options for finer-grained control over the optimizer

[dmarx]

we only get learning rate. I want to give user ability to change the underlying optimizer. doesn't always have to be adamW, and I want to know if different optimizers have their own associated visual aesthetics. the same way you can often tell what app generated an image from how it looks, can you maybe tell what optimizer was used?

can use a getattr, let user type in name of optimizer. use adamw as default if getattr fails to parse user input