Visual In-Context Learning for Large Vision-Language Models

TL;DR

This paper introduces a novel Visual In-Context Learning (VICL) approach for large vision-language models, improving cross-modal interactions and reasoning through retrieval, summarization, and demonstration composition, validated on multiple datasets.

Contribution

The paper proposes VICL, a comprehensive method combining retrieval, summarization, and demonstration composition to enhance in-context learning in large vision-language models.

Findings

VICL improves visual reasoning performance across five datasets.

The method reduces token count and alleviates cross-modal interaction issues.

In-context unlearning effectively resets model knowledge without retraining.

Abstract

In Large Visual Language Models (LVLMs), the efficacy of In-Context Learning (ICL) remains limited by challenges in cross-modal interactions and representation disparities. To overcome these challenges, we introduce a novel Visual In-Context Learning (VICL) method comprising Visual Demonstration Retrieval, Intent-Oriented Image Summarization, and Intent-Oriented Demonstration Composition. Our approach retrieves images via ''Retrieval & Rerank'' paradigm, summarises images with task intent and task-specific visual parsing, and composes language-based demonstrations that reduce token count and alleviate cross-modal interaction problem. Experimental evaluations on five visual reasoning datasets demonstrate the effectiveness of our method. Moreover, our extensive experiments leverage information flow analysis to elucidate the effectiveness of our method, and investigate the impact of length and position of demonstrations for LVLM. The use of in-context unlearning further shows promise in resetting specific model knowledge without retraining.