lstm_out=Bidirectional(LSTM(128, return_sequences=True,name="lstm1"))(x) ###⚠️为什么原本是64，作者这里变成128
lstm_out = LSTM(128,activation='relu')(lstm_out)
print(lstm_out.shape)
l_att = AttLayer()(lstm_out)
output = Dense(1, activation='sigmoid')(l_att)
print(output.shape)
我在中间加了一层LSTM后基本只有断言错误了
2020-04-08
雪球
