pytorch version of DAN loss : I dont't find betals param of multi guassian_kernels

thuml / Xlearn

Transfer Learning Library

465 stars 155 forks source link

pytorch version of DAN loss : I dont't find betals param of multi guassian_kernels #23

Open cch2016 opened 5 years ago

cch2016 commented 5 years ago

In " loss.py " about ,"DAN loss ", I dont't find bata parameter of multi -guassian_kernels and dont't known how it update? Can someone tell me? Thanks.

cch2016 commented 5 years ago

I just see "sum" , mean just "sum" and dont't have "beta " parameter

ChengYeung1222 commented 5 years ago

beta=[1,1,...,1]

hupech commented 4 years ago

In " loss.py " about ,"DAN loss ",it use MK-MMD? why i think it only use MMD In a certain layer. who can tell me?