@inproceedings{wang-etal-2026-mvp,
title = "{MVP}: Multi-View-Guided Token Pruning for Vision-Language Models",
author = "Wang, Chenxi and
Zhong, Dexing and
Shi, Siyu and
Ping, Bowen and
Yan, Hang and
Li, Huayang and
Shao, Huikai",
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.2066/",
pages = "47054--47069",
ISBN = "979-8-89176-459-0",
abstract = "Vision-Language Models (VLMs) achieve strong multimodal understanding by encoding images as sequences of visual tokens, yet cost hundreds and thousands of tokens per image. Existing pruning methods often rely on a single criterion, capturing only a partial view of token necessity. Controlled experiments reveal that token necessity is multi-faceted, requiring the joint exploitation of heterogeneous pruning perspectives. To address this, we propose **MVP-Token**, a novel Multi-View-Guided Token Pruning method to learn token necessity from multiple complementary pruning perspectives. We first establish a dataset **MVP-PruneSet**, which encompasses diverse pruning criteria by running orthogonal pruning methods, including a novel question-guided method **GRASP** at the encoder side. We then propose **MVP-Pruner** to capture such multi-view supervision signals. Specifically, a lightweight dual-attention module predicts soft token-importance scores under multi-view guidance. Extensive experiments demonstrate the effectiveness of MVP-Token across different model architectures. On LLaVA-1.5-7B, our method can surpass the single-view baseline by a clear margin. On Qwen2.5-VL-7B, our method achieves a 1.33{\texttimes} speedup and a 24.6{\%} latency reduction under dynamic resolution and surpasses VisionZip by 1.87{\%} of the average performance over two retention ratios."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-etal-2026-mvp">
<titleInfo>
<title>MVP: Multi-View-Guided Token Pruning for Vision-Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chenxi</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dexing</namePart>
<namePart type="family">Zhong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siyu</namePart>
<namePart type="family">Shi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Bowen</namePart>
<namePart type="family">Ping</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hang</namePart>
<namePart type="family">Yan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Huayang</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Huikai</namePart>
<namePart type="family">Shao</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>Vision-Language Models (VLMs) achieve strong multimodal understanding by encoding images as sequences of visual tokens, yet cost hundreds and thousands of tokens per image. Existing pruning methods often rely on a single criterion, capturing only a partial view of token necessity. Controlled experiments reveal that token necessity is multi-faceted, requiring the joint exploitation of heterogeneous pruning perspectives. To address this, we propose **MVP-Token**, a novel Multi-View-Guided Token Pruning method to learn token necessity from multiple complementary pruning perspectives. We first establish a dataset **MVP-PruneSet**, which encompasses diverse pruning criteria by running orthogonal pruning methods, including a novel question-guided method **GRASP** at the encoder side. We then propose **MVP-Pruner** to capture such multi-view supervision signals. Specifically, a lightweight dual-attention module predicts soft token-importance scores under multi-view guidance. Extensive experiments demonstrate the effectiveness of MVP-Token across different model architectures. On LLaVA-1.5-7B, our method can surpass the single-view baseline by a clear margin. On Qwen2.5-VL-7B, our method achieves a 1.33× speedup and a 24.6% latency reduction under dynamic resolution and surpasses VisionZip by 1.87% of the average performance over two retention ratios.</abstract>
<identifier type="citekey">wang-etal-2026-mvp</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.2066/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>47054</start>
<end>47069</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T MVP: Multi-View-Guided Token Pruning for Vision-Language Models
%A Wang, Chenxi
%A Zhong, Dexing
%A Shi, Siyu
%A Ping, Bowen
%A Yan, Hang
%A Li, Huayang
%A Shao, Huikai
%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 wang-etal-2026-mvp
%X Vision-Language Models (VLMs) achieve strong multimodal understanding by encoding images as sequences of visual tokens, yet cost hundreds and thousands of tokens per image. Existing pruning methods often rely on a single criterion, capturing only a partial view of token necessity. Controlled experiments reveal that token necessity is multi-faceted, requiring the joint exploitation of heterogeneous pruning perspectives. To address this, we propose **MVP-Token**, a novel Multi-View-Guided Token Pruning method to learn token necessity from multiple complementary pruning perspectives. We first establish a dataset **MVP-PruneSet**, which encompasses diverse pruning criteria by running orthogonal pruning methods, including a novel question-guided method **GRASP** at the encoder side. We then propose **MVP-Pruner** to capture such multi-view supervision signals. Specifically, a lightweight dual-attention module predicts soft token-importance scores under multi-view guidance. Extensive experiments demonstrate the effectiveness of MVP-Token across different model architectures. On LLaVA-1.5-7B, our method can surpass the single-view baseline by a clear margin. On Qwen2.5-VL-7B, our method achieves a 1.33× speedup and a 24.6% latency reduction under dynamic resolution and surpasses VisionZip by 1.87% of the average performance over two retention ratios.
%U https://aclanthology.org/2026.emnlp-main.2066/
%P 47054-47069
Markdown (Informal)
[MVP: Multi-View-Guided Token Pruning for Vision-Language Models](https://aclanthology.org/2026.emnlp-main.2066/) (Wang et al., EMNLP 2026)
ACL
- Chenxi Wang, Dexing Zhong, Siyu Shi, Bowen Ping, Hang Yan, Huayang Li, and Huikai Shao. 2026. MVP: Multi-View-Guided Token Pruning for Vision-Language Models. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 47054–47069, Budapest, Hungary. Association for Computational Linguistics.