paper-with-me

Papers

SeePhys: Does Seeing Help Thinking? -- Benchmarking Vision-Based Physics Reasoning

2025-05-25 · Kun Xiang, Heng Li, Terry Jingchen Zhang, Yinya Huang, Zirong Liu, Peixin Qu, Jixi He, Jiaqi Chen, Yu-Jie Yuan, Jianhua Han, Hang Xu, Hanhui Li, Mrinmaya Sachan, Xiaodan Liang

We present SeePhys, a large-scale multimodal benchmark for LLM reasoning grounded in physics questions ranging from middle school to PhD qualifying exams. The benchmark covers 7 fundamental domains spanning the physics discipline, incorporating 21 categories of highly heterogeneous diagrams. In contrast to prior works where visual elements mainly serve auxiliary purposes, our benchmark features a substantial proportion of vision-essential problems (75%) that mandate visual information extraction for correct solutions. Through extensive evaluation, we observe that even the most advanced visual reasoning models (e.g., Gemini-2.5-pro and o4-mini) achieve sub-60% accuracy on our benchmark. These results reveal fundamental challenges in current large language models' visual understanding capabilities, particularly in: (i) establishing rigorous coupling between diagram interpretation and physics reasoning, and (ii) overcoming their persistent reliance on textual cues as cognitive shortcuts.

📄 PDF Abstract BibTeX arXiv:2505.19099

Code (1)

seephys/seephys-project 공식 구현 pytorch

Tasks

BenchmarkingVisual Reasoning

Similar Papers 제목 키워드 기반

Bad Seeing or Bad Thinking? Rewarding Perception for Multimodal Reasoning

2026-05-13 · Haozhe Wang, Qixin Xu, Changpeng Wang, Taofeng Xue 외 arxiv

Achieving robust perception-reasoning synergy is a central goal for advanced Vision-Language Models (VLMs). Recent advancements have pursued this goal via architectural designs or agentic workflows. However, these approa…

Reinforcement LearningMultimodal Reasoning

SeePhys Pro: Diagnosing Modality Transfer and Blind-Training Effects in Multimodal RLVR for Physics Reasoning

2026-05-10 · Kun Xiang, Terry Jingchen Zhang, Zirong Liu, Bokai Zhou 외 arxiv

We introduce SeePhys Pro, a fine-grained modality transfer benchmark that studies whether models preserve the same reasoning capability when critical information is progressively transferred from text to image. Unlike st…

Multimodal Reasoning

Multimodal Reasoning for Science: Technical Report and 1st Place Solution to the ICML 2025 SeePhys Challenge

2025-09-07 · Hao Liang, Ruitao Wu, Bohan Zeng, Junbo Niu 외 arxiv

Multimodal reasoning remains a fundamental challenge in artificial intelligence. Despite substantial advances in text-based reasoning, even state-of-the-art models such as GPT-o3 struggle to maintain strong performance i…

Multimodal Reasoning

Search for or Navigate to? Dual Adaptive Thinking for Object Navigation

2022-08-01 · ICCV 2023 1 · Ronghao Dang, Liuyi Wang, Zongtao He, Shuai Su 외

"Search for" or "Navigate to"? When finding an object, the two choices always come up in our subconscious mind. Before seeing the target, we search for the target based on experience. After seeing the target, we remember…

NavigateObject

Does Thinking More always Help? Understanding Test-Time Scaling in Reasoning Models

2025-06-04 · Soumya Suvra Ghosal, Souradip Chakraborty, Avinash Reddy, Yifu Lu 외

Recent trends in test-time scaling for reasoning models (e.g., OpenAI o1, DeepSeek R1) have led to a popular belief that extending thinking traces using prompts like "Wait" or "Let me rethink" can improve performance. Th…