training backdoor model

[zsaba]

hello. if i use only these commands and don't use the incremental training can i get the 98.48% attack success rate?
python IN_GE.py --hostname #exp_name --remark #remark yourself does this line automatically include the interference layer setting in your code? python trigger_generation.py python backdoored_model.py -a vgg19_bn as i used used the first line of command with incremental training and generated trigger and execute code for backdoored model but i did not get correct results. | T-top1: 42.2700 | T-top5: 100.0000 Best epoch: 267 Best acc: tensor(71.1500, device='cuda:0') for training backdoored model and when i evaluate the model it gives me | T-top1: 0.0000 | T-top5: 39.6500 Test Loss: 2.30267096, Test Acc: 10.00, Test Wm Acc: 0.00

[zsaba]

what are the exact commands that will give 98% attack success rate?

[zsaba]

/netH_epoch_17,sumloss=558.155029,Hloss=319.674500.pth" if i use model with these losses for trigger generation is it good?