Question

当我像这样在Keras中创建模型时：

from keras.models import Model

my_model = model_building_function() # returns a Keras Model
inputs = keras.layers.Input(shape=(None, None, 3))
outputs = my_model(inputs)
my_model = Model(inputs, outputs)

my_model的摘要（使用Model.summary()）将导致如下结果：

__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
input (InputLayer)            (None, None, None, 3)  0                                                                    
__________________________________________________________________________________________________
model_1 (Model)                 multiple             23561152    input[0][0]                                                                                            
__________________________________________________________________________________________________

==================================================================================================
Total params: 23,561,152
Trainable params: 23,454,912
Non-trainable params: 106,240
__________________________________________________________________________________________________

my_model（keras.utils.plot_model）的绘制图也是如此。

有什么办法可以解开模型，使各个图层可见？

Answer 1

您可以获取模型/图层的layers属性，如果是模型，则可以打印其摘要。

def sub_models(layer):

    if isinstance(layer, Model):
        print('summary for model ' + layer.name)
        layer.summary()

    if hasattr(layer, 'layers'):
        for l in layer.layers:
            sub_models(l)

sub_models(my_model)

Keras：解开模型进行绘图/汇总

1 个答案: