Repository files navigation

cuML is a suite of libraries that implement machine learning algorithms and mathematical primitives functions that share compatible APIs with otherRAPIDS projects.

cuML enables data scientists, researchers, and software engineers to runtraditional tabular ML tasks on GPUs without going into the details of CUDAprogramming. In most cases, cuML's Python API matches the API fromscikit-learn.

For large datasets, these GPU-based implementations can complete 10-50x fasterthan their CPU equivalents. For details on performance, see thecuML BenchmarksNotebook.

As an example, the following Python snippet loads input and computes DBSCAN clusters, all on GPU, using cuDF:

importcudffromcuml.clusterimportDBSCAN# Create and populate a GPU DataFramegdf_float=cudf.DataFrame()gdf_float['0']= [1.0,2.0,5.0]gdf_float['1']= [4.0,2.0,1.0]gdf_float['2']= [4.0,2.0,1.0]# Setup and fit clustersdbscan_float=DBSCAN(eps=1.0,min_samples=1)dbscan_float.fit(gdf_float)print(dbscan_float.labels_)

Output:

0    01    12    2dtype: int32

cuML also features multi-GPU and multi-node-multi-GPU operation, usingDask, for agrowing list of algorithms. The following Python snippet reads input from a CSV file and performsa NearestNeighbors query across a cluster of Dask workers, using multiple GPUs on a single node:

Initialize aLocalCUDACluster configured withUCX for fast transport of CUDA arrays

# Initialize UCX for high-speed transport of CUDA arraysfromdask_cudaimportLocalCUDACluster# Create a Dask single-node CUDA cluster w/ one worker per devicecluster=LocalCUDACluster(protocol="ucx",enable_tcp_over_ucx=True,enable_nvlink=True,enable_infiniband=False)

Load data and performk-Nearest Neighbors search.cuml.dask estimators also supportDask.Array as input:

fromdask.distributedimportClientclient=Client(cluster)# Read CSV file in parallel across workersimportdask_cudfdf=dask_cudf.read_csv("/path/to/csv")# Fit a NearestNeighbors model and query itfromcuml.dask.neighborsimportNearestNeighborsnn=NearestNeighbors(n_neighbors=10,client=client)nn.fit(df)neighbors=nn.kneighbors(df)

For additional examples, browse our completeAPIdocumentation, or check out ourexamplewalkthroughnotebooks. Finally, youcan find complete end-to-end examples in thenotebooks-contribrepo.

Seethe RAPIDS ReleaseSelector for the commandline to install either nightly or official release cuML packages via Conda orDocker.

See the buildguide.

Please see ourguide for contributing to cuML.

The RAPIDS team has a number of blogs with deeper technical dives and examples.You can find them here on Medium.

For additional details on the technologies behind cuML, as well as a broader overview of the Python Machine Learning landscape, seeMachine Learning in Python: Main developments and technology trends in data science, machine learning, and artificial intelligence (2020) by Sebastian Raschka, Joshua Patterson, and Corey Nolet.

Please consider citing this when using cuML in a project. You can use the citation BibTeX:

@article{raschka2020machine,title={Machine Learning in Python: Main developments and technology trends in data science, machine learning, and artificial intelligence},author={Raschka, Sebastian and Patterson, Joshua and Nolet, Corey},journal={arXiv preprint arXiv:2002.04803},year={2020}}

Find out more details on theRAPIDS site

The RAPIDS suite of open source software libraries aim to enable execution of end-to-end data science and analytics pipelines entirely on GPUs. It relies on NVIDIA® CUDA® primitives for low-level compute optimization, but exposing that GPU parallelism and high-bandwidth memory speed through user-friendly Python interfaces.