-
Notifications
You must be signed in to change notification settings - Fork 2
/
Copy patheval.py
139 lines (119 loc) · 6.17 KB
/
eval.py
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
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
import os
import argparse
from unittest import result
import torchvision
from torchvision import transforms
import pickle
import attack_generator as attack
import numpy as np
import datetime
from models.resnet import *
from models.wrn_madry import *
from models.wide_resnet import *
parser = argparse.ArgumentParser(description='PyTorch Obtain Natural and Robust Accuracy')
### Experimental setting ###
parser.add_argument('--seed', type=int, default=1, metavar='S', help='random seed (default: 1)')
parser.add_argument('--dataset', type=str, default="cifar10", help="choose from cifar10,svhn,cifar100", choices=['cifar10', 'cifar100', 'svhn'])
parser.add_argument('--data_dir', type=str, default='../data', help='the directory to access to dataset')
parser.add_argument('--split', default='test', help='Test the accuracy of training or test data', choices=['test', 'train'])
parser.add_argument('--gpu',type=str,default='0')
### Attack model setting ###
parser.add_argument('--net', type=str, default='ResNet18', choices=['ResNet18', 'WRN-28-10', 'WRN-32-10', 'WRN-34-10'])
parser.add_argument('--depth', type=int, default=34, help='WRN depth')
parser.add_argument('--width-factor', type=int, default=10, help='WRN width factor')
parser.add_argument('--drop-rate', type=float, default=0.0, help='WRN drop rate')
parser.add_argument('--model_dir', type=str, default=None, help='attack model dir')
parser.add_argument('--pt_name',type=str,default='', help='the name of model checkpoint')
### Attack range setting ###
parser.add_argument('--all_epoch', action='store_true', help='whether to test the accuracy at each epoch')
parser.add_argument('--start_epoch', type=int, default=1)
parser.add_argument('--end_epoch', type=int, default=120)
### Attack strength setting ###
parser.add_argument('--epsilon', type=float, default=0.031, help='perturbation bound')
parser.add_argument('--num_steps', type=int, default=100, help='maximum perturbation step K')
parser.add_argument('--step_size', type=float, default=0.007, help='step size')
args = parser.parse_args()
# settings
os.environ["CUDA_VISIBLE_DEVICES"] = args.gpu
torch.manual_seed(args.seed)
np.random.seed(args.seed)
torch.cuda.manual_seed_all(args.seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
# setup data loader
print('===> Load data')
transform_test = transforms.Compose([
transforms.ToTensor(),
])
if args.dataset == "cifar10":
testset = torchvision.datasets.CIFAR10(root=args.data_dir, train=False, download=True, transform=transform_test)
trainset = torchvision.datasets.CIFAR10(root=args.data_dir, train=True, download=True, transform=transform_test)
test_loader = torch.utils.data.DataLoader(testset, batch_size=128, shuffle=False, num_workers=2)
train_loader = torch.utils.data.DataLoader(trainset, batch_size=128, shuffle=False, num_workers=2)
num_classes = 10
if args.dataset == "svhn":
testset = torchvision.datasets.SVHN(root=args.data_dir, split='test', download=True, transform=transform_test)
trainset = torchvision.datasets.SVHN(root=args.data_dir, split='train', download=True, transform=transform_test)
test_loader = torch.utils.data.DataLoader(testset, batch_size=128, shuffle=False, num_workers=2)
train_loader = torch.utils.data.DataLoader(trainset, batch_size=128, shuffle=False, num_workers=2)
num_classes = 10
if args.dataset == "cifar100":
testset = torchvision.datasets.CIFAR100(root=args.data_dir, train=False, download=True, transform=transform_test)
trainset = torchvision.datasets.CIFAR100(root=args.data_dir, train=True, download=True, transform=transform_test)
test_loader = torch.utils.data.DataLoader(testset, batch_size=128, shuffle=False, num_workers=2)
train_loader = torch.utils.data.DataLoader(trainset, batch_size=128, shuffle=False, num_workers=2)
num_classes = 100
if args.split == 'test':
loader = test_loader
else:
loader = train_loader
print('===> Load model')
net = args.net
if net == 'ResNet18':
model = ResNet18(num_classes).cuda()
if net == 'WRN_28-10':
model = Wide_ResNet_Madry(depth=28).cuda()
if net == 'WRN_32-10':
model = Wide_ResNet_Madry(depth=32).cuda()
if net == 'WRN-34-10':
model = Wide_ResNet(depth=34).cuda()
if args.all_epoch:
model_dir = args.model_dir
print(model_dir)
acc_pkl_name = model_dir + '/learning_curve_epoch{}_{}.pkl'.format(args.start_epoch, args.end_epoch)
print(acc_pkl_name)
natural_acc_list = []
robust_acc_list = []
result = dict()
for epoch in range(args.start_epoch, args.end_epoch + 1, 1):
starttime = datetime.datetime.now()
pt_path = os.path.join(args.dir, "checkpoint_epoch{}.pth.tar".format(epoch))
model.load_state_dict(torch.load(pt_path, map_location="cuda:0")['state_dict'])
model.eval()
_, natural_acc = attack.eval_clean(model, loader, epsilon=args.epsilon)
natural_acc_list.append(natural_acc)
_, robust_acc = attack.eval_AA(model, loader, args.epsilon)
robust_acc_list.append(robust_acc)
result.update({'epoch': epoch})
result.update({"natural_acc": natural_acc_list})
result.update({"robust_acc": robust_acc_list})
endtime = datetime.datetime.now()
time = (endtime - starttime).seconds
print('Epoch:{}\tNatural:{}\tAA:{}\ttime:{}'.format(epoch, natural_acc, robust_acc, time))
with open(acc_pkl_name, 'wb') as f:
pickle.dump(result, f)
f.close()
else:
starttime = datetime.datetime.now()
pt_name = os.path.join(args.model_dir, args.pt_name)
print('===> Load model parameters')
print(pt_name)
model.load_state_dict(torch.load(pt_name, map_location="cuda:0")['state_dict'])
model.eval()
_, natural_acc = attack.eval_clean(model, loader)
loss, CW100_acc = attack.eval_robust(model, loader, perturb_steps=args.num_steps, epsilon=args.epsilon, step_size=args.step_size,
loss_fn="cw", category="Madry", rand_init=True)
_, AA_acc = attack.eval_AA(model, loader, args.epsilon)
endtime = datetime.datetime.now()
time = (endtime - starttime).seconds
print('Natural:{}\tCW100:{}\tAA:{}\ttime:{}'.format(natural_acc, CW100_acc, AA_acc, time))