Course 4 Week 1 Assignment 1 - Exercise 5 conv_backward()

I have a question regarding the final output of dW and db. In the code, we take the sum of dW, db across all m training examples. I’m wondering if we need to divide them by number of training example m in the end? In the standard NN we implemented before, I remember we always divide dW and db by m.


Hi @tangw Do you still need help with this?