forked from Kurumi233/OnlineLabelSmoothing
-
Notifications
You must be signed in to change notification settings - Fork 0
/
train.py
336 lines (275 loc) · 11.9 KB
/
train.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
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
import os
import time
import random
import argparse
import numpy as np
import torch
import torch.nn as nn
from torch.optim import lr_scheduler
from torch.backends import cudnn
from torch.utils.data import DataLoader
from torch.utils.data.distributed import DistributedSampler
import torchvision
from models.model import BaseModel
from models.FocalLoss import FocalLoss
from models.LabelSmoothing import LabelSmoothing
from models.OLS import OnlineLabelSmoothing
from ImageNetLoad import ImageNet
from utils import MultiLossAverageMeter, AverageMeter, accuracy
from plot_result import plot_result
# os.environ["KMP_DUPLICATE_LIB_OK"] = "TRUE"
parser = argparse.ArgumentParser()
parser.add_argument('--model_name', default='res50', type=str)
parser.add_argument('--savepath', default='./Single_gpu', type=str)
parser.add_argument('--loss', default=['ce'], nargs='+', type=str)
parser.add_argument('--loss_w', default=[1.], nargs='+', type=float)
parser.add_argument('--smoothing', default=0.1, type=float)
parser.add_argument('-c', '--num_classes', default=1000, type=int)
parser.add_argument('-p', '--pool_type', default='avg', type=str)
parser.add_argument('--metric', default='linear', type=str)
parser.add_argument('--down', default=0, type=int)
parser.add_argument('--lr', default=0.1, type=float)
parser.add_argument('--weight_decay', default=1e-4, type=float)
parser.add_argument('--momentum', default=0.9, type=float)
parser.add_argument('-s', '--scheduler', default='step', type=str)
parser.add_argument('-r', '--resume', default=None, type=str)
parser.add_argument('--lr_step', default=30, type=int)
parser.add_argument('--warm', default=5, type=int)
parser.add_argument('--print_step', default=500, type=int)
parser.add_argument('--lr_gamma', default=0.1, type=float)
parser.add_argument('--total_epoch', default=250, type=int)
parser.add_argument('-bs', '--batch_size', default=256, type=int)
parser.add_argument('-nw', '--num_workers', default=20, type=int)
parser.add_argument('--multi-gpus', default=0, type=int)
parser.add_argument('--seed', default=2020, type=int)
parser.add_argument('--pretrained', default=0, type=int)
parser.add_argument('--local_rank', default=0, type=int)
parser.add_argument('--sync_bn', default=False, action='store_true')
parser.add_argument('--amp', default=False, action='store_true')
args = parser.parse_args()
print('local_rank:', args.local_rank)
ce_based_loss = ['ce', 'ls', 'fl', 'ols']
def loss_func(x, target, feat=None, training=False):
loss_dict = {}
loss_value = 0.
for l, w in zip(args.loss, args.loss_w):
if training:
criterion[l].train()
else:
criterion[l].eval()
loss = w * criterion[l](x, target)
loss_value += loss
loss_dict[l] = loss.detach().cpu().item()
return loss_dict, loss_value
def train(epoch):
model.train()
loss_meter = MultiLossAverageMeter(args.loss)
top1 = AverageMeter('Acc@1', ':.2f')
top5 = AverageMeter('Acc@5', ':.2f')
t1 = time.time()
s1 = time.time()
for idx, (data, labels) in enumerate(trainloader):
if multi_gpus:
data, labels = data.cuda(non_blocking=True), labels.long().cuda(non_blocking=True)
else:
data, labels = data.to(device), labels.long().to(device)
optimizer.zero_grad()
