Confused with the use of super() in Pytorch for neural net construction

Question

I understand the use of super() to inherit attributes and methods from a class when initializing your class. However, when using this in the context of neural n