Keras 模型输出形状为“（无，）”

Question

我的模型包含一个先前加载的模型，并给出“(None,)”的输出形状：

from tensorflow.keras.models import Sequential, Model
from tensorflow.keras.layers import Activation, Dense, Input, Subtract, Multiply, Lambda

x = Input((158,))
y = model(x)
c = Subtract()([x,y])
c = Multiply()([c,c])
d = Lambda(lambda arg: tf.keras.backend.mean(arg,axis=1), output_shape = (None,1))
e = d(c)

new_model = Model(inputs = x, outputs = e)
new_model.summary()

Model: "model"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
input_1 (InputLayer)            [(None, 158)]        0                                            
__________________________________________________________________________________________________
model_1 (Model)                 (None, 158)          57310       input_1[0][0]                    
__________________________________________________________________________________________________
subtract (Subtract)             (None, 158)          0           input_1[0][0]                    
                                                                 model_1[1][0]                    
__________________________________________________________________________________________________
multiply (Multiply)             (None, 158)          0           subtract[0][0]                   
                                                                 subtract[0][0]                   
__________________________________________________________________________________________________
lambda (Lambda)                 (None,)              0           multiply[0][0]                   
==================================================================================================
Total params: 57,310
Trainable params: 57,310
Non-trainable params: 0
__________________________________________________________________________________________________

这个模型输出正确的值，但它可能会在我的下一步工作中产生问题，所以我想知道这个输出形状意味着什么，如果我必须纠正它（因为我没有看到这样的例子案例在线）。

编辑

具体来说，我不是在调查None值，但事实上它没有说(None,1) ，这是同一件事吗？

例如，这个总结：

Layer (type)                 Output Shape              Param #
=================================================================
dense_1 (Dense)              (None, 2)                 4
_________________________________________________________________
dense_2 (Dense)              (None, 1)                 3
=================================================================
Total params: 7
Trainable params: 7
Non-trainable params: 0
_________________________________________________________________

来源： https : //machinelearningmastery.com/visualize-deep-learning-neural-network-model-keras/

Answer 1

这里没有代表您的batch size 。 批量大小值是动态的，您稍后在.fit()期间定义它，因此在定义之前它不知道大小并且它保持None表示任何正整数值。

您可以阅读此处以更好地理解参数和值。

Answer 2

我设法将最后一层重新整形为(None,1) ，它确实解决了我的代码中的一个问题，我通过向我的模型添加一个Reshape层来做到这一点：

x = Input(158,)
y = model(x)
c = Subtract()([x,y])
c = Multiply()([c,c])
d = Lambda(lambda arg: tf.keras.backend.mean(arg,axis=1), output_shape = (None,1))
e = d(c)
f = Reshape([1])(e)

new_model = Model(inputs = x, outputs = f)

这使：

new_model.summary()

Model: "model_4"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
input_5 (InputLayer)            [(None, 158)]        0                                            
__________________________________________________________________________________________________
model_1 (Model)                 (None, 158)          57310       input_5[0][0]                    
__________________________________________________________________________________________________
subtract_4 (Subtract)           (None, 158)          0           input_5[0][0]                    
                                                                 model_1[5][0]                    
__________________________________________________________________________________________________
multiply_4 (Multiply)           (None, 158)          0           subtract_4[0][0]                 
                                                                 subtract_4[0][0]                 
__________________________________________________________________________________________________
lambda_4 (Lambda)               (None,)              0           multiply_4[0][0]                 
__________________________________________________________________________________________________
reshape_3 (Reshape)             (None, 1)            0           lambda_4[0][0]                   
==================================================================================================
Total params: 57,310
Trainable params: 57,310
Non-trainable params: 0
__________________________________________________________________________________________________