C5_W4_A1 scaled_dot_product_attention "wrong unmasked weights"

Course Q&A Deep Learning Specialization Sequence Models
1

I’m getting an error of “Wrong unmasked weights”. But don’t know where I did wrong.
I used:
dk = tf.shape(k)[-2]
(1 - mask) * (-1.0e9)
tf.nn.softmax
tf.matmul
Can someone help? Thanks!

2

Please click my name and message your function code as text.

3

The problem is with calculating matmul_qk. Pay attention to the transpose operation in Q K^T

4

It works! thank you!