Skip to content
Back to search

cuml

NVIDIA cuML: GPU-Accelerated Machine Learning

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/NVIDIA/cuml with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/NVIDIA/cuml'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/nvidia--cuml/badge.svg)](/listings/nvidia--cuml)

HTML

<a href="/listings/nvidia--cuml"><img src="/api/listings/nvidia--cuml/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 130 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # NVIDIA cuML: GPU-Accelerated Machine Learning
  2. NVIDIA cuML is an open-source CUDA-X Data Science library for
  3. GPU-accelerated machine learning. It provides two ways to run machine learning
  4. workloads on NVIDIA GPUs:
  5. - The `cuml` Python API provides GPU-native estimators with familiar
  6. scikit-learn-style APIs and direct control over machine learning workflows.
  7. - `cuml.accel` accelerates existing scikit-learn, UMAP, and HDBSCAN code without
  8. changing the Python code that uses those libraries.
  9. On representative benchmarks, cuML can accelerate scikit-learn workflows by
  10. up to 50x. Performance depends on the algorithm, dataset, and hardware. See the
  11. [cuML benchmarks](https://docs.nvidia.com/cuml/latest/cuml-accel/benchmarks/) for
  12. results and methodology.
  13. ## Use the GPU-native `cuml` API
  14. The `cuml` Python API follows the familiar scikit-learn fit-predict-transform pattern while keeping data and computation on the GPU. The following example generates sample data and computes DBSCAN clusters on the GPU:
  15. ```python
  16. from cuml.datasets import make_blobs
  17. from cuml.cluster import DBSCAN
  18. # Create sample data
  19. X, y = make_blobs(n_samples=100, centers=3, n_features=2, random_state=42)
  20. # Fit clustering model
  21. dbscan = DBSCAN(eps=1.0, min_samples=5)
  22. dbscan.fit(X)
  23. print(dbscan.labels_)
  24. ```
  25. `cuml` supports clustering, dimensionality reduction, regression,
  26. classification, preprocessing, model selection, model explanation, and
  27. nearest-neighbor workflows. Browse the [API
  28. reference](https://docs.nvidia.com/cuml/latest/api/) for the current list of
  29. estimators and functions.
  30. ## Accelerate existing code with `cuml.accel`
  31. Run an existing Python script through the [`cuml.accel` module](https://docs.nvidia.com/cuml/latest/cuml-accel/):
  32. ```console
  33. python -m cuml.accel script.py
  34. ```
  35. Or load the extension in a Jupyter notebook before importing scikit-learn,
  36. UMAP, or HDBSCAN:
  37. ```python
  38. %load_ext cuml.accel
  39. ```
  40. Supported operations run on the GPU. When an estimator or configuration cannot
  41. be accelerated, `cuml.accel` uses the original CPU implementation so the rest
  42. of the workflow can continue. See the [`cuml.accel` compatibility
  43. documentation](https://docs.nvidia.com/cuml/latest/cuml-accel/compatibility/) for
  44. current coverage and fallback conditions. Use the [logging and profiling
  45. tools](https://docs.nvidia.com/cuml/latest/cuml-accel/logging-and-profiling/) to check
  46. which operations ran on the GPU.
  47. ## Scale beyond one GPU
  48. The `cuml.dask` API provides distributed implementations of selected algorithms
  49. for multi-GPU and multi-node execution with [Dask](https://www.dask.org). See
  50. the [multi-GPU guide](https://docs.nvidia.com/cuml/latest/dask_multigpu_guide/) for
  51. cluster setup, supported algorithms, and examples.
  52. ## Installation
  53. Use the [installation selector](https://docs.nvidia.com/datascience/install#selector) to
  54. generate a command for installing nightly or release cuML packages with conda,
  55. pip, or Docker.
  56. Additional resources:
  57. - [NVIDIA cuML documentation](https://docs.nvidia.com/cuml/)
  58. - [NVIDIA cuML product page](https://developer.nvidia.com/topics/ai/data-science/cuda-x-data-science-libraries/cuml)
  59. - [Walkthrough notebooks](https://github.com/NVIDIA/cuml/tree/main/notebooks)
  60. - [CUDA-X Data Science libraries](https://developer.nvidia.com/topics/ai/data-science/cuda-x-for-data-science)
  61. ## Build and install from source
  62. See the [build guide](BUILD.md).
  63. ## Scikit-learn compatibility
  64. cuML is compatible with scikit-learn version 1.6 or higher.
  65. ## Model serialization and security
  66. cuML models can be serialized with `pickle` or `joblib` and loaded later for
  67. inference. cuML uses cloudpickle so that models trained with `cuml.accel` can be
  68. loaded and used with scikit-learn.
  69. **Only unpickle or deserialize from trusted sources.** The `pickle` module (and
  70. by extension `joblib`) is not secure: malicious payloads can execute arbitrary
  71. code during deserialization and compromise your system. **Do not unpickle or
  72. load data from untrusted or tampered sources.** This applies to `pickle.load()`,
  73. `pickle.loads()`, `joblib.load()`, and any file-based model loading. For
  74. details and patterns, see the [Model Serialization and
  75. Persistence](docs/source/pickling_cuml_models.ipynb) notebook and the [Python
  76. pickle security documentation](https://docs.python.org/3/library/pickle.html).
  77. ## Contributing and support
  78. See the [contributing guide](CONTRIBUTING.md) to contribute to cuML. Report bugs
  79. and request features through [GitHub issues](https://github.com/NVIDIA/cuml/issues).
  80. Join the broader community through the [CUDA-X Data Science libraries
  81. page](https://developer.nvidia.com/topics/ai/data-science/cuda-x-for-data-science#join-the-community).
  82. ## Citation
  83. For additional details on the technologies behind cuML and the broader Python
  84. machine learning landscape, see [_Machine Learning in Python: Main developments
  85. and technology trends in data science, machine learning, and artificial
  86. intelligence_ (2020)](https://arxiv.org/abs/2002.04803) by Sebastian Raschka,
  87. Joshua Patterson, and Corey Nolet.
  88. Please consider citing this work when using cuML in a project:
  89. ```bibtex
  90. @article{raschka2020machine,
  91. title={Machine Learning in Python: Main developments and technology trends in data science, machine learning, and artificial intelligence},
  92. author={Raschka, Sebastian and Patterson, Joshua and Nolet, Corey},
  93. journal={arXiv preprint arXiv:2002.04803},
  94. year={2020}
  95. }
  96. ```

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.