Can I pass arguments to the entrypoint of a SageMaker estimator?

Question

I'm using the SageMaker python sdk and was hoping to pass in some arguments to be used by my entrypoint, I'm not seeing how to do this.

from sagemaker.sklearn.estimator import SKLearn  # sagemaker python sdk

entrypoint = 'entrypoint_script.py'

sklearn = SKLearn(entry_point=entrypoint,  # <-- need to pass args to this
                  train_instance_type=instance_class,
                  role=role,
                  sagemaker_session=sm)

Answer 1

The answer is no as there is no parameter on the Estimator base class, or the fit method, that accepts arguments to pass to the entrypoint.

I resolved this by passing the parameter as part of the hyperparameter dictionary. This gets passed to the entrypoint as arguments.

Answer 2

The entry_point argument here is for a script file that contains the codes you want to run training and prediction.

You can refer to the example here

And the entry_point script above example uses is here