@inproceedings{jia-etal-2025-llm,
title = "{LLM}-empowered Dynamic Prompt Routing for Vision-Language Models Tuning under Long-Tailed Distributions",
author = "Jia, Yongju and
Ma, Jiarui and
Li, Xiangxian and
Zhang, Baiqiao and
Cao, Xianhui and
Liu, Juan and
Bian, Yulong",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-emnlp.799/",
doi = "10.18653/v1/2025.findings-emnlp.799",
pages = "14809--14822",
ISBN = "979-8-89176-335-7",
abstract = "Pre-trained vision-language models (VLMs), such as CLIP, have demonstrated impressive capability in visual tasks, but their fine-tuning often suffers from bias in class-imbalanced scenes. Recent works have introduced large language models (LLMs) to enhance VLM fine-tuning withsupplementaryy semantic information. However, they often overlook inherent class imbalance in VLMs' pre-training, which may lead to bias accumulation in downstream tasks. To address this problem, this paper proposes a Multi-dimensional Dynamic Prompt Routing (MDPR) framework. MDPR constructs a comprehensive knowledge base for classes, spanning multiple visual-semantic dimensions. During fine-tuning, the dynamic routing mechanism aligns global visual classes, retrieves optimal prompts, and balances fine-grained semantics, yielding stable predictions through logits fusion. Extensive experiments on long-tailed benchmarks, including CIFAR-LT, ImageNet-LT, and Places-LT, demonstrate that MDPR achieves comparable results with current SOTA methods. Ablation studies further confirm the effectiveness of our semantic library for tail classes and show that our dynamic routing operates with a slight increase in computational overhead, making MDPR a flexible and efficient enhancement for VLM fine-tuning under data imbalance. The codes are available in \url{https://github.com/Sha843/MDPR}."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="jia-etal-2025-llm">
<titleInfo>
<title>LLM-empowered Dynamic Prompt Routing for Vision-Language Models Tuning under Long-Tailed Distributions</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yongju</namePart>
<namePart type="family">Jia</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiarui</namePart>
<namePart type="family">Ma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiangxian</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Baiqiao</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xianhui</namePart>
<namePart type="family">Cao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Juan</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yulong</namePart>
<namePart type="family">Bian</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christos</namePart>
<namePart type="family">Christodoulopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carolyn</namePart>
<namePart type="family">Rose</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Violet</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-335-7</identifier>
</relatedItem>
<abstract>Pre-trained vision-language models (VLMs), such as CLIP, have demonstrated impressive capability in visual tasks, but their fine-tuning often suffers from bias in class-imbalanced scenes. Recent works have introduced large language models (LLMs) to enhance VLM fine-tuning withsupplementaryy semantic information. However, they often overlook inherent class imbalance in VLMs’ pre-training, which may lead to bias accumulation in downstream tasks. To address this problem, this paper proposes a Multi-dimensional Dynamic Prompt Routing (MDPR) framework. MDPR constructs a comprehensive knowledge base for classes, spanning multiple visual-semantic dimensions. During fine-tuning, the dynamic routing mechanism aligns global visual classes, retrieves optimal prompts, and balances fine-grained semantics, yielding stable predictions through logits fusion. Extensive experiments on long-tailed benchmarks, including CIFAR-LT, ImageNet-LT, and Places-LT, demonstrate that MDPR achieves comparable results with current SOTA methods. Ablation studies further confirm the effectiveness of our semantic library for tail classes and show that our dynamic routing operates with a slight increase in computational overhead, making MDPR a flexible and efficient enhancement for VLM fine-tuning under data imbalance. The codes are available in https://github.com/Sha843/MDPR.</abstract>
<identifier type="citekey">jia-etal-2025-llm</identifier>
<identifier type="doi">10.18653/v1/2025.findings-emnlp.799</identifier>
<location>
<url>https://aclanthology.org/2025.findings-emnlp.799/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>14809</start>
<end>14822</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T LLM-empowered Dynamic Prompt Routing for Vision-Language Models Tuning under Long-Tailed Distributions
%A Jia, Yongju
%A Ma, Jiarui
%A Li, Xiangxian
%A Zhang, Baiqiao
%A Cao, Xianhui
%A Liu, Juan
%A Bian, Yulong
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Findings of the Association for Computational Linguistics: EMNLP 2025
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-335-7
%F jia-etal-2025-llm
%X Pre-trained vision-language models (VLMs), such as CLIP, have demonstrated impressive capability in visual tasks, but their fine-tuning often suffers from bias in class-imbalanced scenes. Recent works have introduced large language models (LLMs) to enhance VLM fine-tuning withsupplementaryy semantic information. However, they often overlook inherent class imbalance in VLMs’ pre-training, which may lead to bias accumulation in downstream tasks. To address this problem, this paper proposes a Multi-dimensional Dynamic Prompt Routing (MDPR) framework. MDPR constructs a comprehensive knowledge base for classes, spanning multiple visual-semantic dimensions. During fine-tuning, the dynamic routing mechanism aligns global visual classes, retrieves optimal prompts, and balances fine-grained semantics, yielding stable predictions through logits fusion. Extensive experiments on long-tailed benchmarks, including CIFAR-LT, ImageNet-LT, and Places-LT, demonstrate that MDPR achieves comparable results with current SOTA methods. Ablation studies further confirm the effectiveness of our semantic library for tail classes and show that our dynamic routing operates with a slight increase in computational overhead, making MDPR a flexible and efficient enhancement for VLM fine-tuning under data imbalance. The codes are available in https://github.com/Sha843/MDPR.
%R 10.18653/v1/2025.findings-emnlp.799
%U https://aclanthology.org/2025.findings-emnlp.799/
%U https://doi.org/10.18653/v1/2025.findings-emnlp.799
%P 14809-14822
Markdown (Informal)
[LLM-empowered Dynamic Prompt Routing for Vision-Language Models Tuning under Long-Tailed Distributions](https://aclanthology.org/2025.findings-emnlp.799/) (Jia et al., Findings 2025)
ACL