cuml-cu12

What it is and what it does

cuML is a GPU-accelerated machine learning library from the RAPIDS ecosystem that brings standard ML algorithms to NVIDIA GPUs. It provides scikit-learn-compatible interfaces for algorithms including random forests, gradient boosting, clustering, and dimensionality reduction, allowing developers to scale training and inference on GPU hardware without rewriting model code.

The package wraps libcuml C++ implementations and depends on cuDF for GPU DataFrames, cuPy for GPU arrays, and RAFT for shared ML primitives. It targets developers who need to accelerate existing scikit-learn workflows on GPU clusters or single-GPU systems, and requires CUDA 12, Python 3.11+, and a compatible NVIDIA GPU. The library is actively maintained as part of the RAPIDS project and integrates with joblib, numba, and other ecosystem tools.

Use it for:

• Train random forest or gradient boosting models on GPU for faster convergence on large datasets.
• Accelerate hyperparameter tuning and cross-validation workflows by moving model training to GPU.
• Build GPU-native ML pipelines in Dask for distributed training across multiple GPUs.
• Replace CPU scikit-learn calls in existing code with GPU equivalents using the same API.
• Perform fast clustering and dimensionality reduction on high-dimensional data.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

GPU-accelerated machine learning algorithms for classification, regression, clustering, and dimensionality reduction, built on CUDA and integrated with scikit-learn-compatible APIs.

Yes, if you have an NVIDIA GPU with CUDA 12 and need to accelerate scikit-learn-compatible workflows. The permissive Apache-2.0 license and active maintenance are favorable. No if you lack GPU hardware or need CPU-only ML. Medium friction from 18 dependencies and CUDA/GPU requirements; verify your GPU and CUDA setup before attempting installation.

Install

cuml-cu12 on PyPI

pip install cuml-cu12

uv add cuml-cu12

poetry add cuml-cu12

Installing cuml-cu12

Before you install

Medium install friction due to 18 runtime dependencies including CUDA toolkit, cuDF, cuPy, and GPU-specific libraries. Active maintenance with recent releases; requires CUDA 12 and Python 3.11+. Pre-built wheels available for x86_64 and aarch64 Linux.

License in practice

Apache-2.0 permissive license allows commercial and private use with minimal restrictions; suitable for most production and research contexts.

Quickstart

pip install cuml-cu12

import cuml
from cuml.ensemble import RandomForestClassifier

rf = RandomForestClassifier()
rf.fit(X_train, y_train)
predictions = rf.predict(X_test)

Requires NVIDIA GPU with CUDA 12 support and CUDA toolkit installed; cuDF and cuPy must match the cu12 variant.

Verify before relying

• Whether all 18 runtime dependencies are required for basic usage or if subsets enable lighter installations.
• Performance gains over CPU alternatives for typical dataset sizes and model types.
• Compatibility with specific GPU architectures beyond the wheel's aarch64 and x86_64 coverage.

Package facts

Evidence: cuml_cu12-26.8.0-cp311-abi3-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl; cuml_cu12-26.8.0-cp311-abi3-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl

Tags

More Artificial Intelligence packages

Further reading