C5_W4_A1 Ex3 Wrong masked weights

Course Q&A Deep Learning Specialization Sequence Models
1

Hi,

My code it is working but I think may be an approximation error, you can see the photo below:

image
image1080×351 69 KB

I’ve checked the code and I really don’t know how to rewrite it in other way.

Thanks in advance

2

Solved! Just forgot to add a -1 to the mask.

3

Happened to have the same problem.

scaled_attention_logits += (1-mask)*-1e9