training_config does not return MetricDefinitions

[shunjd]

Describe the bug
sagemaker.workflow.airflow.training_config does not return the MetricDefinitions when it is provided in the estimator. However, tuning_config does have it.

Related: aws/aws-step-functions-data-science-sdk-python#40

MetricDefinitions in tuning_config

sagemaker-python-sdk/src/sagemaker/workflow/airflow.py

Line 372 in aceb8e0

train_config["AlgorithmSpecification"]["MetricDefinitions"] = tuner.metric_definitions

To reproduce

estimator = sagemaker.estimator.Estimator(
        PCA_IMAGE,
        role=EXECUTION_ROLE,
        train_instance_count=1,
        train_instance_type='ml.c4.xlarge',
        output_path=s3_output_location,
        metric_definitions=[{'Name': 'foo', 'Regex': 'bar=(.*?);'}]
    )

result = training_config(estimator)
print(result)

Expected behavior
MetricDefinitions should be included in the result.

Screenshots or logs
N/A

System information
A description of your system. Please provide:

• SageMaker Python SDK version: latest
• Framework name (eg. PyTorch) or algorithm (eg. KMeans): Estimator/TensforFlow
• Framework version:
• Python version:
• CPU or GPU:
• Custom Docker image (Y/N): N

Additional context
N/A

[ajaykarpur]

Hi @shunjd, thanks for bringing this to our attention. I've submitted a pull request to address this issue: #1453

[laurenyu]

looks like this was released in v1.56.2