Course 2 Week 3 assignment, Ex6

Course Q&A Deep Learning Specialization Improving Deep Neural Networks: Hyperparameter tun
1

I have tried to compute the cost, I can get correct type but not the correct value. What should I do?
{Moderator Edit: Solution Code Removed}

image
image1312×494 23.1 KB

2

Hint: Inputs of tf.keras.losses.categorical_crossentropy are expected to be of shape (number of examples, num_classes)

Given that:

logits -- output of forward propagation (output of the last LINEAR unit), of shape (6, num_examples)
labels -- "true" labels vector, same shape as Z3

But, why you are dividing total loss by logits.shape?

1 Like
3

thank you, because I am thinking maybe taking the mean average will fit into the answer.