C1_W3_Lab_3_custom-layer-activation: super().build(input_shape)

Course Q&A TensorFlow: Advanced Techniques Specialization Custom Models, Layers and Loss Functions with TF
1

Hi

I have a question about “super().build(input_shape)” in the code of the Lab3 of the Week 3.
To be specific, this is found in the link below:

In the given code, we have 2 kinds of super() function.

I understand that we need to have super().__init__() in the first definition __init__() in the Class statement to give access to methods and properties of a parent or sibling class.

Nevertheless, I have no clue what the second ‘super()’ function is doing in the definition of ‘build()’ function. More specifically,

super().build(input_shape)

What is the purpose of this?
And moreover, why don’t we see this in the W1 Assignment code?

Thanks
Kind regards
Michio

2

If you search this specialization’s part of the forum you will find topics similar to this, so have a look on them.

3

It’s good practice to invoke super class method when overriding when there’s bookkeeping functionality there. In this case, it’s not necessay since this line takes care of it (look for the same line for your tensorflow version).

1 Like
4

Thanks @balaji.ambresh