Task level update when number of update increases. This is called after optimization step and learning rate update of each step
(self, num_updates)
source not stored for this graph (policy: none)
no outgoing calls
no test coverage detected