| license: mit | |
| pipeline_tag: text-to-image | |
| # GoT-R1-7B | |
| GoT-R1-7B is a multimodal large language model (MLLM) designed for high-quality text-to-image generation with advanced semantic-spatial reasoning, as introduced in the paper [GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning](https://huggingface.co/papers/2505.17022). | |
| - **Paper:** [GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning](https://huggingface.co/papers/2505.17022) | |
| - **Repository:** [https://github.com/gogoduan/GoT-R1](https://github.com/gogoduan/GoT-R1) | |
| ## Overview | |
| Visual generation models often struggle with complex prompts that specify multiple objects with precise spatial relationships and attributes. 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. The model uses a unified MLLM architecture (based on Janus-Pro) that autoregressively generates a textual reasoning chain followed by image tokens. | |
| ## Usage | |
| To use this model, please follow the installation instructions in the [official GitHub repository](https://github.com/gogoduan/GoT-R1). Inference can be performed using the provided script: | |
| ```bash | |
| python infer.py --ckpt_path <path-to-GoT-R1-7B-weights> | |
| ``` | |
| ## Citation | |
| ```bibtex | |
| @article{duan2025got, | |
| title={GoT-R1: Unleashing Reasoning Capability of MLLM for Visual Generation with Reinforcement Learning}, | |
| author={Duan, Chengqi and Fang, Rongyao 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} | |
| } | |
| ``` |