您正在运行一个期望张量为NumPy ndarray的操作.不幸的是，这是行不通的.您只需要使用 张量运算符来重新实现自定义操作.

You are running an operation which expects a NumPy ndarray on tensors. Unfortunately this will not work. You need to reimplement the custom operation using only tensor operators.

说过，直接使用Tensorflow中的函数也可以，例如import tensorflow中的函数，并在自定义层中使用这些函数，可能会比单独提供Keras后端给您更多的功能.

Having said that, using functions from Tensorflow directly is OK as well, say from import tensorflow and use those inside a custom layer might give you more functions than Keras backend alone.