mbreuss/consistency_models_toy_task

Unofficial minimal implementation of consistency models (CM) proposed by Song et al. 2023 on a 1D toy task in pytorch

30
/ 100
Emerging

This project offers a minimal exploration of consistency models using simple 1D datasets, like single or multiple Gaussian distributions. It allows researchers and machine learning engineers to experiment with Consistency Distillation and Consistency Training techniques. You provide the 1D data configurations, and the project outputs visualizations of the model's performance in generating new data points based on those distributions.

No commits in the last 6 months.

Use this if you are a researcher or machine learning engineer looking to understand and test the fundamental behavior of consistency models on simplified, one-dimensional data distributions.

Not ideal if you need to apply consistency models to complex, real-world datasets like high-dimensional images or tabular data, or if you require a production-ready implementation.

generative-modeling machine-learning-research data-distribution-analysis model-training-experimentation
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 6 / 25
Maturity 16 / 25
Community 8 / 25

How are scores calculated?

Stars

21

Forks

2

Language

Python

License

MIT

Last pushed

May 02, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/mbreuss/consistency_models_toy_task"

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