-
Notifications
You must be signed in to change notification settings - Fork 1
/
main.py
164 lines (128 loc) · 5.44 KB
/
main.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
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
'''Train CIFAR10 with PyTorch.'''
import torch
import torch.nn as nn
import torch.optim as optim
import torch.nn.functional as F
import torch.backends.cudnn as cudnn
import json
import torchvision
import torchvision.transforms as transforms
import os
import argparse
from models import *
from optimizer import *
from utils import *
# Training
def train(epoch):
print('\nEpoch: %d' % epoch)
net.train()
train_loss = 0
correct = 0
total = 0
for batch_idx, (inputs, targets) in enumerate(trainloader):
inputs, targets = inputs.to(device), targets.to(device)
optimizer.zero_grad()
outputs = net(inputs)
loss = criterion(outputs, targets)
loss.backward()
optimizer.step()
train_loss += loss.item()
_, predicted = outputs.max(1)
total += targets.size(0)
correct += predicted.eq(targets).sum().item()
progress_bar(batch_idx, len(trainloader), 'Loss: %.3f | Acc: %.3f%% (%d/%d)'
% (train_loss/(batch_idx+1), 100.*correct/total, correct, total))
train_loss = train_loss/(batch_idx+1)
train_acc = 100.*correct/total
return train_loss, train_acc
# test
def test(epoch, ckpt_name):
global best_acc
net.eval()
test_loss = 0
correct = 0
total = 0
with torch.no_grad():
for batch_idx, (inputs, targets) in enumerate(testloader):
inputs, targets = inputs.to(device), targets.to(device)
outputs = net(inputs)
loss = criterion(outputs, targets)
test_loss += loss.item()
_, predicted = outputs.max(1)
total += targets.size(0)
correct += predicted.eq(targets).sum().item()
progress_bar(batch_idx, len(testloader), 'Loss: %.3f | Acc: %.3f%% (%d/%d)'
% (test_loss/(batch_idx+1), 100.*correct/total, correct, total))
# Save checkpoint.
acc = 100.*correct/total
if acc > best_acc:
print('Saving..')
state = {
'net': net.state_dict(),
'acc': acc,
'epoch': epoch,
}
if not os.path.isdir('checkpoint'):
os.mkdir('checkpoint')
torch.save(state, './checkpoint/'+ ckpt_name + '.pth')
best_acc = acc
test_loss = test_loss/(batch_idx+1)
test_acc = 100.*correct/total
return test_loss, test_acc
if __name__ == "__main__":
parser = argparse.ArgumentParser(description='PyTorch Implementation for DPL on CIFAR100, CIFAR-100')
parser.add_argument('--dataset', type=str, default="CIFAR100", help="dataset ['CIFAR10'/'CIFAR100']")
parser.add_argument('--data_dir', type=str, default="./dataset", help="dataset dirpath")
parser.add_argument('--net', type=str, default="vgg16", help='net type')
parser.add_argument('--batch_size', type=int, default=128, help='batch size for dataloader')
parser.add_argument('--lr', default=0.1, type=float, help='initial learning rate')
parser.add_argument('--epoch', default=200, type=int, help='training epoch')
parser.add_argument('--resume', '-r', action='store_true', help='resume from checkpoint')
parser.add_argument('--alpha', default=0.05, type=float, help='perturbation strength')
parser.add_argument('--G_size', default=0.05, type=float, help="the size of Guide Set ('x%' of the training set size)")
parser.add_argument('--varepsilon', default=0.04, type=float, help="the size of Amended Training Samples ('x%' of the training set size)")
parser.add_argument('--reg_aug', default="aug", type=str, help="the perturbation approach flag 'rep_aug' = ['rep'/'aug']")
args = parser.parse_args()
device = 'cuda' if torch.cuda.is_available() else 'cpu'
best_acc = 0 # best test accuracy
start_epoch = 0 # start from epoch 0 or last checkpoint epoch
# Load network.
print('==> Building model..')
net = load_net(args)
net = net.to(device)
if device == 'cuda':
net = torch.nn.DataParallel(net)
cudnn.benchmark = True
# Load checkpoint.
if args.resume:
print('==> Resuming from checkpoint..')
ckpt_path = './checkpoint/vgg16.pth'
net, best_acc, start_epoch = load_checkpoint(net, ckpt_path)
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(net.parameters(), lr=args.lr,
momentum=0.9, weight_decay=5e-4)
scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(optimizer, T_max = args.epoch)
DPL_iter = 0
best_test_loss = float("inf")
while(True):
print('=== DPL Iteration {} Started ==='.format(DPL_iter))
# Load dataset.
print('==> Loading dataset..')
torch.manual_seed(0)
trainloader, testloader, ori_trainloader = load_data(args)
# Train and test
print('==> Starting to train and test..')
for epoch in range(start_epoch, start_epoch + args.epoch):
ckpt_name = args.net + "_DPL_iter_" + str(DPL_iter)
train_loss, train_acc = train(trainloader, epoch)
test_loss, test_acc = test(testloader, epoch, ckpt_name)
scheduler.step()
# Post-hoc optimization by DPL
print('==> Performing optimization by DPL..')
DPL_optimizer(net, trainloader, testloader, args, DPL_iter, ori_trainloader)
print('=== DPL Iteration {} Finished ==='.format(DPL_iter))
if best_test_loss > test_loss:
best_test_loss = test_loss
DPL_iter += 1
else:
break