In order for the just-laid-out Gaussian process to run, we have to allow our model builds to be run on Google Cloud so that hyperparameter tuning can be carried out.

In order to run hyperparameter tuning, the following are essential components:

• Data file and its location
• Model file
• Hyperparameter configuration file
• Setup file
• __init__ file

Given that we are running the model on Google Cloud ML engine, the data should be residing in a Cloud bucket so that it becomes accessible to ML engine.

This can be done by performing the following in the Cloud shell:

gsutil mb gs://my-mnist-bucket
gsutil cp -r data/mnist.pkl gs://my-mnist-bucket/data/mnist.pkl

Note that, using the preceding steps, we have created a bucket named my-mnist-bucket and copied our data to that bucket. The preceding code should result in creating a directory named data and the mnist.pkl file in that directory: