#102102nd of 168 in ML frameworks and libraries
cuml
GPU-accelerated scikit-learn-style machine learning library
- Stars
- 5.3k
- License
- Apache-2.0
- Last commit
- Nov 2026
- Last release
- Oct 2026
- Language
- Python
5,147 on 21 Apr 20265,320 on 10 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
cuML provides GPU-native estimators with scikit-learn-style fit/predict/transform APIs, covering clustering, dimensionality reduction, regression, classification, preprocessing and nearest neighbors. Its cuml.accel module runs existing scikit-learn, UMAP and HDBSCAN scripts on the GPU without code changes, falling back to CPU for unsupported estimators. cuml.dask adds multi-GPU and multi-node execution for selected algorithms.
Who it is for: Data scientists speeding up scikit-learn workloads on NVIDIA GPUs
Strengths
- cuml.accel runs unmodified scikit-learn, UMAP and HDBSCAN code via `python -m cuml.accel`
- Falls back to the CPU implementation when an estimator cannot be accelerated
- cuml.dask gives multi-GPU and multi-node execution for selected algorithms
- Logging and profiling tools show which operations ran on the GPU
Weaknesses
- Targets NVIDIA GPUs only; no AMD or other accelerator support mentioned
- Speedups vary by algorithm and dataset; the 50x figure is an upper bound
- Distributed support covers only selected algorithms, not the full API
- Models are saved with pickle or joblib, which is unsafe for untrusted files
What it needs
- GPU required
- Needs CUDA, scikit-learn >=1.6, Dask (for multi-GPU)
- Models: scikit-learn-style estimators, UMAP, HDBSCAN
Quick answers
Is cuml open source?
Yes. Its code is on GitHub under the Apache-2.0 license.
Does cuml need a GPU?
Yes, it needs a GPU.
Also in ML frameworks and libraries
See all 168| 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 | 93 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 training scripts | 93 out of 100 |
| #5 | supervisionPython toolkit for loading, annotating and counting computer vision detections | 92 out of 100 |