<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="liu-etal-2026-distilvdr">
    <titleInfo>
        <title>DistilVDR: A Compact End-to-End Visual Document Retriever via Dual-Student Distillation</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Zhuchenyang</namePart>
        <namePart type="family">Liu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Ziyi</namePart>
        <namePart type="family">Wang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yao</namePart>
        <namePart type="family">Zhang</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yu</namePart>
        <namePart type="family">Xiao</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>The Proceedings of Workshop on Document Intelligence and Understanding</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Santosh</namePart>
            <namePart type="family">T.Y.S.S.</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Amit</namePart>
            <namePart type="family">Agarwal</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Vivek</namePart>
            <namePart type="family">Gupta</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Yova</namePart>
            <namePart type="family">Kementchedjhieva</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Aparna</namePart>
            <namePart type="family">Garimella</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Yufang</namePart>
            <namePart type="family">Hou</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">979-8-89176-477-4</identifier>
    </relatedItem>
    <abstract>Visual document retrieval (VDR) is dominated by multi-billion-parameter models that are slow to index at full corpus scale and expensive to serve. Prior compression routes either train a smaller multi-vector encoder from scratch or distil only the query side; neither yields a compact single-vector retriever end-to-end. We present DistilVDR, a 527M end-to-end VDR system distilled bilaterally from a single 8B vision-language teacher under a pointwise cosine alignment loss. All supervision comes from the frozen teacher’s embedding space, which was itself trained with relevance supervision, so the student objective needs no relevance labels, negative sampling, or contrastive term. We match VDR’s text-query and image-document input asymmetry with an asymmetric encoder-only student that concentrates visual capacity on the document side and keeps the query side at 70M parameters. We release two variants that share the same encoders and training and differ only in the document encoder’s visual-tile budget: DistilVDR-HiRes attains 61.74 average NDCG@5 on ViDoRe v1+v2+v3 (86.9% of the 8B teacher) and leads every reproduced sub-1B baseline on the high-resolution-sensitive v3 benchmark, while DistilVDR-Fast attains 59.98 at a 3\times smaller visual-token budget. Both variants store one million documents in a 15.6 timessmaller index than the strongest sub-1B multi-vector baseline,and DistilVDR-Fast indexes 5.2 times faster than that baseline and 18 times faster than the teacher.</abstract>
    <identifier type="citekey">liu-etal-2026-distilvdr</identifier>
    <location>
        <url>https://aclanthology.org/2026.docinsights-1.26/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>385</start>
            <end>399</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T DistilVDR: A Compact End-to-End Visual Document Retriever via Dual-Student Distillation
%A Liu, Zhuchenyang
%A Wang, Ziyi
%A Zhang, Yao
%A Xiao, Yu
%Y T.Y.S.S., Santosh
%Y Agarwal, Amit
%Y Gupta, Vivek
%Y Kementchedjhieva, Yova
%Y Garimella, Aparna
%Y Hou, Yufang
%S The Proceedings of Workshop on Document Intelligence and Understanding
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-477-4
%F liu-etal-2026-distilvdr
%X Visual document retrieval (VDR) is dominated by multi-billion-parameter models that are slow to index at full corpus scale and expensive to serve. Prior compression routes either train a smaller multi-vector encoder from scratch or distil only the query side; neither yields a compact single-vector retriever end-to-end. We present DistilVDR, a 527M end-to-end VDR system distilled bilaterally from a single 8B vision-language teacher under a pointwise cosine alignment loss. All supervision comes from the frozen teacher’s embedding space, which was itself trained with relevance supervision, so the student objective needs no relevance labels, negative sampling, or contrastive term. We match VDR’s text-query and image-document input asymmetry with an asymmetric encoder-only student that concentrates visual capacity on the document side and keeps the query side at 70M parameters. We release two variants that share the same encoders and training and differ only in the document encoder’s visual-tile budget: DistilVDR-HiRes attains 61.74 average NDCG@5 on ViDoRe v1+v2+v3 (86.9% of the 8B teacher) and leads every reproduced sub-1B baseline on the high-resolution-sensitive v3 benchmark, while DistilVDR-Fast attains 59.98 at a 3\times smaller visual-token budget. Both variants store one million documents in a 15.6 timessmaller index than the strongest sub-1B multi-vector baseline,and DistilVDR-Fast indexes 5.2 times faster than that baseline and 18 times faster than the teacher.
%U https://aclanthology.org/2026.docinsights-1.26/
%P 385-399

Markdown (Informal)

[DistilVDR: A Compact End-to-End Visual Document Retriever via Dual-Student Distillation](https://aclanthology.org/2026.docinsights-1.26/) (Liu et al., DocInsights 2026)

• DistilVDR: A Compact End-to-End Visual Document Retriever via Dual-Student Distillation (Liu et al., DocInsights 2026)

ACL

• Zhuchenyang Liu, Ziyi Wang, Yao Zhang, and Yu Xiao. 2026. DistilVDR: A Compact End-to-End Visual Document Retriever via Dual-Student Distillation. In The Proceedings of Workshop on Document Intelligence and Understanding, pages 385–399, Budapest, Hungary. Association for Computational Linguistics.