我正在使用summary（）函数来获取网络配置文件并获取总参数，我遇到了我认为是错误的地方。
我没有使用任何打印功能，而我调用该功能的方式就像：“ model.summary（line_length = 36，position = [15，43，40]）”
网络A：
____________________________________
Layer (type) Output Shape
====================================
conv2d_170 (Co (None, 28, 28, 128)
____________________________________
max_pooling2d_ (None, 14, 14, 128)
____________________________________
flatten_68 (Fl (None, 25088)
____________________________________
dense_132 (Den (None, 128)
____________________________________
dense_133 (Den (None, 128)
____________________________________
dropout_68 (Dr (None, 128)
____________________________________
dense_134 (Den (None, 10)
====================================
Total params: 3,232,522
Trainable params: 3,232,522
Non-trainable params: 0
网络B：
____________________________________
Layer (type) Output Shape
====================================
conv2d_176 (Co (None, 28, 28, 128)
____________________________________
max_pooling2d_ (None, 14, 14, 128)
____________________________________
conv2d_177 (Co (None, 13, 13, 128)
____________________________________
max_pooling2d_ (None, 7, 7, 128)
____________________________________
conv2d_178 (Co (None, 6, 6, 128)
____________________________________
max_pooling2d_ (None, 3, 3, 128)
____________________________________
conv2d_179 (Co (None, 2, 2, 128)
____________________________________
max_pooling2d_ (None, 1, 1, 128)
____________________________________
flatten_71 (Fl (None, 128)
____________________________________
dense_141 (Den (None, 128)
____________________________________
dense_142 (Den (None, 128)
____________________________________
dropout_71 (Dr (None, 128)
____________________________________
dense_143 (Den (None, 10)
====================================
Total params: 234,634
Trainable params: 234,634
Non-trainable params: 0
可以预期，网络B将具有更多可训练的参数，而这不会发生。