Open Vision Reasoner: Transferring Linguistic Cognitive Behavior for Visual Reasoning

• URL: http://arxiv.org/abs/2507.05255v1
• Date: Mon, 07 Jul 2025 17:59:03 GMT
• Title: Open Vision Reasoner: Transferring Linguistic Cognitive Behavior for Visual Reasoning
• Authors: Yana Wei, Liang Zhao, Jianjian Sun, Kangheng Lin, Jisheng Yin, Jingcheng Hu, Yinmin Zhang, En Yu, Haoran Lv, Zejia Weng, Jia Wang, Chunrui Han, Yuang Peng, Qi Han, Zheng Ge, Xiangyu Zhang, Daxin Jiang, Vishal M. Patel,
• Abstract summary: We introduce a two-stage paradigm built on Qwen2.5-VL-7B: a massive linguistic cold-start fine-tuning, followed by multimodal reinforcement learning.<n>This pioneering work reveals three fundamental insights: 1) Behavior transfer emerges surprisingly early in cold start due to linguistic mental imagery, 2) Cold start broadly memorizes visual behaviors, while RL critically discerns and scales up effective patterns.<n>Our resulting model, Open-Vision-Reasoner (OVR), achieves state-of-the-art performance on a suite of reasoning benchmarks, including 95.3% on MATH500, 51.8% on MathVision and 54.6%
• Score: 60.292578064172524
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: The remarkable reasoning capability of large language models (LLMs) stems from cognitive behaviors that emerge through reinforcement with verifiable rewards. This work investigates how to transfer this principle to Multimodal LLMs (MLLMs) to unlock advanced visual reasoning. We introduce a two-stage paradigm built on Qwen2.5-VL-7B: a massive linguistic cold-start fine-tuning, followed by multimodal reinforcement learning (RL) spanning nearly 1,000 steps, surpassing all previous open-source efforts in scale. This pioneering work reveals three fundamental insights: 1) Behavior transfer emerges surprisingly early in cold start due to linguistic mental imagery. 2) Cold start broadly memorizes visual behaviors, while RL critically discerns and scales up effective patterns. 3) Transfer strategically favors high-utility behaviors such as visual reflection. Our resulting model, Open-Vision-Reasoner (OVR), achieves state-of-the-art performance on a suite of reasoning benchmarks, including 95.3% on MATH500, 51.8% on MathVision and 54.6% on MathVerse. We release our model, data, and training dynamics to catalyze the development of more capable, behavior-aligned multimodal reasoners.

Related papers

• Spatial Mental Modeling from Limited Views [71.57140964322559]
  Our new MindCube benchmark with 21,154 questions across 3,268 images exposes this critical gap.<n>Using MindCube, we evaluate how well Vision Language Models (VLMs) build robust spatial mental models.<n>We then explore three approaches to help VLMs approximate spatial mental models, including unseen intermediate views, natural language reasoning chains, and cognitive maps.
  arXiv  Detail & Related papers  (2025-06-26T16:38:19Z)
• OpenVLThinker: Complex Vision-Language Reasoning via Iterative SFT-RL Cycles [91.88062410741833]
  We introduce OpenVLThinker, one of the first open-source large vision-language models (LVLMs) to exhibit sophisticated chain-of-thought reasoning.<n>We show that OpenVLThinker-7B consistently advances performance across six benchmarks demanding mathematical and general reasoning.
  arXiv  Detail & Related papers  (2025-03-21T17:52:43Z)
• Mitigating Visual Forgetting via Take-along Visual Conditioning for Multi-modal Long CoT Reasoning [53.790502697674754]
  We propose Take-along Visual Conditioning (TVC), a strategy that shifts image input to critical reasoning stages.<n>TVC helps the model retain attention to the visual components throughout the reasoning.<n>Our approach achieves state-of-the-art performance on average across five mathematical reasoning benchmarks.
  arXiv  Detail & Related papers  (2025-03-17T16:45:12Z)
• Human Cognitive Benchmarks Reveal Foundational Visual Gaps in MLLMs [65.93003087656754]
  VisFactor is a benchmark that digitizes 20 vision-centric subtests from a well-established cognitive psychology assessment.<n>We evaluate 20 frontier Multimodal Large Language Models (MLLMs) from GPT, Gemini, Claude, LLaMA, Qwen, and SEED families.<n>The best-performing model achieves a score of only 25.19 out of 100, with consistent failures on tasks such as mental rotation, spatial relation inference, and figure-ground discrimination.
  arXiv  Detail & Related papers  (2025-02-23T04:21:32Z)
• Why Vision Language Models Struggle with Visual Arithmetic? Towards Enhanced Chart and Geometry Understanding [94.64781599202882]
  Vision Language Models (VLMs) have achieved remarkable progress in multimodal tasks.<n>They often struggle with visual arithmetic, seemingly simple capabilities like object counting or length comparison.<n>We propose CogAlign, a novel post-training strategy inspired by Piaget's theory of cognitive development.
  arXiv  Detail & Related papers  (2025-02-17T06:54:49Z)
• Socratic Questioning: Learn to Self-guide Multimodal Reasoning in the Wild [35.91285472401222]
  We devise an innovative training and reasoning framework suitable for lightweight Multimodal Large Language Models (MLLMs)<n>Our self-questioning approach organically guides MLLMs to focus on visual clues relevant to the target problem, reducing hallucinations and enhancing the model's ability to describe fine-grained image details.<n>Our experiments on various benchmarks demonstrate SQ's remarkable capabilities in self-questioning, zero-shot visual reasoning and hallucination mitigation.
  arXiv  Detail & Related papers  (2025-01-06T12:16:56Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.