Issue with registration of a SageMaker model

[Kedharkb]

I'm currently encountering an issue when trying to register a SageMaker model in a pipeline using the Python SDK. Here's the relevant code:

 model = Model(
        image_uri=image_uri,
        model_data=step_train.properties.ModelArtifacts.S3ModelArtifacts,
        sagemaker_session=sagemaker_session,
        role=role,
    )

    model_register = model.register(
        content_types=["text/csv"],
        response_types=["text/csv"],
        model_metrics=model_metrics,
        inference_instances=["ml.t2.medium", "ml.m5.large"],
        transform_instances=["ml.m5.large"],
        approval_status="PendingManualApproval",
        model_package_group_name="ReturnModelPackage",
    )

    step_model_register = ModelStep(
        name="ReturnModelRegister", step_args=model_register
    )

When I run this code, I get the following error message:

Traceback (most recent call last):
  File "main_pipeline.py", line 288, in <module>
    pipeline = get_pipeline()
  File "main_pipeline.py", line 240, in get_pipeline
    model_register = model.register(
  File "/usr/local/lib/python3.8/dist-packages/sagemaker/workflow/pipeline_context.py", line 272, in wrapper
    return run_func(*args, **kwargs)
  File "/usr/local/lib/python3.8/dist-packages/sagemaker/model.py", line 425, in register
    model_package = self.sagemaker_session.create_model_package_from_containers(
  File "/usr/local/lib/python3.8/dist-packages/sagemaker/session.py", line 3233, in create_model_package_from_containers
    return self._intercept_create_request(
  File "/usr/local/lib/python3.8/dist-packages/sagemaker/session.py", line 4813, in _intercept_create_request
    return create(request)
  File "/usr/local/lib/python3.8/dist-packages/sagemaker/session.py", line 3231, in submit
    return self.sagemaker_client.create_model_package(**request)
  File "/usr/local/lib/python3.8/dist-packages/botocore/client.py", line 530, in _api_call
    return self._make_api_call(operation_name, kwargs)
  File "/usr/local/lib/python3.8/dist-packages/botocore/client.py", line 919, in _make_api_call
    request_dict = self._convert_to_request_dict(
  File "/usr/local/lib/python3.8/dist-packages/botocore/client.py", line 990, in _convert_to_request_dict
    request_dict = self._serializer.serialize_to_request(
  File "/usr/local/lib/python3.8/dist-packages/botocore/validate.py", line 381, in serialize_to_request
    raise ParamValidationError(report=report.generate_report())
  File "/usr/local/lib/python3.8/dist-packages/botocore/validate.py", line 102, in generate_report
    error_messages.append(self._format_error(error))
  File "/usr/local/lib/python3.8/dist-packages/botocore/validate.py", line 125, in _format_error
    f'Invalid type for parameter {name}, value: {param}, '
  File "/usr/local/lib/python3.8/dist-packages/sagemaker/workflow/entities.py", line 86, in __str__
    raise TypeError(
TypeError: Pipeline variables do not support __str__ operation. Please use `.to_string()` to convert it to string type in execution timeor use `.expr` to translate it to Json for display purpose in Python SDK.

Solution tried: Followed the documentation to register the model using the Model.register() method. Also tried using the RegisterModel() method, which worked without any issue. However, we need to register the model using the Model.register() method.
Request: Please provide a solution to successfully register a SageMaker model in a pipeline using SageMaker Python SDK's Model.register() method without encountering the "TypeError: Pipeline variables do not support str operation" error.

[qidewenwhen]

Hi @Kedharkb, thanks for using SageMaker!

The ModelStep is a new interface we introduced last year which requires a PipelineSession rather than the regular Session to make it work. Please find details in the link.
Without using it, you'll get the TypeError as mentioned in this Github issue.

Can you make the following change and check if it works for you:

from sagemaker.workflow.pipeline_context import PipelineSession

pipeline_session = PipelineSession()

 model = Model(
        image_uri=image_uri,
        model_data=step_train.properties.ModelArtifacts.S3ModelArtifacts,
        sagemaker_session=pipeline_session,  <<<<<<<<<<<<<<<<<<<<
        role=role,
    )

    model_register = model.register(...)

    step_model_register = ModelStep(...)

[qidewenwhen]

Closing the issue as no further requests received. Feel free to reopen if you have any further questions. Thanks!