AniAggarwal/ecad

[ICLR 2026] Code for Evolutionary Caching to Accelerate Your Off-the-Shelf Diffusion Model

33
/ 100
Emerging

This project helps AI researchers and practitioners accelerate image generation using existing diffusion models. It takes an off-the-shelf diffusion model and, through an evolutionary algorithm, discovers optimal caching schedules to significantly speed up the image creation process. The output is a set of optimized caching patterns that make your diffusion model run faster without needing to retrain it, which is ideal for anyone working with generative AI for visual content.

Use this if you are a researcher or practitioner using diffusion models (like PixArt-α or FLUX) and want to generate high-quality images much faster without changing the model's architecture or retraining it.

Not ideal if you are looking to build a diffusion model from scratch, modify its core architecture, or if you don't use diffusion models for image generation.

generative-ai image-synthesis diffusion-models ai-optimization deep-learning-inference
No License No Package No Dependents
Maintenance 10 / 25
Adoption 7 / 25
Maturity 7 / 25
Community 9 / 25

How are scores calculated?

Stars

31

Forks

3

Language

Python

License

Last pushed

Mar 01, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/AniAggarwal/ecad"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.