to the baseline models, the fine-tuned model achieved a COMET score gain of around 0.04 and 0.02, respectively, on translating Wikipedia articles. It also more consistently avoids language and formatting errors.</abstract>
    <identifier type="citekey">vajda-etal-2025-improving</identifier>
    <location>
        <url>https://aclanthology.org/2025.luhme-1.7/</url>
    </location>
    <part>
        <date>2025-10</date>
        <extent unit="page">
            <start>67</start>
            <end>73</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T Improving LLMs for Machine Translation Using Synthetic Preference Data
%A Vajda, Dario
%A Vreš, Domen
%A Šikonja, Marko Robnik
%Y Cardoso, Henrique Lopes
%Y Sousa-Silva, Rui
%Y Koponen, Maarit
%Y Pareja-Lora, Antonio
%S Proceedings of the 2nd LUHME Workshop
%D 2025
%8 October
%I LUHME
%C Bologna, Italy
%F vajda-etal-2025-improving
%X Large language models have emerged as effective machine translation systems. In this paper, we explore how a general instruction-tuned large language model can be improved for machine translation using relatively few easily produced data resources. Using Slovene as a use case, we improve the GaMS-9B-Instruct model using Direct Preference Optimization (DPO) training on a programmatically curated and enhanced subset of a public dataset. As DPO requires pairs of quality-ranked instances, we generated its training dataset by translating English Wikipedia articles using two LLMs, GaMS-9B-Instruct and EuroLLM-9B-Instruct. We ranked the resulting translations based on heuristics coupled with automatic evaluation metrics such as COMET. The evaluation shows that our fine-tuned model outperforms both models involved in the dataset generation. In comparison to the baseline models, the fine-tuned model achieved a COMET score gain of around 0.04 and 0.02, respectively, on translating Wikipedia articles. It also more consistently avoids language and formatting errors.
%U https://aclanthology.org/2025.luhme-1.7/
%P 67-73

Markdown (Informal)

[Improving LLMs for Machine Translation Using Synthetic Preference Data](https://aclanthology.org/2025.luhme-1.7/) (Vajda et al., LUHME 2025)

• Improving LLMs for Machine Translation Using Synthetic Preference Data (Vajda et al., LUHME 2025)

ACL

• Dario Vajda, Domen Vreš, and Marko Robnik Šikonja. 2025. Improving LLMs for Machine Translation Using Synthetic Preference Data. In Proceedings of the 2nd LUHME Workshop, pages 67–73, Bologna, Italy. LUHME.