VTool-R1: VLMs Learn to Think with Images via Reinforcement Learning on Multimodal Tool Use
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866914368863600640 |
|---|---|
| author | Wu, Mingyuan Yang, Jingcheng Jiang, Jize Li, Meitang Yan, Kaizhuo Yu, Hanchao Zhang, Minjia Zhai, Chengxiang Nahrstedt, Klara |
| author_facet | Wu, Mingyuan Yang, Jingcheng Jiang, Jize Li, Meitang Yan, Kaizhuo Yu, Hanchao Zhang, Minjia Zhai, Chengxiang Nahrstedt, Klara |
| contents | Reinforcement Learning Finetuning (RFT) has significantly advanced the reasoning capabilities of large language models (LLMs) by enabling long chains of thought, self-correction, and effective tool use. While recent works attempt to extend RFT to vision-language models (VLMs), these efforts largely produce text-only reasoning conditioned on static image inputs, falling short of true multimodal reasoning in the response. In contrast, test-time methods like Visual Sketchpad incorporate visual steps but lack training mechanisms.
We introduce VTool-R1, the first framework that trains VLMs to generate multimodal chains of thought by interleaving text and intermediate visual reasoning steps. VTool-R1 integrates Python-based visual editing tools into the RFT process, enabling VLMs to learn when and how to generate visual reasoning steps that benefit final reasoning. Trained with outcome-based rewards tied to task accuracy, our approach elicits strategic visual tool use for reasoning without relying on process-based supervision. Experiments on structured visual question answering over charts and tables show that VTool-R1 enhances reasoning performance by teaching VLMs to "think with images" and generate multimodal chain of thoughts with tools. To support future research in multi-turn multi-modal reasoning, we open-source our code at https://github.com/VTOOL-R1/vtool-r1 |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_19255 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | VTool-R1: VLMs Learn to Think with Images via Reinforcement Learning on Multimodal Tool Use Wu, Mingyuan Yang, Jingcheng Jiang, Jize Li, Meitang Yan, Kaizhuo Yu, Hanchao Zhang, Minjia Zhai, Chengxiang Nahrstedt, Klara Machine Learning Artificial Intelligence Reinforcement Learning Finetuning (RFT) has significantly advanced the reasoning capabilities of large language models (LLMs) by enabling long chains of thought, self-correction, and effective tool use. While recent works attempt to extend RFT to vision-language models (VLMs), these efforts largely produce text-only reasoning conditioned on static image inputs, falling short of true multimodal reasoning in the response. In contrast, test-time methods like Visual Sketchpad incorporate visual steps but lack training mechanisms. We introduce VTool-R1, the first framework that trains VLMs to generate multimodal chains of thought by interleaving text and intermediate visual reasoning steps. VTool-R1 integrates Python-based visual editing tools into the RFT process, enabling VLMs to learn when and how to generate visual reasoning steps that benefit final reasoning. Trained with outcome-based rewards tied to task accuracy, our approach elicits strategic visual tool use for reasoning without relying on process-based supervision. Experiments on structured visual question answering over charts and tables show that VTool-R1 enhances reasoning performance by teaching VLMs to "think with images" and generate multimodal chain of thoughts with tools. To support future research in multi-turn multi-modal reasoning, we open-source our code at https://github.com/VTOOL-R1/vtool-r1 |
| title | VTool-R1: VLMs Learn to Think with Images via Reinforcement Learning on Multimodal Tool Use |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2505.19255 |