Combine parameter sharing and independent in tensorflow

Question

I want part of layers are independent,and other layers are weight sharing. For Example:

\'\'\' def generator(self, inputs, reuse=False, scope=\'g_net\'): n, h, w, c =