@inproceedings{liu-etal-2026-rvsd,
title = "{RVSD}: Retrieval Vision Sparse Decoding for Mitigating Visual Hallucinations in Large Vision-Language Models",
author = "Liu, Canjie and
Kang, Jiawen and
Wen, Jinbo and
Zhong, Zishao",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.2009/",
pages = "45789--45806",
ISBN = "979-8-89176-459-0",
abstract = "Large vision-language models have achieved remarkable success in vision-language tasks. However, they remain prone to Visual Hallucinations (VHs), undermining their reliability in real-world applications. Existing solutions typically require curated datasets, additional training, or multi-round decoding, resulting in considerable efficiency overhead. In this paper, we propose **RVSD** (**R**etrieval **V**ision **S**parse **D**ecoding), a training-free and plug-and-play decoding framework that, for the first time, unifies token sparsification and **Semantic-Space Visual Retrieval** (SSVR) within a single decoding pass. Within RVSD, we introduce a **semantics-directed token selection** strategy that selectively sparsifies redundant tokens while preserving critical visual information. We further propose the SSVR mechanism, which reformulates visual compensation as an on-demand cross-modal retrieval process within a shared semantic space. Extensive experiments demonstrate that RVSD achieves state-of-the-art performance in mitigating VHs while maintaining robust suppression capabilities under long-context generation settings. Our code is available at https://github.com/canjie-liu/RVSD."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="liu-etal-2026-rvsd">
<titleInfo>
<title>RVSD: Retrieval Vision Sparse Decoding for Mitigating Visual Hallucinations in Large Vision-Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Canjie</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiawen</namePart>
<namePart type="family">Kang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jinbo</namePart>
<namePart type="family">Wen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zishao</namePart>
<namePart type="family">Zhong</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 2026 Conference on Empirical Methods in Natural Language Processing</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">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Large vision-language models have achieved remarkable success in vision-language tasks. However, they remain prone to Visual Hallucinations (VHs), undermining their reliability in real-world applications. Existing solutions typically require curated datasets, additional training, or multi-round decoding, resulting in considerable efficiency overhead. In this paper, we propose **RVSD** (**R**etrieval **V**ision **S**parse **D**ecoding), a training-free and plug-and-play decoding framework that, for the first time, unifies token sparsification and **Semantic-Space Visual Retrieval** (SSVR) within a single decoding pass. Within RVSD, we introduce a **semantics-directed token selection** strategy that selectively sparsifies redundant tokens while preserving critical visual information. We further propose the SSVR mechanism, which reformulates visual compensation as an on-demand cross-modal retrieval process within a shared semantic space. Extensive experiments demonstrate that RVSD achieves state-of-the-art performance in mitigating VHs while maintaining robust suppression capabilities under long-context generation settings. Our code is available at https://github.com/canjie-liu/RVSD.</abstract>
<identifier type="citekey">liu-etal-2026-rvsd</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.2009/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>45789</start>
<end>45806</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T RVSD: Retrieval Vision Sparse Decoding for Mitigating Visual Hallucinations in Large Vision-Language Models
%A Liu, Canjie
%A Kang, Jiawen
%A Wen, Jinbo
%A Zhong, Zishao
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F liu-etal-2026-rvsd
%X Large vision-language models have achieved remarkable success in vision-language tasks. However, they remain prone to Visual Hallucinations (VHs), undermining their reliability in real-world applications. Existing solutions typically require curated datasets, additional training, or multi-round decoding, resulting in considerable efficiency overhead. In this paper, we propose **RVSD** (**R**etrieval **V**ision **S**parse **D**ecoding), a training-free and plug-and-play decoding framework that, for the first time, unifies token sparsification and **Semantic-Space Visual Retrieval** (SSVR) within a single decoding pass. Within RVSD, we introduce a **semantics-directed token selection** strategy that selectively sparsifies redundant tokens while preserving critical visual information. We further propose the SSVR mechanism, which reformulates visual compensation as an on-demand cross-modal retrieval process within a shared semantic space. Extensive experiments demonstrate that RVSD achieves state-of-the-art performance in mitigating VHs while maintaining robust suppression capabilities under long-context generation settings. Our code is available at https://github.com/canjie-liu/RVSD.
%U https://aclanthology.org/2026.emnlp-main.2009/
%P 45789-45806
Markdown (Informal)
[RVSD: Retrieval Vision Sparse Decoding for Mitigating Visual Hallucinations in Large Vision-Language Models](https://aclanthology.org/2026.emnlp-main.2009/) (Liu et al., EMNLP 2026)
ACL