python-3.x - Keras 功能 API 比 Sequential 慢 / 没有改进

Question

已解决！（必须在顺序模型中设置 trainable=true）

我目前正在将我的 Keras 模型从 Sequential 更改为功能 API。虽然 Sequential 模型在 10 个 epoch 之后确实提高了 1 的准确度，但功能性 API 模型甚至没有达到 0.7 并且没有进一步提高。除了输入层，两个网络应该是相同的。

顺序：

model = Sequential()
model.add(Embedding(20000, 256,input_length = 30))
model.add(SpatialDropout1D(0.4))
model.add(LSTM(256, dropout=0.3, recurrent_dropout=0.3))
model.add(Dense(1,activation='sigmoid'))
model.compile(loss = 'binary_crossentropy', optimizer=Adam(lr=0.0001),metrics = ['accuracy'])
print(model.summary())

输出是：

    _________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
embedding_6 (Embedding)      (None, 30, 256)           5120000   
_________________________________________________________________
spatial_dropout1d_5 (Spatial (None, 30, 256)           0         
_________________________________________________________________
lstm_5 (LSTM)                (None, 256)               525312    
_________________________________________________________________
dense_6 (Dense)              (None, 1)                 257       
=================================================================
Total params: 5,645,569
Trainable params: 5,645,569
Non-trainable params: 0
_________________________________________________________________
None

对于功能 API：

inputs = Input(shape=(31,))
embed = Embedding(20000, 256, trainable=False)(inputs)
drop = (SpatialDropout1D(0.4))(embed)
lstm = LSTM(256, dropout=0.3, recurrent_dropout=0.3)(drop)
acti = Dense(1,activation='sigmoid')(lstm)
model = Model(inputs=inputs, outputs=acti)
model.compile(loss = 'binary_crossentropy', optimizer=Adam(lr=0.0001),metrics = ['accuracy'])
print(model.summary())

结果

Model: "model_5"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_8 (InputLayer)         (None, 31)                0         
_________________________________________________________________
embedding_7 (Embedding)      (None, 31, 256)           5120000   
_________________________________________________________________
spatial_dropout1d_6 (Spatial (None, 31, 256)           0         
_________________________________________________________________
lstm_6 (LSTM)                (None, 256)               525312    
_________________________________________________________________
dense_7 (Dense)              (None, 1)                 257       
=================================================================
Total params: 5,645,569
Trainable params: 525,569
Non-trainable params: 5,120,000
_________________________________________________________________
None

我是否监督了某些事情，或者有人可以解释我的结果吗？

标签： python-3.xkeraslstm