C5W4A1 scaled_dot_product_attention error

Course Q&A Deep Learning Specialization Sequence Models
1

Hello, I am trying to finish Ex3 of the assignment. The test result shows that I got the “Wrong masked weights”. The error is shown below. It seems that the assert result is part of my result. What is my problem? Could anyone please tell me the reason? Thanks a lot.
image|690x354

2

Same problem. My weight is:
tf.Tensor(
[[0.2039163 0.33620113 0.33620113 0.12368149]
[0.18780005 0.18780005 0.51049346 0.11390649]
[0.2772748 0.2772748 0.2772748 0.16817567]], shape=(3, 4), dtype=float32)

3

Hey guys,

make sure you multiplied the matrix values by -1e9 before applying the softmax as it mentioned in the assignment.

@ChaseNG @yum068

3 Likes