W3_A1_dz1 failing

Course Q&A Deep Learning Specialization Neural Networks and Deep Learning
1

{ code removed by mentor }

2

Hello @avnish,

Please post the full error traceback and any other messages that inform you your work isn’t correct. I am removing the code for you since it is against the community guideline.

Raymond

3

Please have another look at the formula for dZ1. You are misinterpreting the last part of the formula. Here’s the full formula with slightly more clear notation about the one dot product:

dZ^{[1]} = \left ( W^{[2]T} \cdot dZ^{[2]} \right ) * g^{[1]'}(Z^{[1]})

There is no dot product multiplication involved in g^{[1]'}(Z^{[1]}): it is just the derivative of the activation function for layer 1. I would expect the code as written to throw a dimension mismatch error in the dZ1 code.

4

Paul -

Thx you very much for your helpful response. This class would not be possible without folks like you !!!

Happy New Year !!!

Avnish