Skip to content

Commit 511756f

Browse files
jzhang38jzhang38
authored
Update index.md
1 parent 1729bc4 commit 511756f

File tree

1 file changed

+23
-24
lines changed
  • content/blogs/fastvideo_post_training

1 file changed

+23
-24
lines changed

‎content/blogs/fastvideo_post_training/index.md‎

Lines changed: 23 additions & 24 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,5 @@
11
+++
2-
title = "Introducing FastWan: 15 Times Faster Video Generation With Sparse Distillation"
2+
title = "Introducing FastWan: Denoising a 5-Second Video in 1 Second via Sparse Distillation"
33
date = 2025-08-01T11:00:00-08:00
44
authors = ["FastVideo Team"]
55
author = "FastVideo Team"
@@ -22,7 +22,7 @@ draft = false
2222

2323
{{< socialBadges github="hao-ai-lab/FastVideo" arxiv-index="2505.13389" demo="https://fastwan.fastvideo.org/" slack="https://join.slack.com/t/fastvideo/shared_invite/zt-38u6p1jqe-yDI1QJOCEnbtkLoaI5bjZQ" discord="https://discord.gg/Dm8F2peD3e" huggingface="https://huggingface.co/FastVideo" >}}
2424

25-
**TL;DR:** We introduce **FastWan**, a family of video generation models, trained via a new recipe we term as “sparse distillation”, to achieve near-realtime video generation. FastWan matches Wan in video quality but is blazingly faster: 50x speedup on diffusion time and **15x end-to-end speedup**. FastWan2.1-1.3B can generate a 5-second 480P video in **12 seconds** on a single RTX 4090 and **near real time** on a single H200. FastWan2.2-5B-FullAttn can generate a 5-second 720P video in 16 seconds on a single H200. All resources — model weights, training recipe, and dataset — are released under the Apache-2.0 license.
25+
**TL;DR:** We introduce **FastWan**, a family of video generation models trained via a new recipe we term as “sparse distillation”. Powered by FastVideo, FastWan2.1-1.3B denoises a 5-second 480P video in **1 second** on a single H200 and **2.8 second** on a **single RTX 4090**. FastWan2.2-5B denoises a 5-second 720P video in **3 seconds** on a single H200. All resources — model weights, training recipe, and dataset — are released under the Apache-2.0 license.
2626

2727

2828
{{<youtube AvCBPBf2o4M>}}
@@ -34,21 +34,21 @@ With this blog, we are releasing the following models and their recipes:
3434
|:-------------------------------------------------------------------------------------------: |:---------------------------------------------------------------------------------------------------------------: |:--------------------------------------------------------------------------------------------------------: |
3535
| [FastWan2.1-T2V-1.3B](https://huggingface.co/FastVideo/FastWan2.1-T2V-1.3B-Diffusers) | [Recipe](https://github.com/hao-ai-lab/FastVideo/tree/main/examples/distill/Wan2.1-T2V/Wan-Syn-Data-480P) | [FastVideo Synthetic Wan2.1 480P](https://huggingface.co/datasets/FastVideo/Wan-Syn_77x448x832_600k) |
3636
| [FastWan2.1-T2V-14B-Preview](https://huggingface.co/FastVideo/FastWan2.1-T2V-14B-Diffusers) | Coming soon! | [FastVideo Synthetic Wan2.1 720P](https://huggingface.co/datasets/FastVideo/Wan-Syn_77x768x1280_250k) |
37-
| [FFastWan2.2-TI2V-5B-FullAttn-Diffusers](https://huggingface.co/FastVideo/FastWan2.2-TI2V-5B-FullAttn-Diffusers) | [Recipe](https://github.com/hao-ai-lab/FastVideo/tree/main/examples/distill/Wan2.2-TI2V-5B-Diffusers/Data-free) | [FastVideo Synthetic Wan2.2 720P](https://huggingface.co/datasets/FastVideo/Wan2.2-Syn-121x704x1280_32k) |
37+
| [FastWan2.2-TI2V-5B-FullAttn-Diffusers](https://huggingface.co/FastVideo/FastWan2.2-TI2V-5B-FullAttn-Diffusers) | [Recipe](https://github.com/hao-ai-lab/FastVideo/tree/main/examples/distill/Wan2.2-TI2V-5B-Diffusers/Data-free) | [FastVideo Synthetic Wan2.2 720P](https://huggingface.co/datasets/FastVideo/Wan2.2-Syn-121x704x1280_32k) |
3838

