ValueError: Shape must be rank 1 but is rank 0 for 'bn_conv1/cond/Reshape_4' (op: 'Reshape') with input shapes: [1,64,1,1], [].

運行demo報錯

ValueError: Shape must be rank 1 but is rank 0 for 'bn_conv1/cond/Reshape_4' (op: 'Reshape') with input shapes: [1,64,1,1], [].

經過查閱，發現是BatchNormalization的問題，如下代碼也會類似報錯

from keras.layers import BatchNormalization, Input

 x = Input(shape=(1, 2, 2)) 

BatchNormalization(axis=1)(x)   

報錯：InvalidArgumentError: Shape must be rank 1 but is rank 0 for 'batch_normalization_1/cond/Reshape_4' (op: 'Reshape') with input shapes: [1,1,1,1], [].

在CPU版的keras 2.2.0上沒有問題，在gup版本的keras有問題。因此將keras 降低版本為2.1.2：

 pip3 uninstall keras

 pip3 install keras==2.1.2  -i http://pypi.douban.com/simple --trusted-host pypi.douban.com

ref：https://blog.csdn.net/weixin_40755306/article/details/84944008