sagemaker - construct model from tar.gz file

Question

I have trained a sagemaker model successfully and the model.tar.gz file is on s3.

Now I want to "reconstruct" the model from that file and then deploy it. I used the following code:

containers = {'us-west-2': '174872318107.dkr.ecr.us-west-2.amazonaws.com/factorization-machines:latest',
              'us-east-1': '382416733822.dkr.ecr.us-east-1.amazonaws.com/factorization-machines:latest',
              'us-east-2': '404615174143.dkr.ecr.us-east-2.amazonaws.com/factorization-machines:latest',
              'eu-west-1': '438346466558.dkr.ecr.eu-west-1.amazonaws.com/factorization-machines:latest'}


fm = sagemaker.model.Model(model_s3_path, containers['eu-west-1'], role=sagemaker.get_execution_role())

I get back an object of type sagemaker.model.Model .

I then seek to deploy the model via

fm_predictor = fm.deploy(instance_type='ml.c4.xlarge', initial_instance_count=1)

The output of this call is

--------------------------------------------------------------------------------------!

But this returns a NoneType object that does not have a predict method. However, the model's endpoint is created.

What am I doing wrong?

Answer 1

In my code samples, the deployment also does not have a relevant return value. But after the deployment, I have to create the predictor with the endpoint name:

endpoint_name = 'rf-scikit-endpoint-xxx'

model.deploy(
  instance_type='ml.c5.large',
  initial_instance_count=1,
  endpoint_name=endpoint_name)

predictor = sagemaker.sklearn.model.SKLearnPredictor(endpoint_name=endpoint_name)

Can you proceed in a similar way?