See: What are the advantages of ReLU over sigmoid function in deep neural networks? The choice to use tanh as a default is likely more about ...
確定! 回上一頁