I did not verify the code by doing some testing. I do believe this line is same as:

"optimizer = optim.SGD([p for p in net.parameters() if p.requires_grad], lr=0.1)". Generally speaking, it only feed the optimizer the parameters which should be updated.