| | --- |
| | license: apache-2.0 |
| | library_name: transformers |
| | pipeline_tag: image-text-to-text |
| | base_model: Qwen/Qwen2.5-VL-7B-Instruct |
| | tags: |
| | - multimodal |
| | - agent |
| | - reinforcement-learning |
| | - qwen |
| | --- |
| | |
| | # PyVision-Image-7B-RL |
| |
|
| | [PyVision-RL: Forging Open Agentic Vision Models via RL](https://arxiv.org/abs/2602.20739) |
| |
|
| | This is **PyVision-Image-7B-RL**, a multimodal agentic vision model post-trained from [Qwen2.5-VL-7B-Instruct](https://huggingface.co/Qwen/Qwen2.5-VL-7B-Instruct) using the PyVision-RL reinforcement learning framework. |
| |
|
| | - **Project Page:** [https://agent-x.space/pyvision-rl/](https://agent-x.space/pyvision-rl/) |
| | - **Repository:** [https://github.com/agents-x-project/PyVision-RL](https://github.com/agents-x-project/PyVision-RL) |
| | - **Paper:** [https://arxiv.org/abs/2602.20739](https://arxiv.org/abs/2602.20739) |
| |
|
| | ## Description |
| |
|
| | Reinforcement learning for agentic multimodal models often suffers from "interaction collapse," where models learn to reduce tool usage and multi-turn reasoning. PyVision-RL is a framework designed to stabilize training and sustain interaction using an oversampling-filtering-ranking rollout strategy combined with an accumulative tool reward. |
| |
|
| | PyVision-Image-7B-RL is specifically optimized for image understanding tasks and sustained multi-turn tool interaction, demonstrating strong performance and efficiency for scalable multimodal agents. |
| |
|
| | ## Citation |
| |
|
| | If you find this work useful, please cite the following paper: |
| |
|
| | ```bibtex |
| | @article{pyvisionrl2026, |
| | title={PyVision-RL: Forging Open Agentic Vision Models via RL}, |
| | author={Zhao, Shitian and Lin, Shaoheng and Li, Ming and Zhang, Haoquan and Peng, Wenshuo and Zhang, Kaipeng and Wei, Chen}, |
| | journal={arXiv:2602.20739}, |
| | year={2026} |
| | } |
| | ``` |
