更多思考，更少准确性？关于视觉-语言模型中推理的双重性质
More Thought, Less Accuracy? On the Dual Nature of Reasoning in Vision-Language Models
AI总结：
研究揭示了视觉-语言模型中推理的双重性质：虽然增强了逻辑推理能力，但可能导致感知基础能力下降，通过提出VAPO方法改进了模型对视觉信息的依赖。
摘要：
推理能力在大型语言模型中已成为关键能力。通过强化学习（RL），尤其是组相对策略优化（GRPO），这些模型能够解决数学和代码生成等复杂任务。在这些进展的基础上，近期研究试图将推理扩展到视觉-语言模型（VLMs），在多样化的视觉任务中取得了有前景的结果。尽管有进展，我们的研究揭示了多模态推理的双重性质：虽然它显著增强了逻辑推理并促进了解决挑战性问题的性能，但可能逐渐损害感知基础，导致在原本基本的视觉问题上出现识别失败。通过进一步分析，我们将这一现象归因于视觉遗忘，即长时间推理导致模型越来越忽视视觉输入。为了解决这一问题，我们提出了一种简单而有效的方法，即视觉锚定策略优化（VAPO），该方法明确引导推理过程向视觉基础轨迹发展。我们的结果模型VAPO-Thinker-7B显著增强了模型对视觉信息的依赖，并在广泛的已建立基准上取得了新的最先进的结果。项目页面：https://xytian1008.github.io/VAPO/
英文摘要：
Reasoning has emerged as a pivotal capability in Large Language Models (LLMs). Through Reinforcement Learning (RL), typically Group Relative Policy Optimization (GRPO), these models are able to solve complex tasks such as mathematics and code generation. Building on these advances, recent research has sought to extend reasoning to Vision-Language Models (VLMs), yielding promising results across diverse visual tasks. Despite this progress, our study uncovers the dual nature of multimodal reasoning: while it substantially enhances logical inference and facilitates performance on challenging problems, it may gradually impair perceptual grounding, leading to recognition failures on otherwise basic visual questions. Through further analysis, we attribute this phenomenon to visual forgetting, wherein prolonged reasoning causes the model to increasingly disregard visual input. To address this, we propose Vision-Anchored Policy Optimization (VAPO), a simple yet effective method that explicitly steers the reasoning process toward visually grounded trajectories. Our result model, VAPO-Thinker-7B, significantly strengthens the model's reliance on visual information and achieves new state-of-the-art results on a wide range of established benchmarks. Project page: https://xytian1008.github.io/VAPO/
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
