Looking Beyond Text: Reducing Language Bias in Large Vision-Language Models via Multimodal Dual-Attention and Soft-Image Guidance
Haozhe Zhao, Shuzheng Si, Liang Chen, Yichi Zhang, Maosong Sun, Baobao Chang, Minjia Zhang
Abstract
Large vision-language models (LVLMs) have achieved impressive results in vision-language tasks. However, Therefore, we propose LACING, designed to address such bias with Mu ̲ L timodal Du ̲ A l-attention Me ̲ C han ̲ I sm (MDA) a ̲ N d Soft-Image ̲ G uidance (SIG). Specifically, MDA adopts a parallel dual-attention mechanism that constructs separate attention for visual and text inputs to enhance integration of visual inputs across model. SIG uses a learnable soft visual prompt during training and inference to replace visual inputs, designed to compel LVLMs to prioritize text inputs during inference. Experiments across different model architectures and scales demonstrate that LACING effectively debiases LVLMs from their language bias, enhancing visual comprehension and reducing hallucinations without additional resources.
BibTeX
@inproceedings{emnlp2025_lookingbeyondtex,
title = {Looking Beyond Text: Reducing Language Bias in Large Vision-Language Models via Multimodal Dual-Attention and Soft-Image Guidance},
author = {Haozhe Zhao and Shuzheng Si and Liang Chen and Yichi Zhang and Maosong Sun and Baobao Chang and Minjia Zhang},
booktitle = {EMNLP 2025},
year = {2025}
}