Changing Environment
Often times, a specific step will require a particular base image to execute inside of. This may be because the image was pre-built with particular dependencies (packages, libraries, drivers, etc) which Jupyter does not need to execute, but may be necessary on the back end. This feature allows to specify the base image required to execute a given step.
Editing the Notebook to Allow Tags¶
First, go to the notebook settings and view "Cell Metadata":
Adding an Environment Specifier¶
You can add alternate base images with the following tag structure:
environment=environment_name
For example:
environment=default
or:
environment=private-training-env
Update SAME File¶
Then in your SAME file, you'll add a section that maps to the specific image you'll need.