<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="junczys-dowmunt-2025-gemba">
    <titleInfo>
        <title>GEMBA V2: Ten Judgments Are Better Than One</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Marcin</namePart>
        <namePart type="family">Junczys-Dowmunt</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2025-11</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <relatedItem type="host">
        <titleInfo>
            <title>Proceedings of the Tenth Conference on Machine Translation</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Barry</namePart>
            <namePart type="family">Haddow</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Tom</namePart>
            <namePart type="family">Kocmi</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Philipp</namePart>
            <namePart type="family">Koehn</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Christof</namePart>
            <namePart type="family">Monz</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <originInfo>
            <publisher>Association for Computational Linguistics</publisher>
            <place>
                <placeTerm type="text">Suzhou, China</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-341-8</identifier>
    </relatedItem>
    <abstract>We introduce GEMBA-MQM V2, an MQM-inspired, reference-free LLM evaluation metric for the WMT25 Metrics Shared Task (Subtask 1). Building on GEMBA/GEMBA-MQM, we prompt GPT-4.1-mini to produce structured MQM error annotations per segment. We map annotations to scores with 25/5/1 severity weights (minor punctuation = 0.1). To reduce stochastic variance, each segment is scored ten times and aggregated with a reciprocal-rank weighted average (RRWA) after removing outliers beyond 2σ. On the WMT24 MQM test sets, GEMBA-MQM V2 ranks first by average correlation, with strong results across languages and evaluation levels; WMT23 results show comparable performance.</abstract>
    <identifier type="citekey">junczys-dowmunt-2025-gemba</identifier>
    <location>
        <url>https://aclanthology.org/2025.wmt-1.67/</url>
    </location>
    <part>
        <date>2025-11</date>
        <extent unit="page">
            <start>926</start>
            <end>933</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T GEMBA V2: Ten Judgments Are Better Than One
%A Junczys-Dowmunt, Marcin
%Y Haddow, Barry
%Y Kocmi, Tom
%Y Koehn, Philipp
%Y Monz, Christof
%S Proceedings of the Tenth Conference on Machine Translation
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-341-8
%F junczys-dowmunt-2025-gemba
%X We introduce GEMBA-MQM V2, an MQM-inspired, reference-free LLM evaluation metric for the WMT25 Metrics Shared Task (Subtask 1). Building on GEMBA/GEMBA-MQM, we prompt GPT-4.1-mini to produce structured MQM error annotations per segment. We map annotations to scores with 25/5/1 severity weights (minor punctuation = 0.1). To reduce stochastic variance, each segment is scored ten times and aggregated with a reciprocal-rank weighted average (RRWA) after removing outliers beyond 2σ. On the WMT24 MQM test sets, GEMBA-MQM V2 ranks first by average correlation, with strong results across languages and evaluation levels; WMT23 results show comparable performance.
%U https://aclanthology.org/2025.wmt-1.67/
%P 926-933

Markdown (Informal)

[GEMBA V2: Ten Judgments Are Better Than One](https://aclanthology.org/2025.wmt-1.67/) (Junczys-Dowmunt, WMT 2025)

• GEMBA V2: Ten Judgments Are Better Than One (Junczys-Dowmunt, WMT 2025)

ACL

• Marcin Junczys-Dowmunt. 2025. GEMBA V2: Ten Judgments Are Better Than One. In Proceedings of the Tenth Conference on Machine Translation, pages 926–933, Suzhou, China. Association for Computational Linguistics.