C4W4 layer style cost error

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

Hi, I’m getting the following error:

InvalidArgumentError Traceback (most recent call last)
in
2 a_S = tf.random.normal([1, 4, 4, 3], mean=1, stddev=4)
3 a_G = tf.random.normal([1, 4, 4, 3], mean=1, stddev=4)
----> 4 J_style_layer_GG = compute_layer_style_cost(a_G, a_G)
5 J_style_layer_SG = compute_layer_style_cost(a_S, a_G)
6

in compute_layer_style_cost(a_S, a_G)
22
23 # Computing gram_matrices for both images S and G (≈2 lines)
—> 24 GS = gram_matrix(a_S)
25 GG = gram_matrix(a_G)
26

in gram_matrix(A)
13
14 #(≈1 line)
—> 15 GA = tf.matmul(A,tf.transpose(A))

and at the end:

InvalidArgumentError: In[0] mismatch In[1] shape: 16 vs. 3: [1,3,16] [16,3,1] 0 0 [Op:BatchMatMulV2]

I’m probably doing something wrong with reshaping the images, but don’t know what.
My code is looks like this:

tensor = tf.transpose(tf.reshape(tensor,shape=[1,-1,n_C]), perm=[0,2,1])

Can You help me with this?

2

These are the dimensions: m=1, n_H=n_W=4 and n_C=3.
a_S should have shape n_C, n_H * n_W i.e. (3, 16). Please confirm at your end.

3

Yes, these are the dimensions. I’m reshaping both a_S and a_G in the same way.

4

You have an additional dimension. The shape of a_S according to your code is (1,3,16) and not (3,16)

5

The values you’re passing as the shape [1, -1, n_C] are not correct.

6

Thank You, everything works now. Problem solved.