@inproceedings{fan-etal-2026-minireranker,
title = "mini{R}eranker: Efficient Multimodal Reranking through Visual Cache Reuse and Interaction Sparsity",
author = "Fan, Yingqi and
Lu, Xuan and
Zhao, Anhao and
Tong, Junlong and
Nie, Ping and
Zou, Kai and
Ma, Yunpu and
Zhang, Wei and
Shen, Xiaoyu",
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.1849/",
pages = "39488--39505",
ISBN = "979-8-89176-462-0",
abstract = "Multimodal large language models (MLLMs) have recently shown strong potential as point-wise rerankers by directly modeling query{--}document relevance through next-token prediction. However, point-wise reranking suffers from substantial repeated computation across query{--}document pairs, while the causal structure of transformers allows only prefix segments to be reused via pre-caching. To address the misalignment of existing query-first and document-first formats with both VQA-style prompting and computation-aware reuse, we propose a vision-first formulation that improves both cache reuse efficiency and reranking performance. However, the remaining cost is still considerable and stems from three main sources: (1) *model depth*, for which we reduce active parameters via early exit; (2) *cross-segment attention*, which we restrict to a narrow interaction band across a few layers; and (3) *visual tokens*, where we reduce the number of tokens via embedder-guided pruning. Together, these designs form *miniReranker*, which reduces reranking runtime to {\ensuremath{<}}1{\%} of the dense implementation under high-reuse settings for a single query, while preserving {\ensuremath{>}}96{\%} of the dense model performance."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="fan-etal-2026-minireranker">
<titleInfo>
<title>miniReranker: Efficient Multimodal Reranking through Visual Cache Reuse and Interaction Sparsity</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yingqi</namePart>
<namePart type="family">Fan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xuan</namePart>
<namePart type="family">Lu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anhao</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Junlong</namePart>
<namePart type="family">Tong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ping</namePart>
<namePart type="family">Nie</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kai</namePart>
<namePart type="family">Zou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yunpu</namePart>
<namePart type="family">Ma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaoyu</namePart>
<namePart type="family">Shen</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>Multimodal large language models (MLLMs) have recently shown strong potential as point-wise rerankers by directly modeling query–document relevance through next-token prediction. However, point-wise reranking suffers from substantial repeated computation across query–document pairs, while the causal structure of transformers allows only prefix segments to be reused via pre-caching. To address the misalignment of existing query-first and document-first formats with both VQA-style prompting and computation-aware reuse, we propose a vision-first formulation that improves both cache reuse efficiency and reranking performance. However, the remaining cost is still considerable and stems from three main sources: (1) *model depth*, for which we reduce active parameters via early exit; (2) *cross-segment attention*, which we restrict to a narrow interaction band across a few layers; and (3) *visual tokens*, where we reduce the number of tokens via embedder-guided pruning. Together, these designs form *miniReranker*, which reduces reranking runtime to \ensuremath<1% of the dense implementation under high-reuse settings for a single query, while preserving \ensuremath>96% of the dense model performance.</abstract>
<identifier type="citekey">fan-etal-2026-minireranker</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1849/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>39488</start>
<end>39505</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T miniReranker: Efficient Multimodal Reranking through Visual Cache Reuse and Interaction Sparsity
%A Fan, Yingqi
%A Lu, Xuan
%A Zhao, Anhao
%A Tong, Junlong
%A Nie, Ping
%A Zou, Kai
%A Ma, Yunpu
%A Zhang, Wei
%A Shen, Xiaoyu
%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 fan-etal-2026-minireranker
%X Multimodal large language models (MLLMs) have recently shown strong potential as point-wise rerankers by directly modeling query–document relevance through next-token prediction. However, point-wise reranking suffers from substantial repeated computation across query–document pairs, while the causal structure of transformers allows only prefix segments to be reused via pre-caching. To address the misalignment of existing query-first and document-first formats with both VQA-style prompting and computation-aware reuse, we propose a vision-first formulation that improves both cache reuse efficiency and reranking performance. However, the remaining cost is still considerable and stems from three main sources: (1) *model depth*, for which we reduce active parameters via early exit; (2) *cross-segment attention*, which we restrict to a narrow interaction band across a few layers; and (3) *visual tokens*, where we reduce the number of tokens via embedder-guided pruning. Together, these designs form *miniReranker*, which reduces reranking runtime to \ensuremath<1% of the dense implementation under high-reuse settings for a single query, while preserving \ensuremath>96% of the dense model performance.
%U https://aclanthology.org/2026.findings-emnlp.1849/
%P 39488-39505
Markdown (Informal)
[miniReranker: Efficient Multimodal Reranking through Visual Cache Reuse and Interaction Sparsity](https://aclanthology.org/2026.findings-emnlp.1849/) (Fan et al., Findings 2026)
ACL
- Yingqi Fan, Xuan Lu, Anhao Zhao, Junlong Tong, Ping Nie, Kai Zou, Yunpu Ma, Wei Zhang, and Xiaoyu Shen. 2026. miniReranker: Efficient Multimodal Reranking through Visual Cache Reuse and Interaction Sparsity. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 39488–39505, Budapest, Hungary. Association for Computational Linguistics.