Databricks Runtime for Machine Learning (Databricks Runtime ML) provides a ready-to-go environment for machine learning and data science. It contains multiple popular libraries, including TensorFlow, PyTorch, Keras, and XGBoost. It also supports distributed training using Horovod.
Databricks Runtime ML lets you start a Databricks cluster with all of the libraries required for distributed training. It ensures the compatibility of the libraries included on the cluster (between TensorFlow and CUDA / cuDNN, for example) and substantially speeds up cluster start-up.
If you require HIPAA compliance, refer to HIPAA-compliant deployment.
Library utilities are not available in Databricks Runtime ML.
Databricks Runtime ML is built on Databricks Runtime. For example, Databricks Runtime 5.0 ML is built on Databricks Runtime 5.0. The libraries included in the base Databricks Runtime are listed in the Databricks runtime release notes.
The Databricks Runtime ML includes a variety of popular ML libraries. The libraries are updated periodically to include new features and fixes.
Databricks has designated a subset of the supported libraries as top-tier libraries. For these libraries, Databricks provides a faster update cadence, updating to the latest upstream package releases with each runtime release (barring dependency conflicts). Databricks also provides advanced support, testing, and embedded optimizations for top-tier libraries.
For a full list of top-tier and other provided libraries, see the following articles for each available runtime:
- Databricks Runtime 6.4 ML
- Databricks Runtime 6.3 ML
- Databricks Runtime 6.2 ML
- Databricks Runtime 6.1 ML
- Databricks Runtime 6.0 ML (Unsupported)
- Databricks Runtime 5.5 LTS ML
- Databricks Runtime 5.4 ML (Unsupported)
- Databricks Runtime 5.3 ML (Unsupported)
- Databricks Runtime 5.2 ML (Beta)
- Databricks Runtime 5.1 ML (Beta)
- Databricks Runtime 5.0 ML (Beta)
- Databricks Runtime 4.1 ML (Beta)
When you create a cluster, select a Databricks Runtime ML version from the Databricks Runtime Version drop-down. Both CPU and GPU-enabled ML runtimes are available.
If you select a GPU-enabled ML runtime, you are prompted to select a compatible Driver Type and Worker Type. Incompatible instance types are grayed out in the drop-downs. GPU-enabled instance types are listed under the GPU-Accelerated label.
Libraries in your workspace that automatically install into all clusters can conflict with the libraries included in Databricks Runtime ML. Before you create a cluster with Databricks Runtime ML, clear the Install automatically on all clusters checkbox for conflicting libraries.
In Databricks Runtime ML the Conda package manager is used to install Python packages. All Python packages are installed inside a single environment:
/databricks/python2 on clusters using Python 2 and
/databricks/python3 on clusters using Python 3. Switching (or activating) Conda environments is not supported.
You can use conda and pip commands to list and install packages.
%sh conda env list
%sh conda install matplotlib -y
When you run shell commands inside notebooks using
%sh, you cannot respond to interactive shells. To avoid blocking, pass the
-y (--yes) flag to
Any modifications to the current environment using this method are restricted to the notebook and the driver. The changes are reset when you detach and reattach the notebook. You can install a package on all workers using a library or an init script.
%sh conda install astropy
In Databricks Runtime 6.0 ML, you can directly use the
%sh conda command to install libraries on the driver node. The
-y option is not required since
always_yes: True is set in the conda configuration file.
To install a package on all cluster nodes, call
conda inside an init script. In your init script, activate the default environment and install packages using
#!/bin/bash set -ex /databricks/python/bin/python -V . /databricks/conda/etc/profile.d/conda.sh conda activate /databricks/python conda install -y astropy
By using this version of Databricks Runtime, you agree to the terms and conditions outlined in the NVIDIA End User License Agreement (EULA) with respect to the CUDA, cuDNN, and Tesla libraries, and the NVIDIA End User License Agreement (with NCCL Supplement) for the NCCL library.