After PR #396, notebook results in NoneType compute_target

[clmccart]

After PR #396, compute_target variable ends up being none when running out of the box. If you create a workspace from the portal, there is no default compute target created.
This command creates a nonetype object:

import os

cluster_type = os.environ.get("AML_COMPUTE_CLUSTER_TYPE", "CPU")
compute_target = ws.get_default_compute_target(cluster_type)

which results in a traceback error on this command:

from azureml.train.estimator import Estimator

script_params = {
    '--data-folder': ds.path('mnist').as_mount(),
    '--regularization': 0.05
}

est = Estimator(source_directory=script_folder,
                script_params=script_params,
                compute_target=compute_target,
                entry_script='train.py',
                conda_packages=['scikit-learn'])

Workaround: create a compute target in the portal and replace command 1 with:

   compute_target = ws.compute_targets[compute_name]

[v-strudm-msft]

Thank you for your post and participation. We will close this thread.