Parameter name in docs is incorrect
Closed this issue · 0 comments
ALaks96 commented
Describe the bug
Docs give the wrong parameter name to launch a training specifying what package (TF, pytorch, etc.) is needed, as a result base docker images (Azure ML images) are not loaded.
To Reproduce
Launch a training using the syntax in the docs:
mount_dataset_name = 'ds_to_mount'
download_dataset_name = 'ds_to_download'
trainer.start_training(training_name, estimator_type='tensorflow',
input_datasets = [mount_dataset_name],
input_datasets_to_download = [download_dataset_name],
compute_target='gp-cluster', gpu_compute=True, script_parameters = arguments)
Expected behavior
Should load Azure ML TF base docker image as base environment for the scheduled training.