I want to create a Keras model consisting of an embedding layer, followed by two LSTMs with dropout 0.5, and lastly a dense layer with a softmax activation.
The first LSTM should propagate the sequential output to the second layer, while in the second I am only interested in getting the hidden state of the LSTM after processing the whole sequence.
I tried the following:
sentence_indices = Input(input_shape, dtype = 'int32')
embedding_layer = pretrained_embedding_layer(word_to_vec_map, word_to_index)
embeddings = embedding_layer(sentence_indices)
# Propagate the embeddings through an LSTM layer with 128-dimensional hidden state
X = LSTM(128, return_sequences=True, dropout = 0.5)(embeddings)
# Propagate X trough another LSTM layer with 128-dimensional hidden state
X = LSTM(128, return_sequences=False, return_state=True, dropout = 0.5)(X)
# Propagate X through a Dense layer with softmax activation to get back a batch of 5-dimensional vectors.
X = Dense(5, activation='softmax')(X)
# Create Model instance which converts sentence_indices into X.
model = Model(inputs=[sentence_indices], outputs=[X])
However I am getting the following error:
ValueError: Layer dense_5 expects 1 inputs, but it received 3 input tensors. Input received: [<tf.Tensor 'lstm_10/TensorArrayReadV3:0' shape=(?, 128) dtype=float32>, <tf.Tensor 'lstm_10/while/Exit_2:0' shape=(?, 128) dtype=float32>, <tf.Tensor 'lstm_10/while/Exit_3:0' shape=(?, 128) dtype=float32>]
Clearly LSTM is not returning an output of the shape I expect. How do I fix this?
