kohpangwei/group_DRO

not being able to reproduce your results on CUB and CelebA

Closed this issue · 2 comments

With the command in repo for MNLI, I am not able to reproduce your results.
Could you send me the command which reproduce your best results on CUB and CeleBA?
Especially for that with Large Weight Decay and Early stop
Thanks a lot.

Hi! We provide sample commands for CUB and CelebA in the README as well (in addition to the command for MNLI). For the exact commands used in our experiments, please see
https://worksheets.codalab.org/worksheets/0x621811fe446b49bb818293bae2ef88c0

Thanks