From Redundancy to Relevance: Information Flow in LVLMs Across Reasoning Tasks

TL;DR

This paper analyzes how information flows through large vision-language models during reasoning tasks, revealing layer-wise convergence patterns and the impact of visual features on model performance.

Contribution

It introduces an integrated attention analysis method combining LLaVA-CAM with gradient-based insights to study visual information flow in LVLMs.

Findings

Information flow converges in shallow layers

Deeper layers show diversified information processing

Flow patterns vary with context and task

Abstract

Large Vision Language Models (LVLMs) achieve great performance on visual-language reasoning tasks, however, the black-box nature of LVLMs hinders in-depth research on the reasoning mechanism. As all images need to be converted into image tokens to fit the input format of large language models (LLMs) along with natural language prompts, sequential visual representation is essential to the performance of LVLMs, and the information flow analysis approach can be an effective tool for determining interactions between these representations. In this paper, we propose integrating attention analysis with LLaVA-CAM, concretely, attention scores highlight relevant regions during forward propagation, while LLaVA-CAM captures gradient changes through backward propagation, revealing key image features. By exploring the information flow from the perspective of visual representation contribution, we observe that it tends to converge in shallow layers but diversify in deeper layers. To validate our analysis, we conduct comprehensive experiments with truncation strategies across various LVLMs for visual question answering and image captioning tasks, and experimental results not only verify our hypothesis but also reveal a consistent pattern of information flow convergence in the corresponding layers, and the information flow cliff layer will be different due to different contexts. The paper's source code can be accessed from \url{https://github.com/zhangbaijin/From-Redundancy-to-Relevance}