通过自我调用代理进行图像思考
Thinking with Images via Self-Calling Agent
AI总结：
本文提出sCoT范式，通过自我调用代理实现无需多模态交错的高效视觉推理，实验显示其在HR-Bench 4K上性能提升达1.9%且训练效率提高75%。
摘要：
图像思考范式通过将视觉信息作为动态元素整合到链式思维（CoT）中，展示了出色的视觉推理能力。然而，通过强化学习优化交错多模态CoT（iMCoT）仍然具有挑战性，因为它依赖于稀缺的高质量推理数据。在本研究中，我们提出了自我调用链式思维（sCoT），一种新的视觉推理范式，将iMCoT重新表述为仅语言的CoT。具体来说，一个主要代理将复杂的视觉推理任务分解为原子子任务，并调用其虚拟副本，即参数共享的子代理，以在孤立上下文中解决它们。sCoT具有显著的训练有效性和效率，因为它不需要在模态之间进行显式的交错。sCoT采用组相对策略优化来强化有效的推理行为以增强优化。在HR-Bench 4K上的实验表明，sCoT相比强大的基线方法，将整体推理性能提高高达1.9%，仅需约75%的GPU小时。代码可在https://github.com/YWenxi/think-with-images-through-self-calling获取。
英文摘要：
Thinking-with-images paradigms have showcased remarkable visual reasoning capability by integrating visual information as dynamic elements into the Chain-of-Thought (CoT). However, optimizing interleaved multimodal CoT (iMCoT) through reinforcement learning remains challenging, as it relies on scarce high-quality reasoning data. In this study, we propose Self-Calling Chain-of-Thought (sCoT), a novel visual reasoning paradigm that reformulates iMCoT as a language-only CoT with self-calling. Specifically, a main agent decomposes the complex visual reasoning task to atomic subtasks and invokes its virtual replicas, i.e. parameter-sharing subagents, to solve them in isolated context. sCoT enjoys substantial training effectiveness and efficiency, as it requires no explicit interleaving between modalities. sCoT employs group-relative policy optimization to reinforce effective reasoning behavior to enhance optimization. Experiments on HR-Bench 4K show that sCoT improves the overall reasoning performance by up to $1.9\%$ with $\sim 75\%$ fewer GPU hours compared to strong baseline approaches. Code is available at https://github.com/YWenxi/think-with-images-through-self-calling.
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。
