Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities across diverse tasks, yet they lag significantly behind humans in spatial reasoning. We investigate this gap through Transformation-Driven Visual Reasoning (TVR), a challenging task requiring identification of object transformations across images under varying viewpoints. While traditional Supervised Fine-Tuning (SFT) fails to generate coherent reasoning paths in cross-view settings, sparse-reward Reinforcement Learning (RL) suffers from inefficient exploration and slow convergence. To address these limitations, we propose STAR-R1, a novel framework that integrates a single-stage RL paradigm with a fine-grained reward mechanism tailored for TVR. Specifically, STAR-R1 rewards partial correctness while penalizing excessive enumeration and passive inaction, enabling efficient exploration and precise reasoning. Comprehensive evaluations demonstrate that STAR-R1 achieves state-of-the-art performance across all 11 metrics, outperforming SFT by 23% in cross-view scenarios. Further analysis reveals STAR-R1's anthropomorphic behavior and highlights its unique ability to compare all objects for improving spatial reasoning. Our work provides critical insights in advancing the research of MLLMs and reasoning models. The codes, model weights, and data will be publicly available at https://github.com/zongzhao23/STAR-R1.\n","updatedAt":"2025-05-27T13:02:59.594Z","author":{"_id":"61bb00f6c4ac95d207b25f1b","avatarUrl":"/avatars/3b6eba701d64518d6f694942f5b2e9a9.svg","fullname":"Zongyang Ma","name":"zyma","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"isUserFollowing":false}},"numEdits":2,"identifiedLanguage":{"language":"en","probability":0.8486011624336243},"editors":["zyma"],"editorAvatarUrls":["/avatars/3b6eba701d64518d6f694942f5b2e9a9.svg"],"reactions":[],"isReport":false}},{"id":"6836697b966c664cb545a67a","author":{"_id":"63d3e0e8ff1384ce6c5dd17d","avatarUrl":"https://cdn-avatars.huggingface.co/v1/production/uploads/1674830754237-63d3e0e8ff1384ce6c5dd17d.jpeg","fullname":"Librarian Bot (Bot)","name":"librarian-bot","type":"user","isPro":false,"isHf":false,"isHfAdmin":false,"isMod":false,"followerCount":318,"isUserFollowing":false},"createdAt":"2025-05-28T01:40:11.000Z","type":"comment","data":{"edited":false,"hidden":false,"latest":{"raw":"This is an automated message from the [Librarian Bot](https://huggingface.co/librarian-bots). I found the following papers similar to this paper. \n\nThe following papers were recommended by the Semantic Scholar API \n\n* [GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning](https://huggingface.co/papers/2505.17022) (2025)\n* [UniVG-R1: Reasoning Guided Universal Visual Grounding with Reinforcement Learning](https://huggingface.co/papers/2505.14231) (2025)\n* [DeepEyes: Incentivizing\"Thinking with Images\"via Reinforcement Learning](https://huggingface.co/papers/2505.14362) (2025)\n* [VideoRFT: Incentivizing Video Reasoning Capability in MLLMs via Reinforced Fine-Tuning](https://huggingface.co/papers/2505.12434) (2025)\n* [G1: Bootstrapping Perception and Reasoning Abilities of Vision-Language Model via Reinforcement Learning](https://huggingface.co/papers/2505.13426) (2025)\n* [GMAI-VL-R1: Harnessing Reinforcement Learning for Multimodal Medical Reasoning](https://huggingface.co/papers/2504.01886) (2025)\n* [VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model](https://huggingface.co/papers/2504.07615) (2025)\n\n\n Please give a thumbs up to this comment if you found it helpful!\n\n If you want recommendations for any Paper on Hugging Face checkout [this](https://huggingface.co/spaces/librarian-bots/recommend_similar_papers) Space\n\n You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: `@librarian-bot recommend`","html":"
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
\nThe following papers were recommended by the Semantic Scholar API
\n- \n
- GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning (2025) \n
- UniVG-R1: Reasoning Guided Universal Visual Grounding with Reinforcement Learning (2025) \n
- DeepEyes: Incentivizing\"Thinking with Images\"via Reinforcement Learning (2025) \n
- VideoRFT: Incentivizing Video Reasoning Capability in MLLMs via Reinforced Fine-Tuning (2025) \n
- G1: Bootstrapping Perception and Reasoning Abilities of Vision-Language Model via Reinforcement Learning (2025) \n
- GMAI-VL-R1: Harnessing Reinforcement Learning for Multimodal Medical Reasoning (2025) \n
- VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model (2025) \n
Please give a thumbs up to this comment if you found it helpful!
