train.sh

# -------- architecture -------------------------
# arch=OMPa
# arch=OMPb
arch=OMPc
# -------- model:vgg16/resnet18 -----------------
# model=vgg11
# model=vgg13
# model=vgg16
# model=vgg19
# model=resnet20
# model=resnet32
model=modela
# -------- hyper-parameters ---------------------
lamb=0.1
num_paths=10
# -------- CIFAR10 ------------------------------
# dataset=CIFAR10
# data_dir='/media/Disk1/KunFang/data/CIFAR10/'
# -------- CIFAR100 -----------------------------
# dataset=CIFAR100
# data_dir='/media/Disk1/KunFang/data/CIFAR100/'
# -------- STL10 --------------------------------
dataset=STL10
data_dir='/media/Disk1/KunFang/data/STL10/'
# -------- model directory ----------------------
model_dir='./save/'
# -----------------------------------------------
gpu_id=1
# -----------------------------------------------
# adv_train=False
adv_train=True
# -----------------------------------------------

python train.py \
    --arch ${arch} \
    --model ${model} \
    --lamb ${lamb} \
    --num_paths ${num_paths} \
    --dataset ${dataset} \
    --data_dir ${data_dir} \
    --model_dir ${model_dir} \
    --gpu_id ${gpu_id} \
    --adv_train ${adv_train}