Work with compute targets in Azure Machine Learning: Review

I clicked 'Create compute instance', gave it a name, and then waited. And waited. Then I did the same thing for a compute cluster. The exercise never asks you to run a script on either resource, so you don't actually see *why* you'd pick one over the other. The text explains the difference (instance for interactive dev, cluster for training jobs), but the hands-on part doesn't reinforce it at all. It's like a car manual telling you about the gas pedal and the brake, but the driving test is just turning the key in the ignition.

I was hoping for a little more. Maybe a small script that fails on a CPU-only instance and needs a cluster, or a scenario where you have to decide which one to provision based on a budget or a time constraint. Instead, it's just pointing and clicking. The sandbox is nice because you don't have to use your own subscription or worry about a bill, but the actual activity is so basic it feels pointless.

I understand this is for beginners, but it's so fundamental that anyone who has done the most basic 'train a model' tutorial in Azure will have already done this. It's more of an interactive glossary entry than a learning module. I finished it in about 20 minutes, most of which was spent watching resources spin up. Feels like a chunk of a larger course was amputated and presented as a standalone thing.