<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="liu-etal-2026-separation">
    <titleInfo>
        <title>From Separation to Entanglement: Dissecting Feed-Forward Network Functional Organization in Vision–Language Models for Ancient Chinese Documents</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">Ruilin</namePart>
        <namePart type="family">Liu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Dongmei</namePart>
        <namePart type="family">Zhu</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Yuting</namePart>
        <namePart type="family">Wei</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Jin</namePart>
        <namePart type="family">Song</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>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
        </titleInfo>
        <name type="personal">
            <namePart type="given">Hung-Yi</namePart>
            <namePart type="family">Lee</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Sunipa</namePart>
            <namePart type="family">Dev</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Jessy</namePart>
            <namePart type="family">Li</namePart>
            <role>
                <roleTerm authority="marcrelator" type="text">editor</roleTerm>
            </role>
        </name>
        <name type="personal">
            <namePart type="given">Desmond</namePart>
            <namePart type="family">Elliott</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">979-8-89176-462-0</identifier>
    </relatedItem>
    <abstract>While Vision-Language Models (VLMs) have demonstrated strong capabilities in general multimodal tasks, they face significant challenges in low-resource, cross-temporal settings such as ancient Chinese documents, where models must overcome complex visual noise and leverage deep linguistic priors. The functional organization of task-relevant components in existing VLMs remains understudied from a computational interpretability perspective. To address this, we introduce AncientVision-3T, an intervention-oriented diagnostic benchmark designed for mechanistic interpretability experiments on VLMs, organized across OCR, image classification, and image understanding tasks of increasing abstraction. Based on this benchmark, we propose the Ancient Chinese Target-Conditioned Gradient-based Neuron analysis framework (AC-TCGN), which identifies and validates task-relevant neurons for dissecting functional organization in general and domain-specific VLMs. Our experiments demonstrate that AC-TCGN effectively identifies neurons that are critical for model predictions. Moreover, our experimental results reveal a transition from separation to entanglement in the functional organization of decoder feed-forward networks: general VLMs show stronger task-wise separability, whereas the ancient Chinese domain-specific VLM shows greater cross-task neuron sharing and reduced task-wise separability. This suggests enhanced coupling between visual form and historical-semantic processing in ancient Chinese document image-text tasks.</abstract>
    <identifier type="citekey">liu-etal-2026-separation</identifier>
    <location>
        <url>https://aclanthology.org/2026.findings-emnlp.1095/</url>
    </location>
    <part>
        <date>2026-10</date>
        <extent unit="page">
            <start>23176</start>
            <end>23189</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Conference Proceedings
%T From Separation to Entanglement: Dissecting Feed-Forward Network Functional Organization in Vision–Language Models for Ancient Chinese Documents
%A Liu, Ruilin
%A Zhu, Dongmei
%A Wei, Yuting
%A Song, Jin
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F liu-etal-2026-separation
%X While Vision-Language Models (VLMs) have demonstrated strong capabilities in general multimodal tasks, they face significant challenges in low-resource, cross-temporal settings such as ancient Chinese documents, where models must overcome complex visual noise and leverage deep linguistic priors. The functional organization of task-relevant components in existing VLMs remains understudied from a computational interpretability perspective. To address this, we introduce AncientVision-3T, an intervention-oriented diagnostic benchmark designed for mechanistic interpretability experiments on VLMs, organized across OCR, image classification, and image understanding tasks of increasing abstraction. Based on this benchmark, we propose the Ancient Chinese Target-Conditioned Gradient-based Neuron analysis framework (AC-TCGN), which identifies and validates task-relevant neurons for dissecting functional organization in general and domain-specific VLMs. Our experiments demonstrate that AC-TCGN effectively identifies neurons that are critical for model predictions. Moreover, our experimental results reveal a transition from separation to entanglement in the functional organization of decoder feed-forward networks: general VLMs show stronger task-wise separability, whereas the ancient Chinese domain-specific VLM shows greater cross-task neuron sharing and reduced task-wise separability. This suggests enhanced coupling between visual form and historical-semantic processing in ancient Chinese document image-text tasks.
%U https://aclanthology.org/2026.findings-emnlp.1095/
%P 23176-23189

Markdown (Informal)

[From Separation to Entanglement: Dissecting Feed-Forward Network Functional Organization in Vision–Language Models for Ancient Chinese Documents](https://aclanthology.org/2026.findings-emnlp.1095/) (Liu et al., Findings 2026)

• From Separation to Entanglement: Dissecting Feed-Forward Network Functional Organization in Vision–Language Models for Ancient Chinese Documents (Liu et al., Findings 2026)

ACL

• Ruilin Liu, Dongmei Zhu, Yuting Wei, and Jin Song. 2026. From Separation to Entanglement: Dissecting Feed-Forward Network Functional Organization in Vision–Language Models for Ancient Chinese Documents. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 23176–23189, Budapest, Hungary. Association for Computational Linguistics.