W4A2 Neural Style Transfer - train_step function

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

I have some trouble to solve exercise 6 in Neural Style Transfer assignment.
I have completed filling the blanks in train_step function,

W4A2 exercise6-2
W4A2 exercise6-2967×129 1.85 KB

and executing the above cell results in the following error.
W4A2 exercise6
W4A2 exercise6890×671 12.6 KB

It says ‘You always must run the last cell before this one.’, but what is the last cell?

2

It means “the previous cell”.

3

And, perhaps these instructions for compute_layer_style_cost() will be useful:

image
image868×103 3.38 KB

4

Thank you for your reply, but the error remains unsolved.
I think my codes written in train_step() function are quite straightforward and correct, but it still doesn’t work and gives the same error message.

5

Did you modify the code in compute_layer_style_cost() and then re-run all of the cells?

6

Problem solved!
The problem was that I used the following code in compute_layer_style_cost function
_, n_H, n_W, n_C = tf.shape(a_G),
instead of using X.get_shape().as_list().

Again, thank you for your accurate and quick reply!

7

That’s good news.

1 Like