SageMaker Distribution Images - Amazon SageMaker
Supported packages and versions
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

SageMaker Distribution Images

Important

Currently, all packages in SageMaker Distribution images are licensed for use with Amazon SageMaker and do not require additional commercial licenses. However, this might be subject to change in the future, and we recommend reviewing the licensing terms regularly for any updates.

SageMaker Distribution is a collection of Docker images, which includes popular libraries and packages for machine learning, data science, and data analytics visualization. The Docker images include deep learning frameworks such as the following:

  • PyTorch

  • TensorFlow

  • Keras

It also includes popular Python packages such as the following:

  • numpy

  • scikit-learn

  • pandas

Within the container, you can use the following IDEs:

  • JupyterLab

  • Code Editor, based on Code-OSS (Visual Studio Code Open Source)

Each SageMaker Distribution image has a GPU variant and a CPU variant.

SageMaker Distribution is available in:

  • Studio

  • Studio Lab

The packages included in the container are guaranteed to be compatible with each other and the runtime is built to work anywhere. You can use the container to run Amazon SageMaker Studio notebooks or SageMaker training jobs. You can also run the container on a local laptop. Use SageMaker Distribution to quickly get started with ML development in your local environment. Seamlessly transition to tasks such as the batch execution of training jobs without needing to reconfigure your runtime environment.

For the list of all supported libraries within SageMaker distribution and their corresponding versions, see the SageMaker Distribution GitHub. You can also use the pre-built and ready-to-use SageMaker Distribution images from the Amazon Elastic Container Registry Gallery.

Supported packages and versions

For the list of the packages that are installed in a version of SageMaker Distribution, see the RELEASE.md file in the build_artifacts directory of the SageMaker DistributionGitHub repository.

Major A major version release of Amazon SageMaker Distribution upgrades all of its core dependencies to the latest compatible version. SageMaker Distribution can add or remove packages in a major version release. Major versions are denoted by the first number in the version string. For example, 1.0, 2.0, 3.0. Half-yearly
Minor A minor version release of Amazon SageMaker Distribution ensures that all of its core dependencies are updated to the latest compatible minor version within the same major version. SageMaker Distribution can add new packages during a minor version release. Minor versions are denoted by the second number in the version string. For example, 1.1, 1.2, or 2.1 Monthly (additional minor versions released on an add needed basis as well)
Patch A patch version release of Amazon SageMaker Distribution ensures that all its core dependencies are updated to the latest compatible patch version within the same minor version. SageMaker Distribution does not add or remove packages during a patch version release. 7 days (overnight fixes also deployed based on the severity)
Important

  • SageMaker Distribution v0.x.y is only used in Studio Classic. SageMaker Distribution v1.x.y is only used in JupyterLab.

  • We try to update the Studio images with new versions regularly. If the packages in the Distribution image are out of date, we recommend waiting for the next update.

  • Some dependencies, such as Python, are treated differently. Amazon SageMaker Distribution allows for a minor upgrade of Python with a release. For example, you can upgrade Python 3.10 to Python 3.11 when you upgrade from version 4.8 to 5.0.