smsharma/consistency-models

Implementation of Consistency Models (Song et al 2023) for few-step image generation in Jax.

27
/ 100
Experimental

This project helps machine learning practitioners and researchers generate high-quality images from scratch with very few computational steps. It takes in training data, such as image datasets like MNIST or CIFAR-10, and outputs a trained model capable of creating new, unique images efficiently. This is ideal for those exploring new generative model architectures or needing fast image synthesis.

No commits in the last 6 months.

Use this if you are a machine learning researcher or engineer interested in experimenting with state-of-the-art, few-step image generation techniques from scratch.

Not ideal if you need to distill an existing, pre-trained diffusion model or require a continuous-time objective that is stable during training without a pre-trained initialization.

generative-AI image-synthesis machine-learning-research diffusion-models model-training
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 6 / 25
Maturity 16 / 25
Community 5 / 25

How are scores calculated?

Stars

19

Forks

1

Language

Jupyter Notebook

License

MIT

Last pushed

Jun 11, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/smsharma/consistency-models"

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