<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="martin-etal-2026-ests">
    <titleInfo>
        <title>ESTS at WMT26: Routing-Informed Expert Pruning for Model Compression</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Liu</namePart>
        <namePart type="family">Martin</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Lucas</namePart>
        <namePart type="family">Bandarkar</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Nanyun</namePart>
        <namePart type="family">Peng</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 Eleventh 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">Budapest, Hungary</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">conference publication</genre>
        <identifier type="isbn">979-8-89176-474-3</identifier>
    </relatedItem>
    <abstract>We describe six submissions under the team name ESTS to the unconstrained WMT26 Model Compression Shared Task for English–Simplified Chinese and English–Egyptian Arabic. We submit three compression operating points per translation direction, all derived from GPT-OSS-20B. We use task-specific routing mass to rank experts and cross-lingual routing divergence to allocate retained capacity across layers, then physically remove low-importance experts. The resulting specialists are recovery-tuned on GPT-5.1-generated synthetic translation data and further compressed by applying MXFP4 quantization to the retained expert projection weights. We additionally implement a robust inference system for the instruction-conditioned WMT26 setting, including category inference, output validation, retries, segmented fallback, and source-owned JSON reconstruction. Across our six submissions, parameter counts range from 4.186B to 7.770B and packed artifact sizes from 4.55 to 6.33~GiB. Internal xCOMET-XL evaluation using GPT-5.1 pseudo-references provides an internal comparison across the submitted compression operating points.</abstract>
    <identifier type="citekey">martin-etal-2026-ests</identifier>
    <location>
        <url>https://aclanthology.org/2026.wmt-main.138/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>2009</start>
            <end>2016</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T ESTS at WMT26: Routing-Informed Expert Pruning for Model Compression
%A Martin, Liu
%A Bandarkar, Lucas
%A Peng, Nanyun
%Y Haddow, Barry
%Y Kocmi, Tom
%Y Koehn, Philipp
%Y Monz, Christof
%S Proceedings of the Eleventh Conference on Machine Translation
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-474-3
%F martin-etal-2026-ests
%X We describe six submissions under the team name ESTS to the unconstrained WMT26 Model Compression Shared Task for English–Simplified Chinese and English–Egyptian Arabic. We submit three compression operating points per translation direction, all derived from GPT-OSS-20B. We use task-specific routing mass to rank experts and cross-lingual routing divergence to allocate retained capacity across layers, then physically remove low-importance experts. The resulting specialists are recovery-tuned on GPT-5.1-generated synthetic translation data and further compressed by applying MXFP4 quantization to the retained expert projection weights. We additionally implement a robust inference system for the instruction-conditioned WMT26 setting, including category inference, output validation, retries, segmented fallback, and source-owned JSON reconstruction. Across our six submissions, parameter counts range from 4.186B to 7.770B and packed artifact sizes from 4.55 to 6.33~GiB. Internal xCOMET-XL evaluation using GPT-5.1 pseudo-references provides an internal comparison across the submitted compression operating points.
%U https://aclanthology.org/2026.wmt-main.138/
%P 2009-2016

Markdown (Informal)

[ESTS at WMT26: Routing-Informed Expert Pruning for Model Compression](https://aclanthology.org/2026.wmt-main.138/) (Martin et al., WMT 2026)

• ESTS at WMT26: Routing-Informed Expert Pruning for Model Compression (Martin et al., WMT 2026)

ACL

• Liu Martin, Lucas Bandarkar, and Nanyun Peng. 2026. ESTS at WMT26: Routing-Informed Expert Pruning for Model Compression. In Proceedings of the Eleventh Conference on Machine Translation, pages 2009–2016, Budapest, Hungary. Association for Computational Linguistics.