Bias set to fals in super-resolution models

Hello,

I’m quite not understanding this concept here in super-resolution models. Most of the implementation on GitHub they are asserting bias as false even when batch norm is false.

Do you have any explanation for this ?