Week 4 Assignment 3 Exercise 1

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

In the assignment → Programming Assignment: Deep Neural Network - Application,
The compute_cost function is returning Nan value.

Used command is -
cost = compute_cost(A2, Y)

Output is below
Cost after iteration 1: nan

2

Please click my name and message your notebook as an attachment.

3

This is the file name.
/notebooks/release/W4A2/Deep Neural Network - Application.ipynb

Not sure how to find the link click my name and message in notebook.

4

File > Download > Download .ipynb and message the notebook as an attachment. Here’s the community faq to get started.

5

Are you sure that you are taking the logarithm of the A values and not the Y values?

If you got that code correct, then the other possibility is that your A values are wrong. Do you mean that error message happened in the test for the compute_cost function or one of the “model” functions?

Also note that there are only 2 assignments in DLS C1 W4, so the title is misleading.

6

PFA the file attached for your reference.

{moderator edit - solution notebook removed}

7

Your forward and back propagation logic for two_layer_model is incorrect. In both directions, you have used “relu” as the activation at both layers. That’s not the way the network is configured, right?

8

Actually that explains the NaN from the cost: your output A2 should only include values between 0 and 1, but with ReLU as the output activation that will not be true. That causes you to take the log of a negative number in the Y = 0 term in the loss function.

9

Thank you for your clarifications. I could rectify the error based on your guidance.