@inproceedings{gao-etal-2026-merta,
title = "{MERTA}: Multi-Expert Retrieval for Token Allocation in Video Large Language Models",
author = "Gao, Bo and
Feng, Donghui and
Wang, Qi and
Gu, Qunshan and
Cheng, Zhengxue and
Song, Li",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.862/",
pages = "18232--18246",
ISBN = "979-8-89176-462-0",
abstract = "Video Large Language Models (VideoLLMs) excel in video understanding but suffer from significant computational overheads due to extensive visual tokens. Existing mainstream solutions including visual token compression and query-based Retrieval-Augmented Generation (Video RAG), face notable limitations: the former typically pursues task-agnostic compression of global video semantics, lacking query awareness, while the latter fails to optimize visual token construction and relies on costly, rigid annotation data. To address these challenges, we propose MERTA (Multi-Expert Retrieval for Token Allocation), a plug-and-play framework that integrates selective multi-expert collaborative retrieval with query-aware token allocation. MERTA selectively leverages lightweight experts to capture multimodal semantics, adaptively identifying key frames by measuring the relevance among queries, visual content, and auxiliary annotations. It then dynamically allocates the visual token budget, assigning higher budgets to key frames while keeping the total visual token budget constant. MERTA integrates seamlessly without requiring architectural modifications, demonstrating superior efficiency and performance over existing methods across multiple video understanding benchmarks."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="gao-etal-2026-merta">
<titleInfo>
<title>MERTA: Multi-Expert Retrieval for Token Allocation in Video Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Bo</namePart>
<namePart type="family">Gao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Donghui</namePart>
<namePart type="family">Feng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qi</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qunshan</namePart>
<namePart type="family">Gu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhengxue</namePart>
<namePart type="family">Cheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Li</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>Video Large Language Models (VideoLLMs) excel in video understanding but suffer from significant computational overheads due to extensive visual tokens. Existing mainstream solutions including visual token compression and query-based Retrieval-Augmented Generation (Video RAG), face notable limitations: the former typically pursues task-agnostic compression of global video semantics, lacking query awareness, while the latter fails to optimize visual token construction and relies on costly, rigid annotation data. To address these challenges, we propose MERTA (Multi-Expert Retrieval for Token Allocation), a plug-and-play framework that integrates selective multi-expert collaborative retrieval with query-aware token allocation. MERTA selectively leverages lightweight experts to capture multimodal semantics, adaptively identifying key frames by measuring the relevance among queries, visual content, and auxiliary annotations. It then dynamically allocates the visual token budget, assigning higher budgets to key frames while keeping the total visual token budget constant. MERTA integrates seamlessly without requiring architectural modifications, demonstrating superior efficiency and performance over existing methods across multiple video understanding benchmarks.</abstract>
<identifier type="citekey">gao-etal-2026-merta</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.862/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>18232</start>
<end>18246</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T MERTA: Multi-Expert Retrieval for Token Allocation in Video Large Language Models
%A Gao, Bo
%A Feng, Donghui
%A Wang, Qi
%A Gu, Qunshan
%A Cheng, Zhengxue
%A Song, Li
%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 gao-etal-2026-merta
%X Video Large Language Models (VideoLLMs) excel in video understanding but suffer from significant computational overheads due to extensive visual tokens. Existing mainstream solutions including visual token compression and query-based Retrieval-Augmented Generation (Video RAG), face notable limitations: the former typically pursues task-agnostic compression of global video semantics, lacking query awareness, while the latter fails to optimize visual token construction and relies on costly, rigid annotation data. To address these challenges, we propose MERTA (Multi-Expert Retrieval for Token Allocation), a plug-and-play framework that integrates selective multi-expert collaborative retrieval with query-aware token allocation. MERTA selectively leverages lightweight experts to capture multimodal semantics, adaptively identifying key frames by measuring the relevance among queries, visual content, and auxiliary annotations. It then dynamically allocates the visual token budget, assigning higher budgets to key frames while keeping the total visual token budget constant. MERTA integrates seamlessly without requiring architectural modifications, demonstrating superior efficiency and performance over existing methods across multiple video understanding benchmarks.
%U https://aclanthology.org/2026.findings-emnlp.862/
%P 18232-18246
Markdown (Informal)
[MERTA: Multi-Expert Retrieval for Token Allocation in Video Large Language Models](https://aclanthology.org/2026.findings-emnlp.862/) (Gao et al., Findings 2026)
ACL