VL Reasoner 7B

🚀 VL-Reasoner-7B

VL-Reasoner-7B is a model that achieves excellent results on various multimodal reasoning benchmarks. It is trained using the GRPO-SSR techniques and serves as the foundation for VL-Rethinker.

🚀 Quick Start

🚀 News: We release our meticulously curated collection of RL training queries for multimodal reasoning: ViRL39K.

Explore further via the following links:

| 🚀Project Page | 📖Paper | 🔗Github | 🤗Data |

✨ Features

• Achieves superior results on various multimodal reasoning benchmarks.
• Trained using the GRPO-SSR techniques.
• Serves as the foundation for VL-Rethinker.

📚 Documentation

For details of our approach and performance comparison, please see our paper. For details of training and evaluation, please see our code repo.

📄 License

This project is under the apache-2.0 license.

📦 Model Information

📚 Citation

If you find this model useful, please cite it as follows:

@article{vl-rethinker,
      title={VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning},
      author = {Wang, Haozhe and Qu, Chao and Huang, Zuming and Chu, Wei and Lin, Fangzhen and Chen, Wenhu},
      journal={arXiv preprint arXiv:2504.08837},
      year={2025}
}