xiangjjj / implicit_alignment

Code for ICML2020 "Implicit Class-Conditioned Domain Alignment for Unsupervised Domain Adaptation"

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Setting

huang123ying opened this issue · comments

when i run the MDD+implicit,should i set the args.self_training==True?

Yes, please set args.self_train=True. This argument is designed exclusively for the target domain so that it can collect pseudo-labels and use them for sampling. More detail please refer to here. BTW, I noticed that I forgot to include it in the readme, but it is in my makefile scripts, such as this. Please let me know if this answers you question. Thanks!

I've just updated the readme to correct this. Please refer to the makefile scripts for the configuration of different datasets.

thank you very much!