---
title: WaDi
canonical_url: "https://www.modelscope.cn/models/gudaochangsheng98/WaDi"
md_url: "https://www.modelscope.cn/models/gudaochangsheng98/WaDi.md"
repository: gudaochangsheng98/WaDi
chinese_name: WaDi
last_updated: 2026-05-25
license: "MIT License"
pipeline_tag: text-to-image-synthesis
tasks:
  - text-to-image-synthesis
parameters: 2.3B
tensor_type:
  - F32
  - F16
library_name:
  - safetensors
  - pytorch
frameworks:
  - PyTorch
language:
  - en
downloads: 46
stars: 0
---

# WaDi

> WaDi - gudaochangsheng98 在 ModelScope 开源的模型。一步图像生成

gudaochangsheng98/WaDi 是 ModelScope 魔搭社区上的 2.3B 参数text-to-image-synthesis模型，采用 MIT License 许可。

- **Repository**: gudaochangsheng98/WaDi
- **License**: MIT License
- **Tasks**: text-to-image-synthesis
- **Parameters**: 2.3B
- **Downloads**: 46
- **Stars**: 0
- **Last updated**: 2026-05-25

Source: https://www.modelscope.cn/models/gudaochangsheng98/WaDi

---

# 🚀 [CVPR 2026]WaDi: Weight Direction-aware Distillation for One-step Image Synthesis

[![arXiv](https://img.shields.io/badge/arXiv-WaDi-<COLOR>.svg)](https://arxiv.org/abs/2603.08258) [![arXiv](https://img.shields.io/badge/paper-WaDi-b31b1b.svg)](https://arxiv.org/pdf/2603.08258) ![Visitors](https://visitor-badge.laobi.icu/badge?page_id=gudaochangsheng/WaDi ) [![HF-SD2.1](https://img.shields.io/badge/HF-WaDi--sd2.1-yellow?logo=huggingface)](https://huggingface.co/gudaochangsheng/WaDi/blob/main/rotated_unet-sdv2-1.safetensors) [![HF-SD1.5](https://img.shields.io/badge/HF-WaDi--sd1.5-yellow?logo=huggingface)](https://huggingface.co/gudaochangsheng/WaDi/blob/main/rotated_unet-sdv1-5.safetensors) [![HF-Pixart-Alpha](https://img.shields.io/badge/HF-WaDi--pixart-yellow?logo=huggingface)](https://huggingface.co/gudaochangsheng/WaDi/blob/main/rotated_transformer.safetensors)
[![MS-SD2.1](https://img.shields.io/badge/ModelScope-WaDi--sd2.1-blue)](https://modelscope.cn/models/gudaochangsheng98/WaDi/file/view/master/rotated_unet-sdv2-1.safetensors?status=2)
[![MS-SD1.5](https://img.shields.io/badge/ModelScope-WaDi--sd1.5-blue)](https://modelscope.cn/models/gudaochangsheng98/WaDi/file/view/master/rotated_unet-sdv1-5.safetensors?status=2)
[![MS-Pixart-alpha](https://img.shields.io/badge/ModelScope-WaDi--pixart-blue)](https://modelscope.cn/models/gudaochangsheng98/WaDi/file/view/master/rotated_transformer.safetensors?status=2)
[![Code](https://img.shields.io/badge/Code-WaDi-black?style=flat&logo=github)](https://github.com/gudaochangsheng/WaDi)
[![Project Page](https://img.shields.io/badge/Project-Page-2ea44f?style=flat-square)](https://gudaochangsheng.github.io/WaDi-Page/)

<div align="center">
  <a href="https://gudaochangsheng.github.io/">Lei Wang</a><sup>1</sup>,
  <a href="https://gudaochangsheng.github.io/WaDi-Page/">Yang Cheng</a><sup>1</sup>,
  <a href="https://sen-mao.github.io/">Senmao Li</a><sup>1</sup>,
  <a href="https://github.com/Martinser">Ge Wu</a><sup>1</sup>,
  <a href="https://yaxingwang.github.io/">Yaxing Wang</a><sup>1,3&dagger;</sup>
  <a href="https://scholar.google.com.hk/citations?user=6CIDtZQAAAAJ&hl=en">Jian Yang</a><sup>1,2&dagger;</sup>
</div>

