Is there any plan to support TeaCache for training-free acceleration?
Open
Feature
- Dominant language
- Python
- Stars
- 133k
- Forks
- 15.7k
- Avg merge
- 1d 7h
- Merged PRs (30d)
- 158
Description
### Feature Idea
TeaCache is a training-free inference acceleration method for visual generation. TeaCache can speedup HunyuanVideo 2x without much visual quality degradation. For example, the inference for a 720p, 129-frame video takes around 50 minutes on a single A800 GPU while TeaCache can sppeedup to 23 minutes. Teacache4FLUX will also be released in a few days. Thanks for your efforts!
https://github.com/LiewFeng/TeaCache.
### Existing Solutions
_No response_
### Other
_No response_
Contributor guide
Assessment
This issue has not been assessed yet.