But for any custom operation that has trainable weights, you should implement your own layer. The example below illustrates the skeleton of a Keras custom layer ...
確定! 回上一頁