arguments method of ProcessingStep results in unwanted upload to S3

[francisco-camargo]

I want to register a model made via SageMaker Pipelines. Following this as an example. Here is the code for model registration:

import logging

from sagemaker.drift_check_baselines import DriftCheckBaselines
from sagemaker.model_metrics import (
    MetricsSource,
    ModelMetrics,
    FileSource,
    )
from sagemaker.workflow.model_step import ModelStep

logger = logging.getLogger()
logger.setLevel(logging.INFO)
logger.addHandler(logging.StreamHandler())

def define_register_step(
    step_name,
    model,
    step_eval,
    step_explainability,
    model_explainability_check_config,
    model_package_group_name,
    model_approval_status,
    ):
    '''
    Register model step that will be conditionally executed
    '''
    
    logger.info(f'\tStart of define_register_step')
    my_args = step_eval.arguments # TODO: This is causing the upload to s3 of sourdir.tar.gz which contains evaluation.py
    logger.info(f'\tstep_eval.arguments: {my_args}')
    
    s3_folder_path = my_args["ProcessingOutputConfig"]["Outputs"][0]["S3Output"]["S3Uri"]
    s3_uri="{}/evaluation.json".format(s3_folder_path)
    model_statistics=MetricsSource(
            s3_uri=s3_uri,
            content_type="application/json"
            )
    model_metrics = ModelMetrics(model_statistics=model_statistics)
    drift_check_baselines = DriftCheckBaselines(
        explainability_constraints=MetricsSource(
            s3_uri=step_explainability.properties.BaselineUsedForDriftCheckConstraints,
            content_type="application/json",
            ),
        explainability_config_file=FileSource(
            s3_uri=model_explainability_check_config.monitoring_analysis_config_uri,
            content_type="application/json",
            ),
        )

    step_args = model.register(
        content_types=["text/csv"],
        response_types=["text/csv"],
        inference_instances=["ml.t2.medium", "ml.m5.large"],
        transform_instances=["ml.m5.large"],
        model_package_group_name=model_package_group_name,
        approval_status=model_approval_status,
        model_metrics=model_metrics,
        drift_check_baselines=drift_check_baselines,
        )
    
    step_register = ModelStep(
        name=step_name,
        step_args=step_args,
        )

    return step_register

The line step_eval.arguments results in an upload of sourdir.tar.gz containing evaluate.py to S3. I don't want this behavior, I just need the function defined in the code above to be able to get the S3 path it needs to read the model metrics of interest. Guidance on how to accomplish this would be much appreciated!