arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2602.13310cs.CVcs.AI
收藏

视觉并行思考器:用于视觉理解的分而治之推理

Visual Para-Thinker: Divide-and-Conquer Reasoning for Visual Comprehension

Zhejiang University(浙江大学);Hunan University(湖南大学);University of Chinese Academy of Sciences(中国科学院大学);Independent Researcher(独立研究者)

Haoran Xu, Hongyu Wang, Jiaze Li, Shunpeng Chen, Zizhao Tong, Jianzhong Ju, Zhenbo Luo, Jian Luan

AI总结:

本文提出视觉并行思考器,通过并行推理框架提升视觉理解能力,结合Pa-Attention和LPRoPE实现高效多模态处理,验证了并行推理在视觉领域的有效性。

摘要:

现有LLM测试时间扩展定律强调通过延长推理长度产生自我反思行为。然而,这种垂直扩展策略常在模型陷入特定思维模式时遇到瓶颈。通过从深度转向并行性,并行思维缓解了探索范围的缩小。然而,将此范式扩展到视觉领域仍是一个开放性问题。本文首先探讨了视觉分区在并行推理中的作用,并提出两种不同策略。基于此,我们引入视觉并行思考器,代表首个针对大规模语言模型的并行推理框架。为保持路径独立性和促进推理多样性,我们的方法整合了Pa-Attention与LPRoPE。利用vLLM框架,我们开发了原生多模态实现,以实现高效并行处理。在V*、CountBench、RefCOCO和HallusionBench等基准数据集上的实验证实,视觉并行思考器成功将并行推理的优势扩展到视觉领域。

英文摘要:

Existing LLM test-time scaling laws emphasize the emergence of self-reflective behaviors through extended reasoning length. Nevertheless, this vertical scaling strategy often encounters plateaus in exploration as the model becomes locked into specific thinking pattern. By shifting from depth to parallelism, parallel thinking mitigates the narrowing of exploration. However, the extension of this paradigm to visual domain remains an open research question. In this paper, we first examine the role of visual partitioning in parallelized reasoning and subsequently propose two distinct strategies. Based on the above, we introduce Visual Para-Thinker, representing the inaugural parallel reasoning framework for MLLMs. To maintain path independence and promote diversity in reasoning, our approach integrates Pa-Attention alongside LPRoPE. Leveraging the vLLM framework, we have developed a native multimodal implementation that facilitates high-efficiency parallel processing. Empirical results on benchmark datasets such as V*, CountBench, RefCOCO, and HallusionBench confirm that Visual Para-Thinker successfully extends the benefits of parallel reasoning to the visual domain.

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。

↑