# AMP
if args.amp:
with torch.cuda.amp.autocast():
out, feat = model(data)
loss_dict, loss = loss_func(out, labels, feat, training=True)
scaler.scale(loss).backward()
scaler.step(optimizer)
scaler.update()
else:
out, feat = model(data)
loss_dict, loss = loss_func(out, labels, feat, training=True)
loss.backward()
optimizer.step()
loss_meter.update(loss_dict, data.size(0))
acc1, acc5 = accuracy(out, labels, topk=(1, 5))
top1.update(acc1.item(), data.size(0))
top5.update(acc5.item(), data.size(0))
if idx % args.print_step == 0:
s2 = time.time()
print(f'rank:{args.local_rank} epoch[{epoch:>3}/{args.total_epoch}] idx[{idx:>3}/{len(trainloader)}] loss[{loss_meter}] acc[@1:{top1.avg:.4f} @5:{top5.avg:.4f}] time:{s2 - s1:.2f}s')
s1 = time.time()
if args.local_rank == 0:
print('=' * 30)
print(f'rank:{args.local_rank} train loss[{loss_meter}] acc[@1:{top1.avg:.4f} @5:{top5.avg:.4f}] time:{time.time() - t1:.2f}s')
if args.local_rank == 0:
with open(os.path.join(savepath, 'log.txt'), 'a+')as f:
f.write('epoch:{} lr:{:.8f} loss[{}] acc[@1:{:.4f} @5:{:.4f}] '.format(epoch, optimizer.param_groups[0]['lr'], loss_meter, top1.avg, top5.avg))
def test(epoch):
model.eval()
loss_meter = MultiLossAverageMeter(args.loss)
top1 = AverageMeter('Acc@1', ':.2f')
top5 = AverageMeter('Acc@5', ':.2f')
with torch.no_grad():
for idx, (data, labels) in enumerate(valloader):
data, labels = data.to(device), labels.long().to(device)
out = model(data)
loss_dict, loss = loss_func(out, labels, training=False)
loss_meter.update(loss_dict, data.size(0))
acc1, acc5 = accuracy(out, labels, topk=(1, 5))
top1.update(acc1.item(), data.size(0))
top5.update(acc5.item(), data.size(0))
print(f'rank:{args.local_rank} test loss[{loss_meter}] acc[@1:{top1.avg:.4f} @5:{top5.avg:.4f}]', end=' ')
global best_acc, best_epoch
if isinstance(model, nn.parallel.distributed.DistributedDataParallel):
state = {
'net': model.module.state_dict(),
'acc': top1.avg,
'epoch': epoch,
}
else:
state = {
'net': model.state_dict(),
'acc': top1.avg,
'epoch': epoch,
}
if 'ols' in args.loss:
state['ols'] = criterion['ols'].matrix.cpu().data
if top1.avg > best_acc:
best_acc = top1.avg
best_epoch = epoch
torch.save(state, os.path.join(savepath, 'best.pth'))
print('*')
else:
print()
torch.save(state, os.path.join(savepath, 'last.pth'))
with open(os.path.join(savepath, 'log.txt'), 'a+')as f:
f.write('test loss[{}] acc[@1:{:.4f} @5:{:.4f}]\n'.format(loss_meter, top1.avg, top5.avg))
if __name__ == '__main__':
best_epoch = 0
best_acc = 0.
use_gpu = False
multi_gpus = False
start_epoch = 0
total = args.total_epoch
if args.seed is not None:
print('use random seed:', args.seed)
torch.manual_seed(args.seed)
torch.cuda.manual_seed(args.seed)
torch.cuda.manual_seed_all(args.seed)
np.random.seed(args.seed)
random.seed(args.seed)
cudnn.deterministic = False
if torch.cuda.is_available():
use_gpu = True
cudnn.benchmark = True
if torch.cuda.device_count() > 1 and args.multi_gpus:
torch.distributed.init_process_group(backend="nccl")
torch.cuda.set_device(args.local_rank)
multi_gpus = True
# loss
criterion = {
'ce': nn.CrossEntropyLoss(),
'fl': FocalLoss(),
'ls': LabelSmoothing(smoothing=args.smoothing),
'ols': OnlineLabelSmoothing(num_classes=args.num_classes, use_gpu=use_gpu)
}
# dataloader
trainset = ImageNet(mode='train')
valset = ImageNet(mode='val')
# dataloader
train_sampler = None
if multi_gpus:
train_sampler = torch.utils.data.distributed.DistributedSampler(trainset)
