VReST: Enhancing Reasoning in Large Vision-Language Models through Tree Search and Self-Reward Mechanism
This addresses a bottleneck in multimodal AI for tasks requiring advanced reasoning, offering a novel approach without additional training.
The paper tackles the problem of limited complex visual reasoning in large vision-language models by proposing VReST, a training-free method using tree search and self-reward, which achieves state-of-the-art performance on three multimodal mathematical reasoning benchmarks.
Large Vision-Language Models (LVLMs) have shown exceptional performance in multimodal tasks, but their effectiveness in complex visual reasoning is still constrained, especially when employing Chain-of-Thought prompting techniques. In this paper, we propose VReST, a novel training-free approach that enhances Reasoning in LVLMs through Monte Carlo Tree Search and Self-Reward mechanisms. VReST meticulously traverses the reasoning landscape by establishing a search tree, where each node encapsulates a reasoning step, and each path delineates a comprehensive reasoning sequence. Our innovative multimodal Self-Reward mechanism assesses the quality of reasoning steps by integrating the utility of sub-questions, answer correctness, and the relevance of vision-language clues, all without the need for additional models. VReST surpasses current prompting methods and secures state-of-the-art performance across three multimodal mathematical reasoning benchmarks. Furthermore, it substantiates the efficacy of test-time scaling laws in multimodal tasks, offering a promising direction for future research.