'Tensor'对象无法使用Keras和嵌入层以及seq2seq模型进行调用

Question

这是我使用的模型：

# Input layer of the encoder :
encoder_input = Input(shape=(None,))

# Hidden layers of the encoder :
encoder_embedding = Embedding(input_dim = num_q_words, output_dim = vec_len)(encoder_input)
encoder_dropout   = (TimeDistributed(Dropout(rate = dropout_rate)))(encoder_embedding)
encoder_LSTM      = LSTM(latent_dim, return_sequences=True)(encoder_dropout)

# Output layer of the encoder :
encoder_LSTM2_layer = LSTM(latent_dim, return_state=True)
encoder_outputs, state_h, state_c = encoder_LSTM2_layer(encoder_LSTM)

# We discard `encoder_outputs` and only keep the states.
encoder_states = [state_h, state_c]

# Input layer of the decoder :
decoder_input = Input(shape=(None,))

# Hidden layers of the decoder :
decoder_embedding_layer = Embedding(input_dim = num_a_words, output_dim = vec_len)
decoder_embedding = decoder_embedding_layer(decoder_input)

decoder_dropout_layer = (TimeDistributed(Dropout(rate = dropout_rate)))
decoder_dropout = decoder_dropout_layer(decoder_embedding)

decoder_LSTM_layer = LSTM(latent_dim, return_sequences=True)
decoder_LSTM = decoder_LSTM_layer(decoder_dropout, initial_state = encoder_states)

decoder_LSTM_2_layer = LSTM(latent_dim, return_sequences=True, return_state=True)
decoder_LSTM_2,_,_ = decoder_LSTM_2_layer(decoder_LSTM)

# Output layer of the decoder :
decoder_dense = Dense(num_a_words, activation='softmax')
decoder_outputs = decoder_dense(decoder_LSTM_2)


# `encoder_input_data` & `decoder_input_data` into `decoder_target_data`
model = Model([encoder_input, decoder_input], decoder_outputs)

但是当我想使用此代码进行推理时：

# Encoder model:
encoder_model = Model(encoder_input, encoder_states)

# Input and Input States for the Decoder:
decoder_state_input_h = Input(shape=(latent_dim,))
decoder_state_input_c = Input(shape=(latent_dim,))
decoder_states_inputs = [decoder_state_input_h, decoder_state_input_c]

# Output from the Decoder:
decoder_outputs, state_h, state_c = decoder_LSTM_layer(decoder_dropout(decoder_input), initial_state=decoder_states_inputs)
decoder_states = [state_h, state_c]
decoder_outputs = decoder_dense(decoder_outputs)

# Decoder Model:
decoder_model = Model(
    [decoder_input] + decoder_states_inputs,
    [decoder_outputs] + decoder_states)

我收到以下错误：

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-54-e293a5c0f9ee> in <module>
      8 
      9 # Output from the Decoder:
---> 10 decoder_outputs, state_h, state_c = decoder_LSTM_layer(decoder_dropout(decoder_input), initial_state=decoder_states_inputs)
     11 decoder_states = [state_h, state_c]
     12 decoder_outputs = decoder_dense(decoder_outputs)

TypeError: 'Tensor' object is not callable

但我不知道如何解决。我是这个域的新手我正在使用2.3.1版的keras和tensorflow 2.1.0版如果有人可以帮助我，我将非常感激！

Answer 1

根据您的模型代码，decoder_dropout是图层的输出：