Unraveling Cross-Modality Knowledge Conflicts in Large Vision-Language Models

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhu, Tinghui, Liu, Qin, Wang, Fei, Tu, Zhengzhong, Chen, Muhao
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909345277542400
author Zhu, Tinghui
Liu, Qin
Wang, Fei
Tu, Zhengzhong
Chen, Muhao
author_facet Zhu, Tinghui
Liu, Qin
Wang, Fei
Tu, Zhengzhong
Chen, Muhao
contents Large Vision-Language Models (LVLMs) have demonstrated impressive capabilities for capturing and reasoning over multimodal inputs. However, these models are prone to parametric knowledge conflicts, which arise from inconsistencies of represented knowledge between their vision and language components. In this paper, we formally define the problem of $\textbf{cross-modality parametric knowledge conflict}$ and present a systematic approach to detect, interpret, and mitigate them. We introduce a pipeline that identifies conflicts between visual and textual answers, showing a persistently high conflict rate across modalities in recent LVLMs regardless of the model size. We further investigate how these conflicts interfere with the inference process and propose a contrastive metric to discern the conflicting samples from the others. Building on these insights, we develop a novel dynamic contrastive decoding method that removes undesirable logits inferred from the less confident modality components based on answer confidence. For models that do not provide logits, we also introduce two prompt-based strategies to mitigate the conflicts. Our methods achieve promising improvements in accuracy on both the ViQuAE and InfoSeek datasets. Specifically, using LLaVA-34B, our proposed dynamic contrastive decoding improves an average accuracy of 2.24%.
format Preprint
id arxiv_https___arxiv_org_abs_2410_03659
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Unraveling Cross-Modality Knowledge Conflicts in Large Vision-Language Models
Zhu, Tinghui
Liu, Qin
Wang, Fei
Tu, Zhengzhong
Chen, Muhao
Computer Vision and Pattern Recognition
Computation and Language
Large Vision-Language Models (LVLMs) have demonstrated impressive capabilities for capturing and reasoning over multimodal inputs. However, these models are prone to parametric knowledge conflicts, which arise from inconsistencies of represented knowledge between their vision and language components. In this paper, we formally define the problem of $\textbf{cross-modality parametric knowledge conflict}$ and present a systematic approach to detect, interpret, and mitigate them. We introduce a pipeline that identifies conflicts between visual and textual answers, showing a persistently high conflict rate across modalities in recent LVLMs regardless of the model size. We further investigate how these conflicts interfere with the inference process and propose a contrastive metric to discern the conflicting samples from the others. Building on these insights, we develop a novel dynamic contrastive decoding method that removes undesirable logits inferred from the less confident modality components based on answer confidence. For models that do not provide logits, we also introduce two prompt-based strategies to mitigate the conflicts. Our methods achieve promising improvements in accuracy on both the ViQuAE and InfoSeek datasets. Specifically, using LLaVA-34B, our proposed dynamic contrastive decoding improves an average accuracy of 2.24%.
title Unraveling Cross-Modality Knowledge Conflicts in Large Vision-Language Models
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2410.03659