Hypoparameters tuning of gaussian process regression using bayesian optimization

I am working on a project to train and deploy a gaussian process regression model in Sagemaker. I created a sagemaker.sklearn estimator with a custom .py file as the endpoint, and successfully trained it with .fit method.

Currently, I am stuck at how to tune the hypoparameters with Bayesian Optimization in parallel with defined thread_nr or/and instance_nr etc.

I went through several sagemaker examples and am aware of the Amazon SageMaker automatic model tuning (AMT) service. But still don't have clue of my next move.

Any tip would be highly appreciated!

Keywords: gaussian process, bayesian optimization, parallel.