What do you need to keep in mind when training ML models that will be deployed on constrained devices such as phones, Raspberry Pis, or even Arduinos? This session goes into the considerations and the code. snippets required for these.