\nIf you want recommendations for any Paper on Hugging Face checkout this Space
\n You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: \n\n@librarian-bot\n\t recommend
STAR-R1: Spatial TrAnsformation Reasoning by Reinforcing Multimodal LLMs
Abstract
STAR-R1, a novel RL framework with a fine-grained reward mechanism, enhances spatial reasoning in multimodal large language models by addressing limitations in traditional SFT and sparse-reward RL.
Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities across diverse tasks, yet they lag significantly behind humans in spatial reasoning. We investigate this gap through Transformation-Driven Visual Reasoning (TVR), a challenging task requiring identification of object transformations across images under varying viewpoints. While traditional Supervised Fine-Tuning (SFT) fails to generate coherent reasoning paths in cross-view settings, sparse-reward Reinforcement Learning (RL) suffers from inefficient exploration and slow convergence. To address these limitations, we propose STAR-R1, a novel framework that integrates a single-stage RL paradigm with a fine-grained reward mechanism tailored for TVR. Specifically, STAR-R1 rewards partial correctness while penalizing excessive enumeration and passive inaction, enabling efficient exploration and precise reasoning. Comprehensive evaluations demonstrate that STAR-R1 achieves state-of-the-art performance across all 11 metrics, outperforming SFT by 23% in cross-view scenarios. Further analysis reveals STAR-R1's anthropomorphic behavior and highlights its unique ability to compare all objects for improving spatial reasoning. Our work provides critical insights in advancing the research of MLLMs and reasoning models. The codes, model weights, and data will be publicly available at https://github.com/zongzhao23/STAR-R1.
Community
📖Paper
Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities across diverse tasks, yet they lag significantly behind humans in spatial reasoning. We investigate this gap through Transformation-Driven Visual Reasoning (TVR), a challenging task requiring identification of object transformations across images under varying viewpoints. While traditional Supervised Fine-Tuning (SFT) fails to generate coherent reasoning paths in cross-view settings, sparse-reward Reinforcement Learning (RL) suffers from inefficient exploration and slow convergence. To address these limitations, we propose STAR-R1, a novel framework that integrates a single-stage RL paradigm with a fine-grained reward mechanism tailored for TVR. Specifically, STAR-R1 rewards partial correctness while penalizing excessive enumeration and passive inaction, enabling efficient exploration and precise reasoning. Comprehensive evaluations demonstrate that STAR-R1 achieves state-of-the-art performance across all 11 metrics, outperforming SFT by 23% in cross-view scenarios. Further analysis reveals STAR-R1's anthropomorphic behavior and highlights its unique ability to compare all objects for improving spatial reasoning. Our work provides critical insights in advancing the research of MLLMs and reasoning models. The codes, model weights, and data will be publicly available at https://github.com/zongzhao23/STAR-R1.
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
- GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning (2025)
- UniVG-R1: Reasoning Guided Universal Visual Grounding with Reinforcement Learning (2025)
- DeepEyes: Incentivizing"Thinking with Images"via Reinforcement Learning (2025)
- VideoRFT: Incentivizing Video Reasoning Capability in MLLMs via Reinforced Fine-Tuning (2025)
- G1: Bootstrapping Perception and Reasoning Abilities of Vision-Language Model via Reinforcement Learning (2025)
- GMAI-VL-R1: Harnessing Reinforcement Learning for Multimodal Medical Reasoning (2025)
- VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model (2025)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment:
@librarian-bot
recommend
Models citing this paper 0
No model linking this paper
Datasets citing this paper 0
No dataset linking this paper
Spaces citing this paper 0
No Space linking this paper