3939

40-
For FastWan2.2-TI2V-5B-FullAttn, since its sequence length is short and doesn't benefit much from VSA, we only apply DMD with full attention. We are actively working on applying sparse distillation to 14B models for both Wan2.1 and Wan2.2 and will be releasing those checkpoints over the following weeks. Follow our progress at our [Github](https://github.com/hao-ai-lab/FastVideo), [Slack](https://join.slack.com/t/fastvideo/shared_invite/zt-38u6p1jqe-yDI1QJOCEnbtkLoaI5bjZQ) and [Discord](https://discord.gg/Dm8F2peD3e)!
40+
For FastWan2.2-TI2V-5B-FullAttn, since its sequence length is short (~20K) and doesn't benefit much from VSA, we only apply DMD with full attention. We are actively working on applying sparse distillation to 14B models for both Wan2.1 and Wan2.2 and will be releasing those checkpoints over the following weeks. Follow our progress at our [Github](https://github.com/hao-ai-lab/FastVideo), [Slack](https://join.slack.com/t/fastvideo/shared_invite/zt-38u6p1jqe-yDI1QJOCEnbtkLoaI5bjZQ) and [Discord](https://discord.gg/Dm8F2peD3e)!
4141

4242
### How Fast is FastWan?
4343
{{< image src="img/fastwan.png" alt="denoising speedup" width="100%" >}}
44-
Compared to FA2 alone, we demonstrate how each module accelerates the DiT denoising time.
45-
| | Wan 2.2 5B 720P | Wan2.1 14B 720P | Wan2.1 1.3B 480P | |
46-
|:-------------------------:|:---------------:|:----------------:|:----------------:|---|
47-
| FA2 | 157.21s | 1746.5s | 95.21s | |
48-
| FA2 + DMD | 4.67s | 52s | 2.88s | |
49-
| FA3+DMD | 3.65s | 37.87s | 2.14s | |
50-
| FA3 + DMD + torch compile | 2.64s | 29.5s | 1.49s | |
51-
| VSA + DMD + torch compile | | 13ss | 0.98s | |
44+
We demonstrate how each module accelerates the DiT denoising time (without text encoder and vae) on a single H200 GPU.
45+
| | Wan 2.2 5B 720P | Wan2.1 14B 720P | Wan2.1 1.3B 480P |
46+
|:-------------------------:|:---------------:|:----------------:|:----------------:|
47+
| FA2 | 157.21s | 1746.5s | 95.21s |
48+
| FA2 + DMD | 4.67s | 52s | 2.88s |
49+
| FA3+DMD | 3.65s | 37.87s | 2.14s |
50+
| FA3 + DMD + torch compile | 2.64s | 29.5s | 1.49s |
51+
| VSA + DMD + torch compile | | 13s | 0.98s |
5252

5353
### Online Demo using FastVideo
5454
Try the FastWan demo [here]()!
@@ -126,20 +126,19 @@ We thank [Anyscale](https://www.anyscale.com/), [MBZUAI](https://mbzuai.ac.ae/),
126126
If you use FastWan for your research, please cite our work:
127127
```bibtex
128128
@software{fastvideo2024,
129-
title = {FastVideo: A Unified Framework for Accelerated Video Generation},
130-
author = {The FastVideo Team},
131-
url = {https://github.com/hao-ai-lab/FastVideo},
132-
year = {2024},
129+
title = {FastVideo: A Unified Framework for Accelerated Video Generation},
130+
author = {The FastVideo Team},
131+
url = {https://github.com/hao-ai-lab/FastVideo},
132+
month = apr,
133+
year = {2024},
133134
}
134135
135-
@misc{zhang2025vsafastervideodiffusion,
136-
title={VSA: Faster Video Diffusion with Trainable Sparse Attention},
137-
author={Peiyuan Zhang and Haofeng Huang and Yongqi Chen and Will Lin and Zhengzhong Liu and Ion Stoica and Eric Xing and Hao Zhang},
138-
year={2025},
139-
eprint={2505.13389},
140-
archivePrefix={arXiv},
141-
primaryClass={cs.CV},
142-
url={https://arxiv.org/abs/2505.13389},
136+
@article{zhang2025faster,
137+
title={Faster video diffusion with trainable sparse attention},
138+
author={Zhang, Peiyuan and Huang, Haofeng and Chen, Yongqi and Lin, Will and Liu, Zhengzhong and Stoica, Ion and Xing, Eric P and Zhang, Hao},
139+
journal={arXiv e-prints},
140+
pages={arXiv--2505},
141+
year={2025}
143142
}
144143
```
145144

0 commit comments

Comments
 (0)