Course5W4 Transformer, final layer, Wrong values in translation


Hi, I am little confused about the implementation of the last layer. It supposed to be easy and straightforward to implement it by following the introduction. But I get error above. I checked my implementation and introduction but still can’t find the reason. Thanks for your help in advance.

Hey @Carrie_Young,
Can you please check your DM?

Cheers,
Elemento

Okay, I have checked it. Thank you.

Hey @Carrie_Young,
In your equation for enc_output, in your implementation of Transformer, you have used dec_padding_mask, whereas, you are supposed to use enc_padding_mask. I suppose the reason behind it is trivial. Let us know if this resolves your issue.

Cheers,
Elemento

Thank you. The problem is solved now.