SCLBD/BackdoorBench

How should we run a label-consistent attack?

Closed this issue a year ago · 2 comments

XaiverYuan commented a year ago

My command is python ./attack/lc_attack.py --yaml_path ../config/attack/lc/cifar10.yaml --dataset cifar10 --dataset_path --save_folder_name xxx --model resnet18
and the result here is

And I checked your git, it does not has that file either... So how should we train the LC attack?

Thank you!

kshitijsachan commented a year ago

any update on this? I would also like to run this attack

CHR-ray commented a year ago

Sorry to response late. You can find it either in the google drive link that we put in the readme, or generate by yourself with code in resource folder (https://github.com/SCLBD/BackdoorBench/tree/main/resource/label-consistent).