I've been following the CNTK Tutorial CNTK 203: Reinforcement Learning Basics and encountered sending the arguments to the Learner in the Brain class in the following way:
arguments = dict(zip(self.loss.arguments, [x,y]))
updated, results =self.trainer.train_minibatch(arguments, outputs=[self.loss.output])
This is the first time this is done in the tutorials. Can anyone understand why the arguments are a dictionary of zip objects? What does zip do in this scenario?
zip is a builtin function in python that creates tuples out of its list arguments. See this for more details. The code in question just createes a dictionary mapping the input variables of the network to the values x and y.