#7373rd of 105 in ML frameworks and libraries
umap
Python UMAP dimension reduction with a scikit-learn API
- Stars
- 8.3k
- License
- BSD-3-Clause
- Last commit
- Oct 2026
- Last release
- Apr 2026
- Language
- Python
8,094 on 1 Apr 20268,282 on 21 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
umap-learn implements Uniform Manifold Approximation and Projection as a scikit-learn-style transformer, usable for 2D visualisation or general non-linear dimension reduction. It supports many distance metrics, sparse input, supervised fitting, transform on new points, densMAP, and Parametric UMAP via TensorFlow. Numba provides the speed, and pynndescent handles nearest neighbours.
Who it is for: Data scientists needing non-linear dimension reduction or visualisation
Strengths
- Drop-in sklearn fit/transform API, works inside pipelines
- Embeds 70000x784 MNIST in about 42 seconds on a 3.1 GHz i7, per README
- Many metrics incl. cosine and correlation; custom numba-JITed functions; sparse input
- Supervised and semi-supervised embedding by passing labels as y
Weaknesses
- Runs on CPU only; GPU use requires the separate torchdr package
- Numba JIT warmup adds delay on first run
- Parametric UMAP needs TensorFlow installed separately
- Plotting needs extra dependencies (matplotlib, datashader, holoviews)
What it needs
- no GPU
- Needs numpy, scipy, scikit-learn, numba, tqdm, pynndescent
Also in ML frameworks and libraries
See all 105| Rank | Project | Score |
|---|---|---|
| 1 | transformersPython library defining open models for text, vision, audio and multimodal tasks | 94 out of 100 |
| 2 | pytorchPython tensor library with GPU support and tape-based autograd for deep learning | 94 out of 100 |
| 3 | ultralyticsPython library and CLI for training and running YOLO vision models | 93 out of 100 |
| #4 | pytorch-image-modelsPyTorch library of image models, pretrained weights and reference training scripts | 92 out of 100 |
| #5 | supervisionPython toolkit for loading, annotating and counting computer vision detections | 91 out of 100 |