Course 1 Week 4 Exercise 2 - initialize_parameters_deep

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

Error: Wrong output for variable W1.
Error: Wrong output for variable W2.

I followed the same pattern

 parameters["W" + str(L)] = np.random.randn(layer_dims[1], layer_dims[0]) * 0.01

just used (l,l-1) for shape W
But don’t know where I am wrong

2

It is a mistake to “hand import” your solutions from the Step by Step exercise. It turns out they supply their own implementations and they end up using a different more sophisticated initialization algorithm for the “deep” case. It just turns out that the simplistic version they had us build in the first exercise in Week 4 just doesn’t work very well in this case. We’ll learn more about that in Course 2 of this series, but there’s too much other new material here in Course 1 to get into that.

3

I tried to do according to for loop. Could you please assist further. I know there is a small mistake to handle…

4

Please say more about which problem you are working on? Are you doing the “Step by Step” exercise? Or the “Application” exercise? And which function?

5

Step by Step exercise.

initialize_parameters_deep Function

just use “np.random.rand(layer_dims[l],layer_dims[l-1]) * 0.01” in for loop

but still getting wrong W1 & W2 values

6

You are using the wrong random function. They literally wrote the code for you in the instructions. It uses “randn”, not “rand”, right?

2 Likes
7

thanks :), silly mistake.

8

I am not getting whats my error please help @paulinpaloalto

{moderator edit - solution code removed}

9

There is a missing close paren on the previous line. I did not have this problem, so you must have accidentally deleted that close parenthesis.

10

Ohh yes I see again a silly mistake …Thank you for your fast reply!!!