README.md

---
license: mit
pipeline_tag: text-to-image
---

# GoT-R1-1B

GoT-R1 is a framework that applies reinforcement learning to enhance semantic-spatial reasoning in visual generation, as presented in [GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning](https://huggingface.co/papers/2505.17022).

## Introduction

Visual generation models often struggle with complex prompts specifying multiple objects with precise spatial relationships. **GoT-R1** addresses this by applying reinforcement learning to enhance semantic-spatial reasoning. Building upon the Generation Chain-of-Thought (GoT) approach, GoT-R1 enables models to autonomously discover effective reasoning strategies through a dual-stage multi-dimensional reward framework.

- **Enhanced Semantic-Spatial Reasoning**: Uses RL to improve planning of complex scenes.
- **Autonomous Reasoning Chain Discovery**: Moves beyond fixed templates to allow the model to explore more effective reasoning paths.
- **Comprehensive MLLM-based Rewards**: Evaluates both the intermediate reasoning process and the final visual output.

## Resources

- **GitHub Repository**: [https://github.com/gogoduan/GoT-R1](https://github.com/gogoduan/GoT-R1)
- **Paper**: [GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning](https://huggingface.co/papers/2505.17022)

## Usage

For installation and setup, please refer to the [official GitHub repository](https://github.com/gogoduan/GoT-R1). To run inference using the provided script from the repository:

```bash
python infer.py --ckpt_path <Your GoT-R1 checkpoint path>
```

## Citation

If you find this work helpful, please consider citing the paper:

```bibtex
@article{duan2025got,
  title={GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning},
  author={Duan, Chengqi and Fang, General and Wang, Yuqing and Wang, Kun and Huang, Linjiang and Zeng, Xingyu and Li, Hongsheng and Liu, Xihui},
  journal={arXiv preprint arXiv:2505.17022},
  year={2025}
}
```