horseee/DeepCache
[CVPR 2024] DeepCache: Accelerating Diffusion Models for Free
Exploits U-Net architectural properties to reuse high-level features across diffusion steps while cheaply updating low-level features, achieving 2.3–4.1x speedups with minimal quality loss. Integrates as a plug-and-play wrapper with Hugging Face Diffusers, supporting Stable Diffusion variants (v1.5, XL, inpainting), Stable Video Diffusion, DDPM, and samplers like DDIM and PLMS without requiring model modifications. Configurable via `cache_interval` and `cache_branch_id` parameters to balance acceleration against output fidelity.
963 stars and 54,901 monthly downloads. Used by 3 other packages. No commits in the last 6 months. Available on PyPI.
Stars
963
Forks
51
Language
Python
License
Apache-2.0
Category
Last pushed
Jun 27, 2024
Monthly downloads
54,901
Commits (30d)
0
Dependencies
3
Reverse dependents
3
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/horseee/DeepCache"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Related models
FlorianFuerrutter/genQC
Generative Quantum Circuits
Gen-Verse/MMaDA
MMaDA - Open-Sourced Multimodal Large Diffusion Language Models (dLLMs with block diffusion,...
kuleshov-group/mdlm
[NeurIPS 2024] Simple and Effective Masked Diffusion Language Model
Shark-NLP/DiffuSeq
[ICLR'23] DiffuSeq: Sequence to Sequence Text Generation with Diffusion Models
ali-vilab/TeaCache
Timestep Embedding Tells: It's Time to Cache for Video Diffusion Model