trainloader = DataLoader(dataset=trainset,
batch_size=args.batch_size,
shuffle=(train_sampler is None),
sampler=train_sampler,
num_workers=args.num_workers,
pin_memory=True)
valloader = DataLoader(dataset=valset,
batch_size=args.batch_size,
shuffle=False,
num_workers=args.num_workers,
pin_memory=True)
# model
model = BaseModel(model_name=args.model_name,
num_classes=args.num_classes,
pretrained=args.pretrained)
if args.resume:
state = torch.load(args.resume)
print('Resume from:{}'.format(args.resume))
model.load_state_dict(state['net'], strict=False)
best_acc = state['acc']
start_epoch = state['epoch'] + 1
if 'ols' in args.loss:
criterion['ols'].matrix = state['ols'].cuda()
# sync_bn
if args.sync_bn and multi_gpus:
model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model)
print('Using SyncBatchNorm')
if multi_gpus:
device = torch.device("cuda", args.local_rank)
model = model.to(device)
model = nn.parallel.DistributedDataParallel(model, device_ids=[args.local_rank])
else:
device = ('cuda:%d' % args.local_rank if torch.cuda.is_available() else 'cpu')
model = model.to(device)
print('Device:', device)
# optim
optimizer = torch.optim.SGD(
[{'params': filter(lambda p: p.requires_grad, model.parameters()), 'lr': args.lr}],
weight_decay=args.weight_decay, momentum=args.momentum)
if args.scheduler == 'step':
scheduler = lr_scheduler.StepLR(optimizer, step_size=args.lr_step, gamma=args.lr_gamma, last_epoch=-1)
elif args.scheduler == 'multi':
scheduler = lr_scheduler.MultiStepLR(optimizer, milestones=[75, 150, 225], gamma=args.lr_gamma, last_epoch=-1)
elif args.scheduler == 'cos':
warm_up_step = args.warm
lambda_ = lambda epoch: (epoch + 1) / warm_up_step if epoch <= warm_up_step else 0.5 * (
np.cos((epoch - warm_up_step) / (args.total_epoch - warm_up_step) * np.pi) + 1)
scheduler = torch.optim.lr_scheduler.LambdaLR(optimizer, lambda_)
else:
raise ValueError('No such scheduler - {}'.format(args.scheduler))
# savepath
loss_str = '_'.join(args.loss)
if 'ls' in args.loss:
loss_str += str(args.smoothing)
savepath = os.path.join(args.savepath, '{}_{}_{}_{}_{}'.format(args.model_name,
args.pool_type,
args.metric,
str(args.down),
loss_str))
# AMP
if args.amp:
scaler = torch.cuda.amp.GradScaler()
print('Using Mixing Accuracy.')
savepath += '_amp'
if args.sync_bn:
savepath += '_syncbn'
savepath += args.scheduler
if args.local_rank == 0:
print('Init_lr={}, Weight_decay={}, Momentum={}'.format(args.lr, args.weight_decay, args.momentum))
print('Loss:', args.loss)
print('Loss_weight:', args.loss_w)
print('Using {} scheduler'.format(args.scheduler))
print('Savepath:', savepath)
os.makedirs(savepath, exist_ok=True)
if args.local_rank == 0 and args.resume is None:
with open(os.path.join(savepath, 'setting.txt'), 'w')as f:
for k, v in vars(args).items():
f.write('{}:{}\n'.format(k, v))
f = open(os.path.join(savepath, 'log.txt'), 'w')
f.close()
start = time.time()
for epoch in range(start_epoch, total):
train(epoch)
scheduler.step()
if args.local_rank == 0:
test(epoch)
if 'ols' in args.loss:
criterion['ols'].update()
end = time.time()
if args.local_rank == 0:
print('total time:{}m{:.2f}s'.format((end - start) // 60, (end - start) % 60))
print('best_epoch:', best_epoch)
print('best_acc:', best_acc)
with open(os.path.join(savepath, 'log.txt'), 'a+')as f:
f.write('# best_acc:{:.4f}, best_epoch:{}'.format(best_acc, best_epoch))
plot_result(txt=os.path.join(savepath, 'log.txt'), savepath=savepath)