@inproceedings{wang-etal-2025-aspo,
title = "{ASPO}: Adaptive Sentence-Level Preference Optimization for Fine-Grained Multimodal Reasoning",
author = "Wang, Yeyuan and
Gao, Dehong and
Long, Rujiao and
Yi, Lei and
Jin, Linbo and
Yang, Libin and
Cai, Xiaoyan",
editor = "Che, Wanxiang and
Nabende, Joyce and
Shutova, Ekaterina and
Pilehvar, Mohammad Taher",
booktitle = "Findings of the Association for Computational Linguistics: ACL 2025",
month = jul,
year = "2025",
address = "Vienna, Austria",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-acl.267/",
doi = "10.18653/v1/2025.findings-acl.267",
pages = "5149--5160",
ISBN = "979-8-89176-256-5",
abstract = "Direct Preference Optimization (DPO) has gained significant attention for its simplicity and computational efficiency in aligning large language models (LLMs). Recent advancements have extended DPO to multimodal scenarios, achieving strong performance. However, traditional DPO relies on binary preference optimization, rewarding or penalizing entire responses without considering fine-grained segment correctness, leading to suboptimal solutions. The root of this issue lies in the absence of fine-grained supervision during the optimization process. To address this, we propose Adaptive Sentence-level Preference Optimization (ASPO), which evaluates individual sentences for more precise preference optimization. By dynamically calculating adaptive rewards at the sentence level based on model predictions, ASPO enhances response content assessment without additional models or parameters. This significantly improves the alignment of multimodal features. Extensive experiments show that ASPO substantially enhances the overall performance of multimodal models."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wang-etal-2025-aspo">
<titleInfo>
<title>ASPO: Adaptive Sentence-Level Preference Optimization for Fine-Grained Multimodal Reasoning</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yeyuan</namePart>
<namePart type="family">Wang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dehong</namePart>
<namePart type="family">Gao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rujiao</namePart>
<namePart type="family">Long</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lei</namePart>
<namePart type="family">Yi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Linbo</namePart>
<namePart type="family">Jin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Libin</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiaoyan</namePart>
<namePart type="family">Cai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: ACL 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wanxiang</namePart>
<namePart type="family">Che</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Joyce</namePart>
<namePart type="family">Nabende</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ekaterina</namePart>
<namePart type="family">Shutova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohammad</namePart>
<namePart type="given">Taher</namePart>
<namePart type="family">Pilehvar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Vienna, Austria</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-256-5</identifier>
</relatedItem>
<abstract>Direct Preference Optimization (DPO) has gained significant attention for its simplicity and computational efficiency in aligning large language models (LLMs). Recent advancements have extended DPO to multimodal scenarios, achieving strong performance. However, traditional DPO relies on binary preference optimization, rewarding or penalizing entire responses without considering fine-grained segment correctness, leading to suboptimal solutions. The root of this issue lies in the absence of fine-grained supervision during the optimization process. To address this, we propose Adaptive Sentence-level Preference Optimization (ASPO), which evaluates individual sentences for more precise preference optimization. By dynamically calculating adaptive rewards at the sentence level based on model predictions, ASPO enhances response content assessment without additional models or parameters. This significantly improves the alignment of multimodal features. Extensive experiments show that ASPO substantially enhances the overall performance of multimodal models.</abstract>
<identifier type="citekey">wang-etal-2025-aspo</identifier>
<identifier type="doi">10.18653/v1/2025.findings-acl.267</identifier>
<location>
<url>https://aclanthology.org/2025.findings-acl.267/</url>
</location>
<part>
<date>2025-07</date>
<extent unit="page">
<start>5149</start>
<end>5160</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T ASPO: Adaptive Sentence-Level Preference Optimization for Fine-Grained Multimodal Reasoning
%A Wang, Yeyuan
%A Gao, Dehong
%A Long, Rujiao
%A Yi, Lei
%A Jin, Linbo
%A Yang, Libin
%A Cai, Xiaoyan
%Y Che, Wanxiang
%Y Nabende, Joyce
%Y Shutova, Ekaterina
%Y Pilehvar, Mohammad Taher
%S Findings of the Association for Computational Linguistics: ACL 2025
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-256-5
%F wang-etal-2025-aspo
%X Direct Preference Optimization (DPO) has gained significant attention for its simplicity and computational efficiency in aligning large language models (LLMs). Recent advancements have extended DPO to multimodal scenarios, achieving strong performance. However, traditional DPO relies on binary preference optimization, rewarding or penalizing entire responses without considering fine-grained segment correctness, leading to suboptimal solutions. The root of this issue lies in the absence of fine-grained supervision during the optimization process. To address this, we propose Adaptive Sentence-level Preference Optimization (ASPO), which evaluates individual sentences for more precise preference optimization. By dynamically calculating adaptive rewards at the sentence level based on model predictions, ASPO enhances response content assessment without additional models or parameters. This significantly improves the alignment of multimodal features. Extensive experiments show that ASPO substantially enhances the overall performance of multimodal models.
%R 10.18653/v1/2025.findings-acl.267
%U https://aclanthology.org/2025.findings-acl.267/
%U https://doi.org/10.18653/v1/2025.findings-acl.267
%P 5149-5160
Markdown (Informal)
[ASPO: Adaptive Sentence-Level Preference Optimization for Fine-Grained Multimodal Reasoning](https://aclanthology.org/2025.findings-acl.267/) (Wang et al., Findings 2025)
ACL