Generated banner for Cuml
Generated logo for Cuml

Cuml

NVIDIAOfficialPublished by NVIDIA

  • Skill
  • Tool integration
  • Data and AI
  • Apache-2.0
  • Medium risk
  • Has scripts

From the author

Use when writing, running, debugging or upgrading Python code that uses NVIDIA cuML (GPU machine learning), cuml.accel (zero-code-change GPU acceleration of scikit-learn, umap-learn and hdbscan), or GPU UMAP/HDBSCAN/RandomForest/KMeans/SVC, including through libraries that wrap cuML (BERTopic, rapids-singlecell). Covers APIs removed or renamed in 2025-2026 releases, verifying code really ran on the GPU, large-data UMAP/HDBSCAN memory recipes, and exporting trained models. Not for cuDF-only dataframe work, PyTorch or XGBoost.
— the description by NVIDIA

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

This skill provides guidance for working with NVIDIA cuML, including the cuml.accel module for GPU acceleration of scikit-learn, umap-learn, and hdbscan. It covers API changes in recent releases, memory management for large datasets, model exporting, and verifying GPU execution. It is designed to help developers navigate frequent library updates and ensure code compatibility with the installed cuML environment.

What it does

  • Identifies removed or renamed APIs in recent cuML releases.
  • Provides instructions for enabling and profiling GPU acceleration via cuml.accel.
  • Offers memory management strategies for large-scale UMAP, HDBSCAN, and KMeans tasks.
  • Explains how to export models between cuML and scikit-learn or other inference formats.
  • Guides users on verifying whether code is executing on the GPU or falling back to the CPU.
  • Details how to handle data type conversions and output formats when using cuDF or NumPy.

When to use it

  • When writing or upgrading Python code that utilizes NVIDIA cuML or cuml.accel.
  • When troubleshooting code that is not running on the GPU as expected.
  • When migrating existing scikit-learn workflows to GPU-accelerated versions.
  • When encountering memory issues with large-scale clustering or manifold learning tasks.
  • When needing to export trained cuML models for use in CPU-only environments.

Advantages

  • Helps maintain compatibility with rapidly evolving cuML API versions.
  • Provides clear methods to verify GPU execution and identify fallback reasons.
  • Offers specific recipes for managing GPU memory during intensive operations.
  • Clarifies differences in numerical results between cuML and CPU-based libraries.

Usage examples

“How do I check if my scikit-learn code is actually running on the GPU with cuml.accel?”
The agent will explain how to use the -v flag or --profile option with python -m cuml.accel to log execution details and identify any CPU fallbacks.
“My UMAP training is running out of memory. How can I optimize it?”
The agent will suggest using build_algo='nn_descent' with specific knn_n_clusters settings to batch the kNN build and reduce memory usage.
“How do I convert a trained cuML RandomForest model for use on a CPU-only machine?”
The agent will provide the est.as_sklearn() method to export the model as a standard scikit-learn object.
“I am getting a ModuleNotFoundError for a cuML module. What should I do?”
The agent will guide you to check your installed cuML version and use the provided script to identify if the API has been removed or renamed in your version.

Install

  1. Download the skill as a ZIP:

    https://autogineer.com/agent-skills/nvidia/cuml/download
  2. Unzip it into the cross-client skills folder of your project:

    mkdir -p .agents/skills && unzip cuml.zip -d .agents/skills

Documentation by NVIDIA

cuML

cuML releases every two months and removes deprecated APIs, so your training data is likely to contain APIs that no longer exist. Look things up in the installed cuML instead of recalling them.

Ground facts

  • Install from pypi.org: pip install cuml-cu13 (CUDA 13 driver, >= 580) or cuml-cu12. No extra index URL is needed. pip install cuml installs an unrelated placeholder.

  • Check the installation guide for supported GPUs, operating systems, Python versions, CUDA/driver requirements, and install commands. Linux only; on Windows use WSL2.

  • Google Colab GPU runtimes ship cuML preinstalled: check pip list before installing anything there.

  • cuml.accel runs unmodified scikit-learn / umap-learn / hdbscan code on the GPU and falls back to the CPU for anything unsupported. It accelerates RandomForest, LogisticRegression, Ridge, KMeans, DBSCAN, PCA, SVC, KNN, UMAP, HDBSCAN, scalers, encoders and more.

  • Enable it with python -m cuml.accel script.py, %load_ext cuml.accel (first notebook cell), CUML_ACCEL_ENABLED=1, or cuml.accel.install() before importing sklearn.

Step 1: look it up in the installed cuML

  • Check the versions and answer for them: python -c "import cuml, sklearn; print(cuml.__version__, sklearn.__version__)".

