LightCache: Memory-Efficient, Training-Free Acceleration for Video Generation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866916993102249984 |
|---|---|
| author | Xiao, Yang Li, Gen Deng, Kaiyuan Wu, Yushu Zhan, Zheng Wang, Yanzhi Ma, Xiaolong Hui, Bo |
| author_facet | Xiao, Yang Li, Gen Deng, Kaiyuan Wu, Yushu Zhan, Zheng Wang, Yanzhi Ma, Xiaolong Hui, Bo |
| contents | Training-free acceleration has emerged as an advanced research area in video generation based on diffusion models. The redundancy of latents in diffusion model inference provides a natural entry point for acceleration. In this paper, we decompose the inference process into the encoding, denoising, and decoding stages, and observe that cache-based acceleration methods often lead to substantial memory surges in the latter two stages. To address this problem, we analyze the characteristics of inference across different stages and propose stage-specific strategies for reducing memory consumption: 1) Asynchronous Cache Swapping. 2) Feature chunk. 3) Slicing latents to decode. At the same time, we ensure that the time overhead introduced by these three strategies remains lower than the acceleration gains themselves. Compared with the baseline, our approach achieves faster inference speed and lower memory usage, while maintaining quality degradation within an acceptable range. The Code is available at https://github.com/NKUShaw/LightCache . |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_05367 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | LightCache: Memory-Efficient, Training-Free Acceleration for Video Generation Xiao, Yang Li, Gen Deng, Kaiyuan Wu, Yushu Zhan, Zheng Wang, Yanzhi Ma, Xiaolong Hui, Bo Computer Vision and Pattern Recognition Machine Learning Training-free acceleration has emerged as an advanced research area in video generation based on diffusion models. The redundancy of latents in diffusion model inference provides a natural entry point for acceleration. In this paper, we decompose the inference process into the encoding, denoising, and decoding stages, and observe that cache-based acceleration methods often lead to substantial memory surges in the latter two stages. To address this problem, we analyze the characteristics of inference across different stages and propose stage-specific strategies for reducing memory consumption: 1) Asynchronous Cache Swapping. 2) Feature chunk. 3) Slicing latents to decode. At the same time, we ensure that the time overhead introduced by these three strategies remains lower than the acceleration gains themselves. Compared with the baseline, our approach achieves faster inference speed and lower memory usage, while maintaining quality degradation within an acceptable range. The Code is available at https://github.com/NKUShaw/LightCache . |
| title | LightCache: Memory-Efficient, Training-Free Acceleration for Video Generation |
| topic | Computer Vision and Pattern Recognition Machine Learning |
| url | https://arxiv.org/abs/2510.05367 |