one_shot_learning_network.py 文件源码

def __call__(self, inputs, name, training=False):
        """
        Runs the bidirectional LSTM, produces outputs and saves both forward and backward states as well as gradients.
        :param inputs: The inputs should be a list of shape [sequence_length, batch_size, 64]
        :param name: Name to give to the tensorflow op
        :param training: Flag that indicates if this is a training or evaluation stage
        :return: Returns the LSTM outputs, as well as the forward and backward hidden states.
        """
        with tf.name_scope('bid-lstm' + name), tf.variable_scope('bid-lstm', reuse=self.reuse):
            with tf.variable_scope("encoder"):
                fw_lstm_cells_encoder = [rnn.LSTMCell(num_units=self.layer_sizes[i], activation=tf.nn.tanh)
                                 for i in range(len(self.layer_sizes))]
                bw_lstm_cells_encoder = [rnn.LSTMCell(num_units=self.layer_sizes[i], activation=tf.nn.tanh)
                                 for i in range(len(self.layer_sizes))]



                outputs, output_state_fw, output_state_bw = rnn.stack_bidirectional_rnn(
                    fw_lstm_cells_encoder,
                    bw_lstm_cells_encoder,
                    inputs,
                    dtype=tf.float32
                )
            print("out shape", tf.stack(outputs, axis=0).get_shape().as_list())
            with tf.variable_scope("decoder"):
                fw_lstm_cells_decoder = [rnn.LSTMCell(num_units=self.layer_sizes[i], activation=tf.nn.tanh)
                                         for i in range(len(self.layer_sizes))]
                bw_lstm_cells_decoder = [rnn.LSTMCell(num_units=self.layer_sizes[i], activation=tf.nn.tanh)
                                         for i in range(len(self.layer_sizes))]
                outputs, output_state_fw, output_state_bw = rnn.stack_bidirectional_rnn(
                    fw_lstm_cells_decoder,
                    bw_lstm_cells_decoder,
                    outputs,
                    dtype=tf.float32
                )


        self.reuse = True
        self.variables = tf.get_collection(tf.GraphKeys.TRAINABLE_VARIABLES, scope='bid-lstm')
        return outputs, output_state_fw, output_state_bw