ModelTC/HarmoniCa

[ICML 2025] This is the official PyTorch implementation of "🎵 HarmoniCa: Harmonizing Training and Inference for Better Feature Caching in Diffusion Transformer Acceleration".

28
/ 100
Experimental

This project helps machine learning engineers and researchers accelerate the process of generating high-quality images using diffusion models. It takes existing diffusion transformer models and applies a specialized training framework to make them run faster. The result is significantly quicker image generation without losing visual quality, making it ideal for tasks requiring rapid visual output.

No commits in the last 6 months.

Use this if you are a machine learning engineer working with diffusion models like DiT or PixArt-alpha and need to speed up both their training and inference times for image generation.

Not ideal if you are looking for a simple, off-the-shelf image generation tool for end-users or if your primary focus is on non-transformer-based diffusion models.

deep-learning-optimization generative-AI diffusion-models AI-model-acceleration image-synthesis
Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 8 / 25
Maturity 15 / 25
Community 3 / 25

How are scores calculated?

Stars

45

Forks

1

Language

Python

License

Apache-2.0

Last pushed

Jul 10, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/ModelTC/HarmoniCa"

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