xinyun/TeaCache
1
0
Fork 0
mirror of https://git.datalinker.icu/ali-vilab/TeaCache synced 2025-12-08 20:34:24 +08:00
Code Issues Packages Projects Releases Wiki Activity

Update README.md

Browse Source
This commit is contained in:
Feng Liu 2025-01-20 15:14:01 +08:00 committed by GitHub
parent 611cc61f51
commit 9fb5047786
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 8 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
README.md
View File

@ -59,10 +59,10 @@
![visualization](./assets/tisser.png) ![visualization](./assets/tisser.png)
## Introduction 🫖 ## 🫖 Introduction
We introduce Timestep Embedding Aware Cache (TeaCache), a training-free caching approach that estimates and leverages the fluctuating differences among model outputs across timesteps, thereby accelerating the inference. TeaCache works well for Video Diffusion Models, Image Diffusion models and Audio Diffusion Models. For more details and results, please visit our [project page](https://github.com/LiewFeng/TeaCache). We introduce Timestep Embedding Aware Cache (TeaCache), a training-free caching approach that estimates and leverages the fluctuating differences among model outputs across timesteps, thereby accelerating the inference. TeaCache works well for Video Diffusion Models, Image Diffusion models and Audio Diffusion Models. For more details and results, please visit our [project page](https://github.com/LiewFeng/TeaCache).
## Latest News 🔥 ## 🔥 Latest News
- **If you like our project, please give us a star ⭐ on GitHub for the latest update.** - **If you like our project, please give us a star ⭐ on GitHub for the latest update.**
- [2025/01/20] 🔥 Support [CogVideoX1.5-5B](https://github.com/THUDM/CogVideo) for both T2V and I2V. Thanks [@zishen-ucap](https://github.com/zishen-ucap). - [2025/01/20] 🔥 Support [CogVideoX1.5-5B](https://github.com/THUDM/CogVideo) for both T2V and I2V. Thanks [@zishen-ucap](https://github.com/zishen-ucap).
- [2025/01/07] 🔥 Support [TangoFlux](https://github.com/declare-lab/TangoFlux). TeaCache works well for Audio Diffusion Models! - [2025/01/07] 🔥 Support [TangoFlux](https://github.com/declare-lab/TangoFlux). TeaCache works well for Audio Diffusion Models!
@ -74,7 +74,7 @@ We introduce Timestep Embedding Aware Cache (TeaCache), a training-free caching
- [2024/12/06] 🎉 Release the [code](https://github.com/LiewFeng/TeaCache) of TeaCache. Support [Open-Sora](https://github.com/hpcaitech/Open-Sora), [Open-Sora-Plan](https://github.com/PKU-YuanGroup/Open-Sora-Plan) and [Latte](https://github.com/Vchitect/Latte). - [2024/12/06] 🎉 Release the [code](https://github.com/LiewFeng/TeaCache) of TeaCache. Support [Open-Sora](https://github.com/hpcaitech/Open-Sora), [Open-Sora-Plan](https://github.com/PKU-YuanGroup/Open-Sora-Plan) and [Latte](https://github.com/Vchitect/Latte).
- [2024/11/28] 🎉 Release the [paper](https://arxiv.org/abs/2411.19108) of TeaCache. - [2024/11/28] 🎉 Release the [paper](https://arxiv.org/abs/2411.19108) of TeaCache.
## Community Contributions 🧩 ## 🧩 Community Contributions
If you develop/use TeaCache in your projects, welcome to let us know. If you develop/use TeaCache in your projects, welcome to let us know.
- [ConsisID](https://github.com/PKU-YuanGroup/ConsisID) supports TeaCache. Thanks [@SHYuanBest](https://github.com/SHYuanBest). - [ConsisID](https://github.com/PKU-YuanGroup/ConsisID) supports TeaCache. Thanks [@SHYuanBest](https://github.com/SHYuanBest).
- [ComfyUI-HunyuanVideoWrapper](https://github.com/kijai/ComfyUI-HunyuanVideoWrapper) supports TeaCache4HunyuanVideo. Thanks [@kijai](https://github.com/kijai), [ctf05](https://github.com/ctf05) and [DarioFT](https://github.com/DarioFT). - [ComfyUI-HunyuanVideoWrapper](https://github.com/kijai/ComfyUI-HunyuanVideoWrapper) supports TeaCache4HunyuanVideo. Thanks [@kijai](https://github.com/kijai), [ctf05](https://github.com/ctf05) and [DarioFT](https://github.com/DarioFT).
@ -86,7 +86,7 @@ If you develop/use TeaCache in your projects, welcome to let us know.
- [ComfyUI-TangoFlux](https://github.com/LucipherDev/ComfyUI-TangoFlux) supports TeaCache. Thanks [@LucipherDev](https://github.com/LucipherDev). - [ComfyUI-TangoFlux](https://github.com/LucipherDev/ComfyUI-TangoFlux) supports TeaCache. Thanks [@LucipherDev](https://github.com/LucipherDev).
## Supported Models 🎉 ## 🎉 Supported Models
**Text to Video** **Text to Video**
- [TeaCache4Open-Sora](./eval/teacache/README.md) - [TeaCache4Open-Sora](./eval/teacache/README.md)
- [TeaCache4Open-Sora-Plan](./eval/teacache/README.md) - [TeaCache4Open-Sora-Plan](./eval/teacache/README.md)
@ -109,22 +109,22 @@ If you develop/use TeaCache in your projects, welcome to let us know.
**Text to Audio** **Text to Audio**
- [TeaCache4TangoFlux](./TeaCache4TangoFlux/README.md) - [TeaCache4TangoFlux](./TeaCache4TangoFlux/README.md)
## Instructions for Supporting Other Models 🤖 ## 🤖 Instructions for Supporting Other Models
- **Welcome for PRs to support other models.** - **Welcome for PRs to support other models.**
