@inproceedings{jung-etal-2026-fork,
title = "Fork-Merge Decoding: Enhancing Multimodal Understanding in Audio-Visual Large Language Models",
author = "Jung, Chaeyoung and
Jang, Youngjoon and
Choi, Jongmin and
Chung, Joon Son",
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.1392/",
pages = "29610--29629",
ISBN = "979-8-89176-462-0",
abstract = "The goal of this work is to enhance balanced multimodal understanding in audio-visual large language models (AV-LLMs) by addressing modality bias without additional training. In current AV-LLMs, audio and video features are typically processed jointly in the decoder. While this strategy facilitates unified multimodal understanding, it may introduce modality bias, where the model disproportionately relies on a single dominant modality. In this work, we provide an in-depth analysis of this phenomenon and propose Fork-Merge Decoding (FMD) as a targeted solution. FMD is a simple, training-free inference strategy that rectifies this imbalance without architectural modifications. Specifically, FMD isolates audio and visual inputs in the early decoding layers (fork) to enforce modality-specific reasoning, before integrating them for joint interpretation (merge). This two-stage mechanism effectively mitigates the dominance by ensuring balanced contributions from both modalities. Experiments on VideoLLaMA2, video-SALMONN, and Qwen2.5-Omni demonstrate that FMD consistently enhances multimodal understanding across diverse benchmarks."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jung-etal-2026-fork">
<titleInfo>
<title>Fork-Merge Decoding: Enhancing Multimodal Understanding in Audio-Visual Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Chaeyoung</namePart>
<namePart type="family">Jung</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Youngjoon</namePart>
<namePart type="family">Jang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jongmin</namePart>
<namePart type="family">Choi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Joon</namePart>
<namePart type="given">Son</namePart>
<namePart type="family">Chung</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>The goal of this work is to enhance balanced multimodal understanding in audio-visual large language models (AV-LLMs) by addressing modality bias without additional training. In current AV-LLMs, audio and video features are typically processed jointly in the decoder. While this strategy facilitates unified multimodal understanding, it may introduce modality bias, where the model disproportionately relies on a single dominant modality. In this work, we provide an in-depth analysis of this phenomenon and propose Fork-Merge Decoding (FMD) as a targeted solution. FMD is a simple, training-free inference strategy that rectifies this imbalance without architectural modifications. Specifically, FMD isolates audio and visual inputs in the early decoding layers (fork) to enforce modality-specific reasoning, before integrating them for joint interpretation (merge). This two-stage mechanism effectively mitigates the dominance by ensuring balanced contributions from both modalities. Experiments on VideoLLaMA2, video-SALMONN, and Qwen2.5-Omni demonstrate that FMD consistently enhances multimodal understanding across diverse benchmarks.</abstract>
<identifier type="citekey">jung-etal-2026-fork</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.1392/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>29610</start>
<end>29629</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Fork-Merge Decoding: Enhancing Multimodal Understanding in Audio-Visual Large Language Models
%A Jung, Chaeyoung
%A Jang, Youngjoon
%A Choi, Jongmin
%A Chung, Joon Son
%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 jung-etal-2026-fork
%X The goal of this work is to enhance balanced multimodal understanding in audio-visual large language models (AV-LLMs) by addressing modality bias without additional training. In current AV-LLMs, audio and video features are typically processed jointly in the decoder. While this strategy facilitates unified multimodal understanding, it may introduce modality bias, where the model disproportionately relies on a single dominant modality. In this work, we provide an in-depth analysis of this phenomenon and propose Fork-Merge Decoding (FMD) as a targeted solution. FMD is a simple, training-free inference strategy that rectifies this imbalance without architectural modifications. Specifically, FMD isolates audio and visual inputs in the early decoding layers (fork) to enforce modality-specific reasoning, before integrating them for joint interpretation (merge). This two-stage mechanism effectively mitigates the dominance by ensuring balanced contributions from both modalities. Experiments on VideoLLaMA2, video-SALMONN, and Qwen2.5-Omni demonstrate that FMD consistently enhances multimodal understanding across diverse benchmarks.
%U https://aclanthology.org/2026.findings-emnlp.1392/
%P 29610-29629
Markdown (Informal)
[Fork-Merge Decoding: Enhancing Multimodal Understanding in Audio-Visual Large Language Models](https://aclanthology.org/2026.findings-emnlp.1392/) (Jung et al., Findings 2026)
ACL