---
title: SSDD
canonical_url: "https://www.modelscope.cn/models/facebook/SSDD"
md_url: "https://www.modelscope.cn/models/facebook/SSDD.md"
repository: facebook/SSDD
last_updated: 2026-03-14
license: cc-by-nc-4.0
parameters: 2.5B
tensor_type:
  - F32
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
downloads: 159
stars: 0
tags:
  - SSDD
  - "arxiv:2510.04961"
---

# SSDD

> SSDD - facebook 在 ModelScope 开源的模型。SSDD: Single-Step Diffusion Decoder for Efficient Image Tokenization

facebook/SSDD 是 ModelScope 魔搭社区上的 2.5B 参数机器学习模型，采用 cc-by-nc-4.0 许可。

- **Repository**: facebook/SSDD
- **License**: cc-by-nc-4.0
- **Parameters**: 2.5B
- **Tags**: SSDD, arxiv:2510.04961
- **Downloads**: 159
- **Stars**: 0
- **Last updated**: 2026-03-14

Source: https://www.modelscope.cn/models/facebook/SSDD

---

# SSDD: Single-Step Diffusion Decoder for Efficient Image Tokenization

Weights for the [SSDD: Single-Step Diffusion Decoder for Efficient Image Tokenization](https://arxiv.org/abs/2510.04961) paper.
You can load and use those weights using [the official implementation on github](https://github.com/facebookresearch/SSDD).

## Citation

If you find our models and paper helpful, please consider citing our work:
```
@article{vallaeys2025ssdd,
    title={SSDD: Single-Step Diffusion Decoder for Efficient Image Tokenization}, 
    author={Théophane Vallaeys and Jakob Verbeek and Matthieu Cord},
    year={2025},
    volume={2510.04961},
    journal={arXiv},
    url={https://arxiv.org/abs/2510.04961},
}
```