<div align="center">
  <sup>1</sup> PCA Lab, VCIP, College of Computer Science, Nankai University &nbsp;&nbsp;
  <sup>2</sup> PCA Lab, School of Intelligence Science and Technology, Nanjing University &nbsp;&nbsp;
  <sup>3</sup> Shenzhen Futian, NKIARI &nbsp;&nbsp;
</div>

<div align="center">
  &dagger;Corresponding authors
</div>

<div align="center">
<img src="./abstract.png" alt="demo" style="zoom:150%;" />
  <br>
</div>


## 📦 Model Weights

| Model | Hugging Face | ModelScope |
|-------|--------------|------------|
| WaDi-SD2.1 | [![Download HF](https://img.shields.io/badge/HuggingFace-Download-yellow?logo=huggingface)](https://huggingface.co/gudaochangsheng/WaDi/blob/main/rotated_unet-sdv2-1.safetensors) | [![Download MS](https://img.shields.io/badge/ModelScope-Download-blue)](https://modelscope.cn/models/gudaochangsheng98/WaDi/file/view/master/rotated_unet-sdv2-1.safetensors?status=2) |
| WaDi-SD1.5 | [![Download HF](https://img.shields.io/badge/HuggingFace-Download-yellow?logo=huggingface)](https://huggingface.co/gudaochangsheng/WaDi/blob/main/rotated_unet-sdv1-5.safetensors) | [![Download MS](https://img.shields.io/badge/ModelScope-Download-blue)](https://modelscope.cn/models/gudaochangsheng98/WaDi/file/view/master/rotated_unet-sdv1-5.safetensors?status=2) |
| WaDi-PixArt | [![Download HF](https://img.shields.io/badge/HuggingFace-Download-yellow?logo=huggingface)](https://huggingface.co/gudaochangsheng/WaDi/blob/main/rotated_transformer.safetensors) | [![Download MS](https://img.shields.io/badge/ModelScope-Download-blue)](https://modelscope.cn/models/gudaochangsheng98/WaDi/file/view/master/rotated_transformer.safetensors?status=2) |

## 🎬 Inference
```shell
# inference sd model
python infer_sd_model.py
# inference pixart
python infer_pixart.py
  ```

## Training SwiftBrushV2
[An unofficial implementation of SwiftBrushV2](https://github.com/gudaochangsheng/SwiftBrushV2)

## Citation

If you find WaDi useful, please consider giving our repository a star (⭐) and citing our [paper](https://arxiv.org/abs/2603.08258).

```
@InProceedings{Wang_2026_CVPR,
    author    = {Wang, Lei and Cheng, Yang and Li, Senmao and Wu, Ge and Wang, Yaxing and Yang, Jian},
    title     = {WaDi: Weight Direction-aware Distillation for One-step Image Synthesis},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2026},
    pages     = {5574-5584}
}
@inproceedings{li2025one,
      title={One-Way Ticket: Time-Independent Unified Encoder for Distilling Text-to-Image Diffusion Models}, 
      author={Li, Senmao and Wang, Lei and Wang, Kai and Liu, Tao and Xie, Jiehang and van de Weijer, Joost and Khan, Fahad Shahbaz and Yang, Shiqi and Wang, Yaxing and Yang, Jian},
      booktitle={IEEE Conference on Computer Vision and Pattern Recognition (CVPR)}, 
      year={2025},
}
```
## Acknowledgement

This project is based on [Diffusers](https://github.com/huggingface/diffusers). Thanks for their awesome works.
We sincerely acknowledge the excellent and inspiring prior work, [TiUE](https://github.com/sen-mao/Loopfree) and [SwiftBrush](https://github.com/VinAIResearch/SwiftBrush).
## Contact
If you have any questions, please feel free to reach out to me at  `scitop1998@gmail.com`.
