Create or update a Studio app with a training plan using the SageMaker API or AWS CLI

To use SageMaker training plans for your SageMaker Studio app, specify the ARN of the training plan in the TrainingPlanArn parameter of the ResourceSpec when calling the CreateApp API operation.

The following example shows how to create a JupyterLab or Code Editor app with a training plan using the AWS CLI. Replace the --app-type value with JupyterLab or CodeEditor as needed. The InstanceType must match the instance type of your training plan:

aws sagemaker create-app \
  --domain-id d-xxxxxxxxxxxx \
  --space-name my-space \
  --app-type JupyterLab \
  --app-name default \
  --resource-spec '{
    "InstanceType": "instance-type",
    "TrainingPlanArn": "arn:aws:sagemaker:us-east-1:123456789012:training-plan/my-training-plan"
  }'

After you create the app, you can verify the training plan association by calling the DescribeApp API. The response includes the TrainingPlanArn in the ResourceSpec if the app was created with a training plan.

aws sagemaker describe-app \
  --domain-id d-xxxxxxxxxxxx \
  --space-name my-space \
  --app-type JupyterLab \
  --app-name default

To update the training plan for an existing app, delete the current app and create a new one with the updated training plan ARN using the create-app command shown above:

aws sagemaker delete-app \
  --domain-id d-xxxxxxxxxxxx \
  --space-name my-space \
  --app-type JupyterLab \
  --app-name default