A small gradient means that the weights and biases of the initial layers will not be updated effectively with each training session.
確定! 回上一頁