创建RNN的关注层时，无法调用“模块”对象

Question

我正在尝试实现RNN。这是我的代码：

def attention_layer(inputs, layer):
    attention = layer(inputs)
    attention = Dense(1, use_bias=False)(attention)
    attention = Flatten()(attention)
    attention = Activation('softmax')(attention)
    attention = RepeatVector(params['rnn_dim']*2)(attention)
    attention = Permute ([ 2 , 1 ]) (attention)
    # apply the attention
    sent_representation = merge([inputs, attention], mode='mul')
    sent_representation = Lambda(lambda xin: K.sum(xin, axis=1) 
    (sent_representation))
    # print sent_representation.get_shape()
    return sent_representation

问题：

当我尝试将其合并时发生错误：

  

'模块'对象不可调用

有人可以帮我定义一下吗？