ot-triton-lab/flash-sinkhorn

FlashSinkhorn: IO-Aware Entropic Optimal Transport in PyTorch + Triton. Streaming Sinkhorn with O(nd) memory.

56
/ 100
Established

When comparing two sets of data points, like images or molecular structures, FlashSinkhorn quickly calculates the 'distance' or similarity between them. It takes in your data points and outputs a cost or a plan for how to transform one set into the other, even for very large datasets. This is designed for scientists, engineers, or researchers working with high-dimensional data.

183 stars. Available on PyPI.

Use this if you need to compare large collections of data points efficiently, such as for machine learning tasks, image analysis, or molecular shape comparison, and memory usage is a concern.

Not ideal if your data is small enough that standard methods work fine, or if you don't have access to a CUDA-enabled GPU.

data-comparison optimal-transport point-cloud-analysis machine-learning-research computational-science
Maintenance 10 / 25
Adoption 10 / 25
Maturity 22 / 25
Community 14 / 25

How are scores calculated?

Stars

183

Forks

19

Language

Python

License

MIT

Last pushed

Mar 03, 2026

Commits (30d)

0

Dependencies

3

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/ot-triton-lab/flash-sinkhorn"

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