Studio Lab pre-installed environments - Amazon SageMaker

Studio Lab pre-installed environments

Amazon SageMaker Studio Lab uses conda environments to manage packages (or libraries) for your projects. This guide explains what conda environments are, how to interact with them, and the different pre-installed environments available in Studio Lab.

A conda environment is a directory that contains a collection of packages you have installed. It allows you to create isolated environments with specific package versions, preventing conflicts between projects with different dependencies.

You can interact with conda environments in Studio Lab in two ways:

  • Terminal: Use the terminal to create, activate, and manage environments.

  • JupyterLab Notebook: When opening a JupyterLab notebook, select the kernel with the environment name you wish to use, to use the packages installed in that environment.

For a walkthrough on managing environments, see Manage your environment

Studio Lab comes with several pre-installed environments that are either persistent or non-persistent memory environments. Any changes made to persistent memory environments will remain for your next session. Any changes to non-persistent memory environments will not remain for your next sessions, but the packages within will be updated and tested for compatability by Amazon SageMaker. Here's an overview of each environment and its use case:

  • sagemaker-distribution: A non-persistent environment managed by Amazon SageMaker. It contains popular packages for machine learning, data science, and visualization. This environment is regularly updated and tested for compatibility. Use this environment if you want a fully-managed setup with common packages pre-installed.

    The sagemaker-distribution environment is closely related to the environment used in Amazon SageMaker Studio Classic, so after graduating from Studio Lab to Studio Classic the notebooks should run similarly. For information on exporting your environment from Studio Lab to Studio Classic, see Export an Amazon SageMaker Studio Lab environment to Amazon SageMaker Studio Classic.

  • default: A persistent environment with minimal packages pre-installed. Use this environment if you want to customize it significantly by installing additional packages.

  • studiolab: A persistent environment where JupyterLab and related packages installed. Use this environment for configuring the JupyterLab user interface and installing Jupyter server extensions.

  • studiolab-safemode: A non-persistent environment activated automatically when there's an issue with your project runtime. Use this environment for troubleshooting purposes. For information on troubleshooting, see Troubleshooting.

  • base: A non-persistent environment used for system tooling. This environment is not intended for customer use.

To view the packages in an environment, run the command conda list.

For more information on installing packages within your environment, see Customize your environment.

If you plan to graduate from Studio Lab to Amazon SageMaker Studio Classic, see Export an Amazon SageMaker Studio Lab environment to Amazon SageMaker Studio Classic.

For information on SageMaker images and their versions, see Amazon SageMaker images available for use with Studio Classic.