-
Notifications
You must be signed in to change notification settings - Fork 7
/
run.sh
38 lines (21 loc) · 2.45 KB
/
run.sh
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
# Plain Training
# python train.py --mode 0 --att 0 --train_mode 0 --test_mode 0 --zero_norm 0 --save_dir "model_chkpt_new/chkpt_scaled/" --gpu "0,1,2,3" &
# python train.py --mode 0 --att 0 --train_mode 0 --test_mode 0 --zero_norm 1 --save_dir "model_chkpt_new/chkpt_norm/" --gpu "4,5,6,7"
# python train.py --mode 0 --att 1 --train_mode 2 --test_mode 0 --train_max_iter 7 --load_adv_dir "model_chkpt_new/chkpt_plain/" --save_dir "model_chkpt_new/chkpt_pgd7/" --gpu "1,3"
# Adversarial Training
# (NORM :: NO_RETAIN :: FGSM)
python train.py --mode 0 --att 1 --train_mode 2 --test_mode 0 --train_max_iter 1 --zero_norm 1 --store_adv 0 --load_adv_dir "model_chkpt_new/chkpt_norm/" --save_dir "model_chkpt_new/chkpt__norm__noretain__pgd1/" --gpu "0,1" &
# # (NORM :: NO_RETAIN :: PGD7)
python train.py --mode 0 --att 1 --train_mode 2 --test_mode 0 --train_max_iter 7 --zero_norm 1 --store_adv 0 --load_adv_dir "model_chkpt_new/chkpt_norm/" --save_dir "model_chkpt_new/chkpt__norm__noretain__pgd7/" --gpu "2,3" &
# # (NORM :: RETAIN :: FGSM)
python train.py --mode 0 --att 1 --train_mode 2 --test_mode 0 --train_max_iter 1 --zero_norm 1 --store_adv 1 --load_adv_dir "model_chkpt_new/chkpt_norm/" --save_dir "model_chkpt_new/chkpt__norm__retain__pgd1/" --gpu "4,5" &
# # (NORM :: RETAIN :: PGD7)
python train.py --mode 0 --att 1 --train_mode 2 --test_mode 0 --train_max_iter 7 --zero_norm 1 --store_adv 1 --load_adv_dir "model_chkpt_new/chkpt_norm/" --save_dir "model_chkpt_new/chkpt__norm__retain__pgd7/" --gpu "6,7" &
# (NORM :: NO_RETAIN :: FGSM)
# python train.py --mode 0 --att 1 --train_mode 2 --test_mode 0 --train_max_iter 1 --store_adv 0 --load_adv_dir "model_chkpt_new/chkpt_scaled/" --save_dir "model_chkpt_new/chkpt__scaled__noretain__pgd1/" --gpu "0,1" &
# (NORM :: NO_RETAIN :: PGD7)
python train.py --mode 0 --att 1 --train_mode 2 --test_mode 0 --train_max_iter 7 --store_adv 0 --load_adv_dir "model_chkpt_new/chkpt_scaled/" --save_dir "model_chkpt_new/chkpt__scaled__noretain__pgd7/" --gpu "2,3"
# (NORM :: RETAIN :: FGSM)
# python train.py --mode 0 --att 1 --train_mode 2 --test_mode 0 --train_max_iter 1 --store_adv 1 --load_adv_dir "model_chkpt_new/chkpt_scaled/" --save_dir "model_chkpt_new/chkpt__scaled__retain__pgd1/" --gpu "4,5" &
# (NORM :: RETAIN :: PGD7)
# python train.py --mode 0 --att 1 --train_mode 2 --test_mode 0 --train_max_iter 7 --store_adv 1 --load_adv_dir "model_chkpt_new/chkpt_scaled/" --save_dir "model_chkpt_new/chkpt__scaled__retain__pgd7/" --gpu "6,7"