Decorrelated-Adversarial-Learning
Decorrelated-Adversarial-Learning copied to clipboard
about feature renorm
hi ~ i find you used norm feature in backbone , can you explain this op? And compared to un-norm feature ,is it usefully
code: renorm(2,0,1e-5).mul(1e-5)
please refer to this link
https://pytorch.org/docs/stable/generated/torch.renorm.html