Course 4 week 1 exercise3

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

{Moderator’s Edit: Solution Code Removed}

I don’t have any idea what should I do for weight and biases and if my code has any problem please tell me

2

Hey @Raoof2002,
Welcome and we are glad that you could be a part of our community :partying_face: Let’s statr to deal with the issues one-by-one. First of all, you haven’t taken the stride into account. Second, you have hard-coded the filter-size as 2, which is incorrect. Once you correct these out, you will see that you will be on the right track, and then, please do give another try in figuring out what should be weights and biases. Let us know if this helps.

P.S. - Posting solution code is strictly against the community guidelines. No harm done, I am removing it for now. You are more than welcome to post the error stacks, for future references.

Cheers,
Elemento

1 Like
3

Hey @saifkhanengr,
Should I remove the code now?

Cheers,
Elemento

4

Hello @Raoof2002! Let’s discuss this step by step.

First, h has to do with n_H, not n_W. Same for w. And you are not using m in a loop.

Second, vert_start and vert_end depends on what? Guess it. What is stride and f. Think about it. Same goes for horiz_start and horiz_end.

Third, while defining a_slice_prev , you know that A_prev_pad is a 4D tensor, right? So, what you are missing here?

After that, we will discuss the W and b.

Best,
Saif.

5

Thanks, Vishesh, for waiting for me to complete my response. Yes, you can remove the code.

Best,
Saif.