0

I'm very new to DL and I've been trying to use a seq2seq model to classify text (sentiment analysis) from this repo. The dataset I've used is amazon review polarity (first 2000 rows).Data-set basically consists of labels and corresponding text. My model is as follows:

sequence_input = Input(shape=(MAX_SEQUENCE_LENGTH,), dtype='int32') #MAX_SEQUENCE_LENGTH = 1000
embedded_sequences = embedding_layer(sequence_input)
l_gru = Bidirectional(GRU(100, return_sequences=True))(embedded_sequences)
l_att = AttLayer()(l_gru)
preds = Dense(2, activation='softmax')(l_att)
model = Model(sequence_input, preds)

model.compile(optimizer='rmsprop', loss='sparse_categorical_crossentropy',metrics=['accuracy'])
print("model fitting - attention GRU network")
model.summary()
model.fit(x_train, y_train, validation_data=(x_val, y_val),
          epochs=5,verbose = 1, batch_size=50)

model.save('s2s.h5')

Output:

model fitting - attention GRU network
Model: "model_23"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_24 (InputLayer)        (None, 1000)              0         
_________________________________________________________________
embedding_11 (Embedding)     (None, 1000, 100)         1276800   
_________________________________________________________________
bidirectional_24 (Bidirectio (None, 1000, 200)         120600    
_________________________________________________________________
att_layer_24 (AttLayer)      (None, 1000, 200)         200       
_________________________________________________________________
dense_23 (Dense)             (None, 1000, 2)           402       
=================================================================
Total params: 1,398,002
Trainable params: 1,398,002
Non-trainable params: 0

---------------------------------------------------------------------------

ValueError                                Traceback (most recent call last)

<ipython-input-112-ec37b13f1d7e> in <module>()
      1 model.fit(x_train, y_train, validation_data=(x_val, y_val),
----> 2           epochs=5,verbose = 1, batch_size=50)
      3 
      4 model.save('s2s.h5')

2 frames

/usr/local/lib/python3.6/dist-packages/keras/engine/training_utils.py in standardize_input_data(data, names, shapes, check_batch_axis, exception_prefix)
    129                         ': expected ' + names[i] + ' to have ' +
    130                         str(len(shape)) + ' dimensions, but got array '
--> 131                         'with shape ' + str(data_shape))
    132                 if not check_batch_axis:
    133                     data_shape = data_shape[1:]

ValueError: Error when checking target: expected dense_22 to have 3 dimensions, but got array with shape (1600, 2)

Dimension of test and validation data set:

print(x_train.shape)
print(x_val.shape)
print(y_train.shape)
print(y_val.shape)

Output:

(1600, 1000)
(400, 1000)
(1600, 2)
(400, 2)

I also referred other similar questions like: this. But couldn't find any leads.I am ready to provide more details on my implementation if these don't sufffice. Thanks in advance.

Aivi
  • 1
  • 3

1 Answers1

0

After experimenting around a little, I realised that I'd been trying to use a 2D input while the actual code was using 3D input. I referred this question which had an almost similar query and the solution to my query.

Aivi
  • 1
  • 3