Unveiling Visual Perception in Language Models: An Attention Head Analysis Approach

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Bi, Jing, Guo, Junjia, Tang, Yunlong, Wen, Lianggong Bruce, Liu, Zhang, Xu, Chenliang
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918194854232064
author Bi, Jing
Guo, Junjia
Tang, Yunlong
Wen, Lianggong Bruce
Liu, Zhang
Xu, Chenliang
author_facet Bi, Jing
Guo, Junjia
Tang, Yunlong
Wen, Lianggong Bruce
Liu, Zhang
Xu, Chenliang
contents Recent advancements in Multimodal Large Language Models (MLLMs) have demonstrated remarkable progress in visual understanding. This impressive leap raises a compelling question: how can language models, initially trained solely on linguistic data, effectively interpret and process visual content? This paper aims to address this question with systematic investigation across 4 model families and 4 model scales, uncovering a unique class of attention heads that focus specifically on visual content. Our analysis reveals a strong correlation between the behavior of these attention heads, the distribution of attention weights, and their concentration on visual tokens within the input. These findings enhance our understanding of how LLMs adapt to multimodal tasks, demonstrating their potential to bridge the gap between textual and visual understanding. This work paves the way for the development of AI systems capable of engaging with diverse modalities.
format Preprint
id arxiv_https___arxiv_org_abs_2412_18108
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Unveiling Visual Perception in Language Models: An Attention Head Analysis Approach
Bi, Jing
Guo, Junjia
Tang, Yunlong
Wen, Lianggong Bruce
Liu, Zhang
Xu, Chenliang
Computer Vision and Pattern Recognition
Recent advancements in Multimodal Large Language Models (MLLMs) have demonstrated remarkable progress in visual understanding. This impressive leap raises a compelling question: how can language models, initially trained solely on linguistic data, effectively interpret and process visual content? This paper aims to address this question with systematic investigation across 4 model families and 4 model scales, uncovering a unique class of attention heads that focus specifically on visual content. Our analysis reveals a strong correlation between the behavior of these attention heads, the distribution of attention weights, and their concentration on visual tokens within the input. These findings enhance our understanding of how LLMs adapt to multimodal tasks, demonstrating their potential to bridge the gap between textual and visual understanding. This work paves the way for the development of AI systems capable of engaging with diverse modalities.
title Unveiling Visual Perception in Language Models: An Attention Head Analysis Approach
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2412.18108