README.md

---
base_model:
- Qwen/Qwen2.5-VL-72B-Instruct
language:
- en
license: apache-2.0
tags:
- transformers
- multimodal
pipeline_tag: visual-question-answering
---
# VL-Reasoner-72B

**🚀 News:** <u>We release our meticulously curated collection of RL training queries for multimodal reasoning: [ViRL39K](https://huggingface.co/datasets/TIGER-Lab/ViRL39K).</u>

**VL-Reasoner-72B** achieves superior results on various multimodal reasoning benchmarks.

It is trained using the **GRPO-SSR** techniques, serving as the foundation for [**VL-Rethinker**](https://huggingface.co/TIGER-Lab/VL-Rethinker-72B/).

For details of our approach and performance comparison, please see our [paper](https://arxiv.org/abs/2504.08837).

For details of training and evaluation, please see our [code repo](https://github.com/TIGER-AI-Lab/VL-Rethinker/).

Explore further via the following links:

| [**🚀Project Page**](https://tiger-ai-lab.github.io/VL-Rethinker/) | [**📖Paper**](https://arxiv.org/abs/2504.08837) | [**🔗Github**](https://github.com/TIGER-AI-Lab/VL-Rethinker/) | [**🤗Data**](https://huggingface.co/datasets/TIGER-Lab/ViRL39K)  |

## Citation

If you feel this model useful, please give us a free cite:
```
@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}
}	        
```