-
Notifications
You must be signed in to change notification settings - Fork 0
/
cifar.py
210 lines (170 loc) · 6.94 KB
/
cifar.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
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
'''Train CIFAR with PyTorch.
e.g.
python3 cifar.py --netName=PreActResNet18 --cifar=10 --bs=512
'''
from __future__ import print_function
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 torchvision
import torchvision.transforms as transforms
import os
import argparse
import random
#from models import *
import models as models
from utils import *
model_names = sorted(name for name in models.__dict__
if not name.startswith("__")
and callable(models.__dict__[name]))
# print(model_names)
os.environ["HDF5_USE_FILE_LOCKING"] = "FALSE"
os.environ["CUDA_VISIBLE_DEVICES"] = "0,1,2,3"
parser = argparse.ArgumentParser(description='PyTorch CIFAR10 Training')
parser.add_argument('--lr', default=0.1, type=float, help='learning rate')
parser.add_argument('--resume', '-r',default=False, action='store_true', help='resume from checkpoint')
parser.add_argument('--netName', default='PreActResNet18', choices=model_names, type=str, help='choosing network')
parser.add_argument('--bs', default=512, type=int, help='batch size')
parser.add_argument('--es', default=150, type=int, help='epoch size')
parser.add_argument('--cifar', default=100, type=int, help='dataset classes number')
parser.add_argument('--fix_seed', default=123, help='Fix random seed')
args = parser.parse_args()
device = 'cuda' if torch.cuda.is_available() else 'cpu'
print(device)
best_acc = 0 # best test accuracy
start_epoch = 0 # start from epoch 0 or last checkpoint epoch
if args.fix_seed>0:
# Seed model
random.seed(args.fix_seed)
torch.manual_seed(args.fix_seed)
cudnn.deterministic = True
print("SEED MODEL: Fix seed as ", args.fix_seed)
else:
print("SEED MODEL: Using random seed.")
# Data
print('==> Preparing data..')
transform_train = transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),
])
transform_test = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465), (0.2023, 0.1994, 0.2010)),
])
if args.cifar ==100:
trainset = torchvision.datasets.CIFAR100(root='./data', train=True, download=True, transform=transform_train)
else:
args.cifar=10
trainset = torchvision.datasets.CIFAR10(root='./data', train=True, download=True, transform=transform_train)
trainloader = torch.utils.data.DataLoader(trainset, batch_size=args.bs, shuffle=True, num_workers=4)
if args.cifar ==100:
testset = torchvision.datasets.CIFAR100(root='./data', train=False, download=True, transform=transform_test)
else:
testset = torchvision.datasets.CIFAR10(root='./data', train=False, download=True, transform=transform_test)
testloader = torch.utils.data.DataLoader(testset, batch_size=args.bs, shuffle=False, num_workers=4)
# Model
print('==> Building model..')
try:
net = models.__dict__[args.netName](num_classes=args.cifar)
except:
net = models.__dict__[args.netName]()
para_numbers = count_parameters(net)
print("Total parameters number is: "+ str(para_numbers))
net = net.to(device)
if device == 'cuda':
net = torch.nn.DataParallel(net)
cudnn.benchmark = True
if args.resume:
# Load checkpoint.
print('==> Resuming from checkpoint..')
assert os.path.isdir('checkpoint'), 'Error: no checkpoint directory found!'
checkpoint_path = './checkpoint/ckpt_cifar_'+str(args.cifar)+'_'+args.netName+'.t7'
checkpoint = torch.load(checkpoint_path)
net.load_state_dict(checkpoint['net'])
best_acc = checkpoint['acc']
print("BEST_ACCURACY: "+str(best_acc))
start_epoch = checkpoint['epoch']
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(net.parameters(), lr=args.lr, momentum=0.9, weight_decay=5e-4)
# Training
def train(epoch):
adjust_learning_rate(optimizer, epoch, args.lr)
print('\nEpoch: %d Learning rate: %f' % (epoch, optimizer.param_groups[0]['lr']))
print("\nAllocated GPU memory:", torch.cuda.memory_allocated())
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))
file_path='../records/cifar100/cifar_' + str(args.cifar) + '_' +args.netName+'_train.txt'
record_str=str(epoch)+'\t'+"%.3f"%(train_loss/(batch_idx+1))+'\t'+"%.3f"%(100.*correct/total)+'\n'
write_record(file_path,record_str)
def test(epoch):
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))
file_path = '../records/cifar100/cifar_' + str(args.cifar) + '_' +args.netName+ '_test.txt'
record_str = str(epoch) + '\t' + "%.3f" % (test_loss / (batch_idx + 1)) + '\t' + "%.3f" % (
100. * correct / total) + '\n'
write_record(file_path, record_str)
# 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')
save_path = './checkpoint/ckpt_cifar_' + str(args.cifar) + '_' + args.netName + '.t7'
torch.save(state, save_path)
best_acc = acc
for epoch in range(start_epoch, start_epoch+args.es):
train(epoch)
test(epoch)
# write statistics to files
statis_path = '../records/cifar100/STATS_'+args.netName+'.txt'
if not os.path.exists(statis_path):
# os.makedirs(statis_path)
os.system(r"touch {}".format(statis_path))
f = open(statis_path, 'w')
statis_str="============\nDivces:"+device+"\n"
statis_str+='\n===========\nargs:\n'
statis_str+=args.__str__()
statis_str+='\n==================\n'
statis_str+="BEST_accuracy: "+str(best_acc)
statis_str+='\n==================\n'
statis_str+="Total parameters: "+str(para_numbers)
f.write(statis_str)
f.close()