Greetings!!
I am trying to complete the Neural Machine Translation programming assignment that is a part of the DLS C5 (sequence models), W3.
I see that my code is passing all the tests. But, Exercise 2 - modelf, is getting failed by the grader with the following message:
Code Cell UNQ_C1: Function 'one_step_attention' is correct.
Code Cell UNQ_C2: Unexpected error (ValueError('Input 0 of layer dense is incompatible with the layer: expected axis -1 of input shape to have value 128 but received input with shape [None, 30, 320]')) occurred during function check. We expected function `modelf` to return type <class 'bool'>. Please check that this function is defined properly.
If you see many functions being marked as incorrect, try to trace back your steps & identify if there is an incorrect function that is being used in other steps.
This dependency may be the cause of the errors.
Two other things I noticed:
- Model summary is different as you can see from below:
Model: "functional_22"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_10 (InputLayer) [(None, 30, 37)] 0
__________________________________________________________________________________________________
s0 (InputLayer) [(None, 64)] 0
__________________________________________________________________________________________________
bidirectional_9 (Bidirectional) (None, 30, 64) 17920 input_10[0][0]
__________________________________________________________________________________________________
repeat_vector (RepeatVector) (None, 30, 64) 0 s0[0][0]
lstm[110][1]
lstm[111][1]
lstm[112][1]
lstm[113][1]
lstm[114][1]
lstm[115][1]
lstm[116][1]
lstm[117][1]
lstm[118][1]
__________________________________________________________________________________________________
concatenate (Concatenate) (None, 30, 128) 0 bidirectional_9[0][0]
repeat_vector[110][0]
bidirectional_9[0][0]
repeat_vector[111][0]
bidirectional_9[0][0]
repeat_vector[112][0]
bidirectional_9[0][0]
repeat_vector[113][0]
bidirectional_9[0][0]
repeat_vector[114][0]
bidirectional_9[0][0]
repeat_vector[115][0]
bidirectional_9[0][0]
repeat_vector[116][0]
bidirectional_9[0][0]
repeat_vector[117][0]
bidirectional_9[0][0]
repeat_vector[118][0]
bidirectional_9[0][0]
repeat_vector[119][0]
__________________________________________________________________________________________________
dense (Dense) (None, 30, 10) 1290 concatenate[110][0]
concatenate[111][0]
concatenate[112][0]
concatenate[113][0]
concatenate[114][0]
concatenate[115][0]
concatenate[116][0]
concatenate[117][0]
concatenate[118][0]
concatenate[119][0]
__________________________________________________________________________________________________
dense_1 (Dense) (None, 30, 1) 11 dense[110][0]
dense[111][0]
dense[112][0]
dense[113][0]
dense[114][0]
dense[115][0]
dense[116][0]
dense[117][0]
dense[118][0]
dense[119][0]
__________________________________________________________________________________________________
attention_weights (Activation) (None, 30, 1) 0 dense_1[110][0]
dense_1[111][0]
dense_1[112][0]
dense_1[113][0]
dense_1[114][0]
dense_1[115][0]
dense_1[116][0]
dense_1[117][0]
dense_1[118][0]
dense_1[119][0]
__________________________________________________________________________________________________
dot (Dot) (None, 1, 64) 0 attention_weights[110][0]
bidirectional_9[0][0]
attention_weights[111][0]
bidirectional_9[0][0]
attention_weights[112][0]
bidirectional_9[0][0]
attention_weights[113][0]
bidirectional_9[0][0]
attention_weights[114][0]
bidirectional_9[0][0]
attention_weights[115][0]
bidirectional_9[0][0]
attention_weights[116][0]
bidirectional_9[0][0]
attention_weights[117][0]
bidirectional_9[0][0]
attention_weights[118][0]
bidirectional_9[0][0]
attention_weights[119][0]
bidirectional_9[0][0]
__________________________________________________________________________________________________
c0 (InputLayer) [(None, 64)] 0
__________________________________________________________________________________________________
lstm (LSTM) [(None, 64), (None, 33024 dot[110][0]
s0[0][0]
c0[0][0]
dot[111][0]
lstm[110][1]
lstm[110][2]
dot[112][0]
lstm[111][1]
lstm[111][2]
dot[113][0]
lstm[112][1]
lstm[112][2]
dot[114][0]
lstm[113][1]
lstm[113][2]
dot[115][0]
lstm[114][1]
lstm[114][2]
dot[116][0]
lstm[115][1]
lstm[115][2]
dot[117][0]
lstm[116][1]
lstm[116][2]
dot[118][0]
lstm[117][1]
lstm[117][2]
dot[119][0]
lstm[118][1]
lstm[118][2]
__________________________________________________________________________________________________
dense_2 (Dense) (None, 11) 715 lstm[110][1]
lstm[111][1]
lstm[112][1]
lstm[113][1]
lstm[114][1]
lstm[115][1]
lstm[116][1]
lstm[117][1]
lstm[118][1]
lstm[119][1]
==================================================================================================
Total params: 52,960
Trainable params: 52,960
Non-trainable params: 0
_____________________________________________________________________________________
As you can see from the attached screenshot, the accuracy is somewhat low:
Also, it translated the following input entry:
source: 21th of August 2016
output: 2016-08-20
I am unable to figure what the issues are. please help me resolve this.