[Bug fix] Create the model object in a pipeline

batch-inference/seedcode/build.py

@@ -169,7 +169,9 @@ def create_cfn_params_tags_file(config, export_params_file, export_tags_file):
         json.dumps({
             'region':args.aws_region,
             'pipeline_name':f'{args.sagemaker_project_name}-{args.sagemaker_project_id}-BatchInference',
-            'base_job_prefix':f'{args.sagemaker_project_name}-{args.sagemaker_project_id}'
+            'base_job_prefix':f'{args.sagemaker_project_name}-{args.sagemaker_project_id}',
+            'model_package_arn': model_package_arn,
+            'model_execution_role': args.model_execution_role
         })
     )
 

batch-inference/seedcode/endpoint-config-template.yml

@@ -49,27 +49,14 @@ Parameters:
       The body of the SageMaker Pipeline.
 
 Resources:
-  ModelToDeploy:
-    Type: AWS::SageMaker::Model
-    Properties:
-      ExecutionRoleArn: !Ref ModelExecutionRoleArn
-      Containers: 
-        - ModelPackageName: !Ref ModelPackageName
-      Tags:
-        - Key: sagemaker:project-name
-          Value: !Sub ${SageMakerProjectName}
-        - Key: sagemaker:project-id
-          Value: !Sub ${SageMakerProjectId}
   BatchPipeline:
     Type: AWS::SageMaker::Pipeline
     Properties: 
       PipelineDescription: The SM Pipeline that executes the batch inference
       PipelineName: !Sub ${SageMakerProjectName}-${StageName}-BatchPipeline
       RoleArn: !Ref ModelExecutionRoleArn
       PipelineDefinition: 
-        PipelineDefinitionBody: !Sub
-          - ${PipelineDefinitionBody}
-          - { ModelName: !GetAtt ModelToDeploy.ModelName }
+          PipelineDefinitionBody: !Ref PipelineDefinitionBody
       Tags:
         - Key: sagemaker:project-name
           Value: !Ref SageMakerProjectName
@@ -78,7 +65,7 @@ Resources:
   EventRule:
     Type: AWS::Events::Rule
     Properties:
-      Name: !Sub ${SageMakerProjectName}-${StageName}-SchedExecRule
+      Name: !Sub sagemaker-${SageMakerProjectName}-${StageName}-SchedExecRule
       ScheduleExpression: !Sub rate(${ScheduleExpressionforPipeline})
       Targets:
         - Arn: !Sub arn:aws:sagemaker:${AWS::Region}:${AWS::AccountId}:pipeline/${SageMakerProjectName}-${StageName}-batch-pipeline

batch-inference/seedcode/pipelines/batch_inference/pipeline.py

@@ -2,7 +2,7 @@
 
 import boto3
 import sagemaker
-from sagemaker import model
+from sagemaker import model, ModelPackage
 import sagemaker.session
 from sagemaker.workflow.parameters import (
     ParameterInteger,
@@ -68,37 +68,44 @@ def get_pipeline_custom_tags(new_tags, region, sagemaker_project_arn=None):
 
 def get_pipeline(
     region,
+    model_package_arn,
+    model_execution_role,
     pipeline_name="AbalonePipelineBatchInference",
     base_job_prefix="Abalone",
 ):
     """Gets a SageMaker ML Pipeline instance working with on abalone data.
 
     Args:
         region: AWS region to create and run the pipeline.
-        model_name: Name of the SageMaker Model to deploy
-
+        model_package_arn: The Amazon Resource Name (ARN) of the SageMaker model package group to deploy
+        model_execution_role: The Amazon Resource Name (ARN) of the IAM role that SageMaker can assume to access model artifacts and docker image for deployment
+        base_job_prefix: Base job prefix for jobs in the SageMaker pipeline
     Returns:
         an instance of a pipeline
     """
     sagemaker_session = get_session(region)
 
     #### PARAMETERS
-    model_name = ParameterString("ModelName", default_value='${ModelName}')
     batch_inference_instance_count = ParameterInteger("BatchInstanceCount", default_value=1)
     batch_inference_instance_type = ParameterString("BatchInstanceType", default_value='ml.m5.xlarge')
     input_path = ParameterString("InputPath", default_value=f"s3://sagemaker-servicecatalog-seedcode-{region}/dataset/abalone-dataset.csv")
     output_path = ParameterString("OutputPath")
 
+    model = ModelPackage(
+        role=model_execution_role,
+        model_package_arn=model_package_arn,
+        sagemaker_session=sagemaker_session
+    )
+
     #### SAGEMAKER CONSTRUCTS
-    transform = Transformer(
-        model_name=model_name,
+    transform = model.transformer(
         instance_count=1,
-        instance_type='ml.m5.xlarge',
+        instance_type='ml.m5.large',
         output_path=output_path,
-        base_transform_job_name=f"{base_job_prefix}/batch-transform-job",
         max_payload=10,
         accept='text/csv'
     )
+    transform.base_transform_job_name = f"{base_job_prefix}/batch-transform-job",
 
     #### STEPS
     transform_step = TransformStep(
@@ -110,7 +117,7 @@ def get_pipeline(
     #### PIPELINE
     pipeline = Pipeline(
         name=pipeline_name,
-        parameters=[model_name, batch_inference_instance_count, batch_inference_instance_type, input_path, output_path],
+        parameters=[batch_inference_instance_count, batch_inference_instance_type, input_path, output_path],
         steps=[transform_step],
         sagemaker_session=sagemaker_session
     )