In this hands-on demonstration of Cloud ML Engine, we are going to take a pre-configured Tensorflow learning model, train it on a local machine in a variety of configurations, then train it on Cloud ML Engine itself.
To best follow along and practice yourself, we have prepared a series of scripts that will automate everything we demonstrate in this lesson, although we encourage you to go through the steps manually on your first pass in order to become familiar with them.
To download the scripts, run the below command in your Cloud Shell session:
gsutil -m cp gs://gcp-course-exercise-scripts/data-engineer/cloud-ml_engine/* .
Note that we did not go through the hyperparameter tuning script actions due to the length of time required for it to complete; however you are free to do it on your own.
Please be aware that since resources are charged hourly, a long-running training job could add up financially.
Also be aware that we created a 'fun' superscript that combined many of the other scripts. You may use it for experimentation if you wish.