  • Before using a cuML class, function or argument, confirm it exists and read its documentation: python -c "import inspect, cuml; print(inspect.signature(cuml.svm.SVC))", help(cuml.manifold.UMAP). Docstrings list valid values, for example the accepted build_kwds keys and output_type values.

  • Run the code and read the errors: removed arguments raise TypeError, removed modules ModuleNotFoundError, invalid values usually list the valid ones.

  • Treat deprecated APIs as removed: do not recommend anything that emits a FutureWarning or is marked deprecated in its docstring, even if it still works in the installed version. Deprecation warnings only appear when the code runs, so run the code you suggest (on small data if needed) with python -W always::FutureWarning and read the warnings; they name the replacement.

  • For cuml.accel, see below for how to find out what runs on the GPU.

Step 2: check code for removed APIs

Run python scripts/check_removed_apis.py <files or dirs> (path relative to this skill) on the user's code and on code you write, before presenting it. It reports removed APIs whose replacement the installed cuML does not reveal, with the replacement, for example cuml.fil (use nvForest), SVC(probability=True) (use CalibratedClassifierCV) and renamed UMAP/HDBSCAN build_kwds keys that are silently ignored.

The checker only knows about these few APIs. An empty result does not mean the code is current, and it says nothing about whether cuml.accel runs it on the GPU; still do Step 1 and check cuml.accel fallbacks as described below.

Behavior that changed without an API change:

  • RandomForest is reproducible with random_state alone (since 25.10). Do not set n_streams=1 "for determinism"; that advice is outdated and only slows training.

  • KMeans.transform returns Euclidean, not squared, distances (since 26.10).

cuml.accel

  • Activate before sklearn is imported. In a notebook that already imported sklearn, restart the kernel and make %load_ext cuml.accel the first cell.

  • With cudf.pandas: python -m cudf.pandas -m cuml.accel script.py, or %load_ext cudf.pandas then %load_ext cuml.accel.

  • Prove where code ran instead of assuming: python -m cuml.accel -v script.py logs "ran on GPU" / "falling back to CPU: <reason>"; --profile (or %%cuml.accel.profile, or with cuml.accel.profile():) prints a per-call GPU/CPU table. cuml.accel.is_proxy(est) only says the estimator is wrapped, not that it ran on the GPU.

  • Unsupported parameters and inputs fall back to the CPU silently. Do not recall fallback lists: run with -v or --profile and read the reasons, or, without running, read the documentation for the installed release from GitHub: https://raw.githubusercontent.com/NVIDIA/cuml/v<cuml.__version__>/docs/source/cuml-accel/compatibility.rst (for example v26.08.00; for 26.06 and earlier the file is limitations.rst; for nightly builds such as 26.12.00a5 use main instead of the tag).

  • Single GPU only. To use several GPUs, run separate processes with CUDA_VISIBLE_DEVICES.

  • Models trained under accel pickle as plain sklearn/umap/hdbscan objects and load without cuML. The pickle bytes still contain the string cuml (a reconstructor), so do not use "no cuml in the bytes" as a check; load it in a CPU-only environment instead.

  • Results are comparable in quality, not numerically identical, to the CPU libraries. See references/parity-and-numerics.md before treating a difference as a bug.

Large data (UMAP, HDBSCAN, KMeans)

  • UMAP build_algo="auto" already uses NN-Descent above 50,000 rows (unless random_state is set, which forces brute force). HDBSCAN defaults to build_algo="brute_force".

  • Out of memory on the kNN build: batch it with build_algo="nn_descent", build_kwds={"knn_n_clusters": 4, "knn_overlap_factor": 2} (increase knn_n_clusters to 8, 16, ... to use less memory). Unknown keys are ignored without a warning, so spell them exactly.

  • Oversubscribe GPU memory with managed memory: import rmm; rmm.reinitialize(managed_memory=True) before creating estimators. cuml.accel enables managed memory by default (not on WSL2).

  • KMeans on host data larger than the GPU: KMeans(device_buffer_samples=...) streams host batches (26.08+).

  • HDBSCAN soft clustering: fit with prediction_data=True, then use cuml.cluster.hdbscan.all_points_membership_vectors, membership_vector and approximate_predict. There is no HDBSCAN.predict and no .prediction submodule.

Exporting models

HaveWantUse
cuML estimatorscikit-learn object for a CPU-only machineest.as_sklearn()
scikit-learn estimatorcuML estimatorcuml.<Estimator>.from_sklearn(skl)
cuML RandomForestTreelite / fast GPU or CPU inferencerf.as_treelite(), rf.as_nvforest()
sklearn / XGBoost / LightGBM forestfast inferencenvforest.load_from_sklearn(model), nvforest.load_model(path)
accel-trained modelONNXskl2onnx on the proxy (not DBSCAN, TSNE, UMAP, HDBSCAN, NearestNeighbors)

Outputs

Outputs mirror the input container: NumPy in, NumPy out; cuDF in, cuDF out. With cuDF or dask_cudf input, fitted attributes such as cluster_centers_ are cuDF objects, so call .to_pandas() / .to_numpy() before handing them to pandas or matplotlib. Force a type with output_type="cupy" on the estimator, cuml.set_global_output_type("cupy"), or with cuml.using_output_type("cupy"): (global and context settings take precedence). Non-float input is converted to float32 by default.

Files

LICENSE

Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.


                                 Apache License
                           Version 2.0, January 2004
                        http://www.apache.org/licenses/

   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION

   1. Definitions.

      "License" shall mean the terms and conditions for use, reproduction,
      and distribution as defined by Sections 1 through 9 of this document.

      "Licensor" shall mean the copyright owner or entity authorized by
      the copyright owner that is granting the License.

      "Legal Entity" shall mean the union of the acting entity and all
      other entities that control, are controlled by, or are under common
      control with that entity. For the purposes of this definition,
      "control" means (i) the power, direct or indirect, to cause the
      direction or management of such entity, whether by contract or
      otherwise, or (ii) ownership of fifty percent (50%) or more of the
      outstanding shares, or (iii) beneficial ownership of such entity.

      "You" (or "Your") shall mean an individual or Legal Entity
      exercising permissions granted by this License.

      "Source" form shall mean the preferred form for making modifications,
      including but not limited to software source code, documentation
      source, and configuration files.

      "Object" form shall mean any form resulting from mechanical
      transformation or translation of a Source form, including but
      not limited to compiled object code, generated documentation,
      and conversions to other media types.

      "Work" shall mean the work of authorship, whether in Source or
      Object form, made available under the License, as indicated by a
      copyright notice that is included in or attached to the work
      (an example is provided in the Appendix below).

      "Derivative Works" shall mean any work, whether in Source or Object
      form, that is based on (or derived from) the Work and for which the
      editorial revisions, annotations, elaborations, or other modifications
      represent, as a whole, an original work of authorship. For the purposes
      of this License, Derivative Works shall not include works that remain
      separable from, or merely link (or bind by name) to the interfaces of,
      the Work and Derivative Works thereof.

      "Contribution" shall mean any work of authorship, including
      the original version of the Work and any modifications or additions
      to that Work or Derivative Works thereof, that is intentionally
      submitted to Licensor for inclusion in the Work by the copyright owner
      or by an individual or Legal Entity authorized to submit on behalf of
      the copyright owner. For the purposes of this definition, "submitted"
      means any form of electronic, verbal, or written communication sent
      to the Licensor or its representatives, including but not limited to
      communication on electronic mailing lists, source code control systems,
      and issue tracking systems that are managed by, or on behalf of, the
      Licensor for the purpose of discussing and improving the Work, but
      excluding communication that is conspicuously marked or otherwise
      designated in writing by the copyright owner as "Not a Contribution."

      "Contributor" shall mean Licensor and any individual or Legal Entity
      on behalf of whom a Contribution has been received by Licensor and
      subsequently incorporated within the Work.

   2. Grant of Copyright License. Subject to the terms and conditions of
      this License, each Contributor hereby grants to You a perpetual,
      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
      copyright license to reproduce, prepare Derivative Works of,
      publicly display, publicly perform, sublicense, and distribute the
      Work and such Derivative Works in Source or Object form.

   3. Grant of Patent License. Subject to the terms and conditions of
      this License, each Contributor hereby grants to You a perpetual,
      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
      (except as stated in this section) patent license to make, have made,
      use, offer to sell, sell, import, and otherwise transfer the Work,
      where such license applies only to those patent claims licensable
      by such Contributor that are necessarily infringed by their
      Contribution(s) alone or by combination of their Contribution(s)
      with the Work to which such Contribution(s) was submitted. If You
      institute patent litigation against any entity (including a
      cross-claim or counterclaim in a lawsuit) alleging that the Work
      or a Contribution incorporated within the Work constitutes direct
      or contributory patent infringement, then any patent licenses
      granted to You under this License for that Work shall terminate
      as of the date such litigation is filed.