- If the custom model is based on or has similar model structure to the models we've supported, you can try to directly transfer TeaCache to the custom model. For example, rescaling coefficients for CogVideoX-5B can be directly applied to CogVideoX1.5, ConsisID and rescaling coefficients for FLUX can be directly applied to TangoFlux. - If the custom model is based on or has similar model structure to the models we've supported, you can try to directly transfer TeaCache to the custom model. For example, rescaling coefficients for CogVideoX-5B can be directly applied to CogVideoX1.5, ConsisID and rescaling coefficients for FLUX can be directly applied to TangoFlux.
- Otherwise, you can refer to these successful attempts, e.g., [1](https://github.com/ali-vilab/TeaCache/issues/20), [2](https://github.com/ali-vilab/TeaCache/issues/18). - Otherwise, you can refer to these successful attempts, e.g., [1](https://github.com/ali-vilab/TeaCache/issues/20), [2](https://github.com/ali-vilab/TeaCache/issues/18).
## Acknowledgement 💐 ## 💐 Acknowledgement
This repository is built based on [VideoSys](https://github.com/NUS-HPC-AI-Lab/VideoSys), [Diffusers](https://github.com/huggingface/diffusers), [Open-Sora](https://github.com/hpcaitech/Open-Sora), [Open-Sora-Plan](https://github.com/PKU-YuanGroup/Open-Sora-Plan), [Latte](https://github.com/Vchitect/Latte), [CogVideoX](https://github.com/THUDM/CogVideo), [HunyuanVideo](https://github.com/Tencent/HunyuanVideo), [ConsisID](https://github.com/PKU-YuanGroup/ConsisID), [FLUX](https://github.com/black-forest-labs/flux), [Mochi](https://github.com/genmoai/mochi), [LTX-Video](https://github.com/Lightricks/LTX-Video), [Lumina-T2X](https://github.com/Alpha-VLLM/Lumina-T2X) and [TangoFlux](https://github.com/declare-lab/TangoFlux). Thanks for their contributions! This repository is built based on [VideoSys](https://github.com/NUS-HPC-AI-Lab/VideoSys), [Diffusers](https://github.com/huggingface/diffusers), [Open-Sora](https://github.com/hpcaitech/Open-Sora), [Open-Sora-Plan](https://github.com/PKU-YuanGroup/Open-Sora-Plan), [Latte](https://github.com/Vchitect/Latte), [CogVideoX](https://github.com/THUDM/CogVideo), [HunyuanVideo](https://github.com/Tencent/HunyuanVideo), [ConsisID](https://github.com/PKU-YuanGroup/ConsisID), [FLUX](https://github.com/black-forest-labs/flux), [Mochi](https://github.com/genmoai/mochi), [LTX-Video](https://github.com/Lightricks/LTX-Video), [Lumina-T2X](https://github.com/Alpha-VLLM/Lumina-T2X) and [TangoFlux](https://github.com/declare-lab/TangoFlux). Thanks for their contributions!
## License 🔒 ## 🔒 License
* The majority of this project is released under the Apache 2.0 license as found in the [LICENSE](./LICENSE) file. * The majority of this project is released under the Apache 2.0 license as found in the [LICENSE](./LICENSE) file.
* For [VideoSys](https://github.com/NUS-HPC-AI-Lab/VideoSys), [Diffusers](https://github.com/huggingface/diffusers), [Open-Sora](https://github.com/hpcaitech/Open-Sora), [Open-Sora-Plan](https://github.com/PKU-YuanGroup/Open-Sora-Plan), [Latte](https://github.com/Vchitect/Latte), [CogVideoX](https://github.com/THUDM/CogVideo), [HunyuanVideo](https://github.com/Tencent/HunyuanVideo), [ConsisID](https://github.com/PKU-YuanGroup/ConsisID), [FLUX](https://github.com/black-forest-labs/flux), [Mochi](https://github.com/genmoai/mochi), [LTX-Video](https://github.com/Lightricks/LTX-Video), [Lumina-T2X](https://github.com/Alpha-VLLM/Lumina-T2X), and [TangoFlux](https://github.com/declare-lab/TangoFlux), please follow their LICENSE. * For [VideoSys](https://github.com/NUS-HPC-AI-Lab/VideoSys), [Diffusers](https://github.com/huggingface/diffusers), [Open-Sora](https://github.com/hpcaitech/Open-Sora), [Open-Sora-Plan](https://github.com/PKU-YuanGroup/Open-Sora-Plan), [Latte](https://github.com/Vchitect/Latte), [CogVideoX](https://github.com/THUDM/CogVideo), [HunyuanVideo](https://github.com/Tencent/HunyuanVideo), [ConsisID](https://github.com/PKU-YuanGroup/ConsisID), [FLUX](https://github.com/black-forest-labs/flux), [Mochi](https://github.com/genmoai/mochi), [LTX-Video](https://github.com/Lightricks/LTX-Video), [Lumina-T2X](https://github.com/Alpha-VLLM/Lumina-T2X), and [TangoFlux](https://github.com/declare-lab/TangoFlux), please follow their LICENSE.
* The service is a research preview. Please contact us if you find any potential violations. (liufeng20@mails.ucas.ac.cn) * The service is a research preview. Please contact us if you find any potential violations. (liufeng20@mails.ucas.ac.cn)
## Citation 📖 ## 📖 Citation
If you find TeaCache is useful in your research or applications, please consider giving us a star ⭐ and citing it by the following BibTeX entry. If you find TeaCache is useful in your research or applications, please consider giving us a star ⭐ and citing it by the following BibTeX entry.
``` ```