Don't Look Only Once: Towards Multimodal Interactive Reasoning with
  Selective Visual Revisitation

Jiwan Chung; Junhyeok Kim; Siyeol Kim; Jaeyoung Lee; Min Soo Kim; Youngjae Yu

返回文献探索

Paper · arXiv 2505.18842

Don't Look Only Once: Towards Multimodal Interactive Reasoning with Selective Visual Revisitation

Jiwan Chung, Junhyeok Kim, Siyeol Kim, Jaeyoung Lee, Min Soo Kim, Youngjae Yu

36 upvotesMay 24, 2025arXiv 预印本

AI 摘要

v1 enhances Multimodal Large Language Models by enabling selective and dynamic visual region retrieval during inference, improving performance on multimodal reasoning tasks.

Multimodal Large Language Models (MLLMs)point-and-copy mechanismvisual tokensmultimodal reasoning tracesvisual grounding annotationsMathVistaMathVisionMathVersegrounded multimodal reasoning

Abstract

We present v1, a lightweight extension to Multimodal Large Language Models (MLLMs) that enables selective visual revisitation during inference. While current MLLMs typically consume visual input only once and reason purely over internal memory, v1 introduces a simple point-and-copy mechanism that allows the model to dynamically retrieve relevant image regions throughout the reasoning process. This mechanism augments existing architectures with minimal modifications, enabling contextual access to visual tokens based on the model's evolving hypotheses. To train this capability, we construct v1g, a dataset of 300K multimodal reasoning traces with interleaved visual grounding annotations. Experiments on three multimodal mathematical reasoning benchmarks -- MathVista, MathVision, and MathVerse -- demonstrate that v1 consistently improves performance over comparable baselines, particularly on tasks requiring fine-grained visual reference and multi-step reasoning. Our results suggest that dynamic visual access is a promising direction for enhancing grounded multimodal reasoning. Code, models, and data will be released to support future research.