   4. Redistribution. You may reproduce and distribute copies of the
      Work or Derivative Works thereof in any medium, with or without
      modifications, and in Source or Object form, provided that You
      meet the following conditions:

      (a) You must give any other recipients of the Work or
          Derivative Works a copy of this License; and

      (b) You must cause any modified files to carry prominent notices
          stating that You changed the files; and

      (c) You must retain, in the Source form of any Derivative Works
          that You distribute, all copyright, patent, trademark, and
          attribution notices from the Source form of the Work,
          excluding those notices that do not pertain to any part of
          the Derivative Works; and

      (d) If the Work includes a "NOTICE" text file as part of its
          distribution, then any Derivative Works that You distribute must
          include a readable copy of the attribution notices contained
          within such NOTICE file, excluding those notices that do not
          pertain to any part of the Derivative Works, in at least one
          of the following places: within a NOTICE text file distributed
          as part of the Derivative Works; within the Source form or
          documentation, if provided along with the Derivative Works; or,
          within a display generated by the Derivative Works, if and
          wherever such third-party notices normally appear. The contents
          of the NOTICE file are for informational purposes only and
          do not modify the License. You may add Your own attribution
          notices within Derivative Works that You distribute, alongside
          or as an addendum to the NOTICE text from the Work, provided
          that such additional attribution notices cannot be construed
          as modifying the License.

      You may add Your own copyright statement to Your modifications and
      may provide additional or different license terms and conditions
      for use, reproduction, or distribution of Your modifications, or
      for any such Derivative Works as a whole, provided Your use,
      reproduction, and distribution of the Work otherwise complies with
      the conditions stated in this License.

   5. Submission of Contributions. Unless You explicitly state otherwise,
      any Contribution intentionally submitted for inclusion in the Work
      by You to the Licensor shall be under the terms and conditions of
      this License, without any additional terms or conditions.
      Notwithstanding the above, nothing herein shall supersede or modify
      the terms of any separate license agreement you may have executed
      with Licensor regarding such Contributions.

   6. Trademarks. This License does not grant permission to use the trade
      names, trademarks, service marks, or product names of the Licensor,
      except as required for reasonable and customary use in describing the
      origin of the Work and reproducing the content of the NOTICE file.

   7. Disclaimer of Warranty. Unless required by applicable law or
      agreed to in writing, Licensor provides the Work (and each
      Contributor provides its Contributions) on an "AS IS" BASIS,
      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
      implied, including, without limitation, any warranties or conditions
      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
      PARTICULAR PURPOSE. You are solely responsible for determining the
      appropriateness of using or redistributing the Work and assume any
      risks associated with Your exercise of permissions under this License.

   8. Limitation of Liability. In no event and under no legal theory,
      whether in tort (including negligence), contract, or otherwise,
      unless required by applicable law (such as deliberate and grossly
      negligent acts) or agreed to in writing, shall any Contributor be
      liable to You for damages, including any direct, indirect, special,
      incidental, or consequential damages of any character arising as a
      result of this License or out of the use or inability to use the
      Work (including but not limited to damages for loss of goodwill,
      work stoppage, computer failure or malfunction, or any and all
      other commercial damages or losses), even if such Contributor
      has been advised of the possibility of such damages.

   9. Accepting Warranty or Additional Liability. While redistributing
      the Work or Derivative Works thereof, You may choose to offer,
      and charge a fee for, acceptance of support, warranty, indemnity,
      or other liability obligations and/or rights consistent with this
      License. However, in accepting such obligations, You may act only
      on Your own behalf and on Your sole responsibility, not on behalf
      of any other Contributor, and only if You agree to indemnify,
      defend, and hold each Contributor harmless for any liability
      incurred by, or claims asserted against, such Contributor by reason
      of your accepting any such warranty or additional liability.

   END OF TERMS AND CONDITIONS

   APPENDIX: How to apply the Apache License to your work.

      To apply the Apache License to your work, attach the following
      boilerplate notice, with the fields enclosed by brackets "[]"
      replaced with your own identifying information. (Don't include
      the brackets!)  The text should be enclosed in the appropriate
      comment syntax for the file format. We also recommend that a
      file or class name and description of purpose be included on the
      same "printed page" as the copyright notice for easier
      identification within third-party archives.

   Copyright [yyyy] [name of copyright owner]

   Licensed under the Apache License, Version 2.0 (the "License");
   you may not use this file except in compliance with the License.
   You may obtain a copy of the License at

       http://www.apache.org/licenses/LICENSE-2.0

   Unless required by applicable law or agreed to in writing, software
   distributed under the License is distributed on an "AS IS" BASIS,
   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
   See the License for the specific language governing permissions and
   limitations under the License.

Requirements and compatibility

Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta

“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.

Needs: Needs Python

Security

What our scanner noticed· Medium risk

  • Large base64-encoded blobs
  • Instructions that use the network
  • Scripts that your agent may run

This skill includes scripts that your agent may run. Read them before you use it.

A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.

Attribution and licence

Published by
NVIDIA
Publisher
NVIDIA (Official)
Source
NVIDIA/skills · skills/cuml · commit fce3cca
Licence
Apache License 2.0 (redistribution allowed)
  • LICENSE (added by AutoGineer from LICENSE-APACHE)
  • ATTRIBUTION.md (written by AutoGineer)
Synced

AutoGineer is not affiliated with or endorsed by NVIDIA or NVIDIA.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download cuml.zip