library_name: transformers
license: apache-2.0
datasets:
- Franklin0/ReasonGen-R1-RL-Geneval-12k
- Franklin0/ReasonGen-R1-RL-DPG-5k
- Franklin0/ReasonGen-R1-RL-T2I-11k
base_model:
- deepseek-ai/Janus-Pro-7B
Model Card for Model ID
An autoregressive image generation with text-based chain-of-thought. Official checkpoint for the paper "ReasonGen-R1: Cot for Autoregressive Image generation models through SFT and RL".
Website: https://aka.ms/reasongen
Code: https://github.com/Franklin-Zhang0/Image-RL
🚀 ReasonGen-R1: Cot for Autoregressive Image generation models through SFT and RL
📥 Model Download |
⚡ Quick Start |
📜 License |
📖 Citation
📄 Paper Link
1. Introduction
Although chain-of-thought (CoT) reasoning and reinforcement learning (RL) have driven breakthroughs in NLP, their integration into generative vision models remains underexplored. We introduce ReasonGen-R1, a two-stage framework that first imbues an autoregressive image generator with explicit text-based “thinking” skills via supervised fine-tuning (SFT) on a newly generated reasoning dataset of written rationales, and then refines its outputs using Generation-Reward Proximal Optimization (GRPO). Text-based CoT reasoning dataset for image synthesis. We automatically generate and release a corpus of step-by-step, model-crafted rationales paired with visual prompts, enabling controlled planning of object layouts, styles, and scene compositions. RL refinement with GRPO. Our GRPO algorithm uses reward signals from a pretrained vision–language model to assess overall visual quality, optimizing the policy in each update. Evaluations on Geneval, DPG, and the T2I benchmark demonstrate that ReasonGen-R1 consistently outperforms strong baselines and prior state-of-the-art models. We will open-source our generated reasoning dataset and training code to accelerate further advances in text-based reasoning–driven image generation.
2. Model Download
We release ReasonGen-R1 to the public to support a broader and more diverse range of research within both academic and commercial communities. Please note that the use of this model is subject to the terms outlined in License section. Commercial usage is permitted under these terms.
Huggingface
| Model | Download |
|---|---|
| ReasonGen-R1 | 🤗 Hugging Face |
| ReasonGen-R1-SFT-Only | 🤗 Hugging Face |
3. Quick Start
Installation
You can install the necessary dependencies by running the following command:
cd ~
mkdir project
cd project
conda create -n image_rl python==3.12 -y
conda activate image_rl
pip3 install torch==2.6.0 torchvision --index-url https://download.pytorch.org/whl/cu124
pip3 install flash-attn --no-build-isolation
git clone https://github.com/Franklin-Zhang0/Image-RL.git
cd Image-RL
pip install -r requirements.txt
pip install -e .
pip install -e ./Janus
Evaluation Environment Installation (Optional)
If you want to run the evaluation code, you can install the evaluation environment by running the following commands:
# Geneval
cd ~
mkdir project
cd project
git clone https://github.com/djghosh13/geneval.git
cd geneval
conda deactivate
conda create -n geneval python=3.9 -y
conda activate geneval
pip install torch==1.13.1 torchvision==0.14.1 torchaudio==0.13.1
pip install mmcv-full==1.7.0 -f https://download.openmmlab.com/mmcv/dist/cu117/torch1.13/index.html
pip install mmengine==0.7.3
pip install pandas
pip install numpy==1.23.1
pip install open-clip-torch
pip install clip-benchmark
git clone https://github.com/open-mmlab/mmdetection.git
cd mmdetection; git checkout 2.x
pip install -v -e .
cd ../
bash ./evaluation/download_models.sh "./models"
# DPG
cd ~
cd project
git clone https://github.com/TencentQQGYLab/ELLA.git
cd ELLA
cp ~/project/ReasonGen-R1/requirements-for-dpg_bench.txt .
conda deactivate
conda create -n dpg_test python=3.9 -y
conda activate dpg_test
conda install conda-forge::fairseq -y
pip install -r requirements-for-dpg_bench.txt
Once the eval environment is setup, you can use the following commands to run the evaluation:
bash -i benchmark/geneval.sh
bash -i benchmark/dpg_eval.sh
Inference
To inference with the ReasonGen-R1 model, you can use the following command:
python Image-RL/Janus/cot_generate_inference.py
SFT Training
To train the SFT model from Janus-Pro-7B model on the ReasonGen-R1-SFT-200k dataset, you can use the following command:
bash Image-RL/examples/janus_sft.sh
RL Training
To train the RL model from the ReasonGen-R1-SFT model, you can use the following command:
bash Image-RL/Janus/janus_rl.py
5. Acknowledgements
We would like to thank Verl, upon which our repo is built.
4. Citation
@article{yu2025reasongen,
title={ReasonGen-R1: Cot for Autoregressive Image generation models through SFT and RL},
author={Yu Zhang, Yunqi Li, Yifan Yang, Rui Wang， Yuqin Yang, Qi Dai, Jianming Bao, Dongdong Chen, Chong Luo, Lili Qiu},
year={2025}
}