[CVPR2025] Decouple-Then-Merge: Finetune Diffusion Models as Multi-Task Learning

🔥 News

• [2025/02/27] DeMe has been accepted to CVPR 2025! 🤗🤗

📝 TODO

• Training code
• Inference code
• Project page

✒️ Citation

If you find our repo useful for your research, please consider citing our paper:

 @article{ma2024decouple,
 title={Decouple-Then-Merge: Towards Better Training for Diffusion Models},
 author={Ma, Qianli and Ning, Xuefei and Liu, Dongrui and Niu, Li and Zhang, Linfeng},
 journal={arXiv preprint arXiv:2410.06664},
 year={2024}
 }

❤️ Acknowledgement

Our code is built upon diffusers and task vectors. We also refer to the model soups and dnn mode connectivity. Thanks to the contributors for their great work!