actually doing the training for the basic control

[yamins81]

What is the loss function that we'll be using for the control? How will we handle multiple labels? Softmax matching?

[yamins81]

(1) figure out that model
(2) figuring out the loss function
(3) figure out what optimizer, with which learning rate schedule
(4) putting it into tfutils script
(5) actually run it on machines at Stanford