zhangcheng-007

I want to know why you cancel Gaussian fuzzy logic. If I'm not mistaken, the paper will train y-channel network (YCrCb) according to srcnn. Thank you for your help.

代码中的tanh_beta能够实现论文中分段线性函数app的效果吗？或者说fc9与fc10层输出的组合，我不太明白fc10层输出的用途。