C4W1 Assignment - Exercise 3 Decoder Function

[roarmarketing]

Hello!

Please see the unit test output error message I received for the ‘Decoder’ function exercise. Any hints or direction would be greatly appreciated! Thanks in advance…

[paulinpaloalto]

Here’s the output of that function with instrumentation to show the shapes at each internal step. That should help you to figure out where things go off the rails.

dec1
target.shape (64, 15)
dec2
x.shape (64, 15, 256)
dec3
x.shape (64, 15, 256)
ca1
ca2
ca3
ca4
dec4
x.shape (64, 15, 256)
dec5
x.shape (64, 15, 256)
dec6
logits.shape (64, 15, 12000)
Tensor of contexts has shape: (64, 14, 256)
Tensor of right-shifted translations has shape: (64, 15)
Tensor of logits has shape: (64, 15, 12000)

[roarmarketing]

Thanks so much!!

[roarmarketing]

I discovered my issue. I had return_sequences = False for the post_attention RNN. Once I changed it to True, the unit test worked.