-
Notifications
You must be signed in to change notification settings - Fork 0
/
train.py
177 lines (138 loc) · 5.86 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
import os
import sys
import random
import builtins
import warnings
import importlib
import subprocess
import torch
import torch.multiprocessing as mp
import torch.backends.cudnn as cudnn
from torch.distributed import init_process_group
from torch.nn.parallel import DistributedDataParallel as DDP
from data import DataPrefetcher
from utils import resume, get_state_dict, save_checkpoint
from utils.log import setup_logger, setup_writer
from utils.dist import synchronize
from utils.default_argparse import default_argument_parser
def main_worker(gpu, ngpus_per_node, args):
args.gpu = gpu
# ------------ set environment variables for distributed training ------------------------------------- #
# suppress printing if not master
if args.multiprocessing_distributed and args.gpu != 0:
def print_pass(*args):
pass
builtins.print = print_pass
if args.gpu is not None:
print("Use GPU: {} for training".format(args.gpu))
if args.rank == -1:
args.rank = args.gpu
if args.multiprocessing_distributed:
# For multiprocessing distributed training, rank needs to be the
# global rank among all the processes
args.rank = (args.num_machines-1) * ngpus_per_node + gpu
init_process_group(backend=args.dist_backend, init_method=args.dist_url,
world_size=args.world_size, rank=args.rank)
# get the Exp file
if not args.exp_file:
print('Exp file missing.')
sys.exit(1)
else:
sys.path.insert(0, os.path.dirname(args.exp_file))
current_exp = importlib.import_module(os.path.basename(args.exp_file).split(".")[0])
trainer = current_exp.Trainer()
# update some config if necessary
updated_config = trainer.update(args.exp_options)
# make dir for experiment output
file_name = os.path.join(args.output_dir, args.experiment_name)
if args.rank == 0:
if not os.path.exists(args.output_dir):
os.mkdir(args.output_dir)
if not args.resume:
if os.path.exists(file_name):
raise ValueError('Experiment name conflicts.')
else:
os.mkdir(file_name)
synchronize()
# setup the logger and writer
logger = setup_logger(file_name, distributed_rank=args.rank, filename='train_log.txt', mode='a')
writer = setup_writer(file_name, distributed_rank=args.rank)
# setup model, dataloader and optimizer
trainer.build_dataloader(args)
trainer.build_model()
trainer.build_optimizer(args)
if args.rank == 0:
logger.info('args: {}'.format(args))
hyper_param = []
for k in trainer.__dict__:
if 'model' not in k:
hyper_param.append(str(k) + '=' + str(trainer.__dict__[k]))
logger.info('Hyper-parameters: {}'.format(', '.join(hyper_param)))
if updated_config:
logger.opt(ansi=True).info("List of override configs:\n<blue>{}</blue>\n".format(updated_config))
if args.rank == 0:
logger.info('Model: ')
logger.info(str(trainer.model))
# put the model onto gpu
torch.cuda.set_device(gpu)
trainer.model.cuda(gpu)
if trainer.CLS:
trainer.classifier.cuda(gpu)
if ngpus_per_node > 1:
trainer.model = DDP(trainer.model, device_ids=[gpu])
if trainer.CLS:
trainer.classifier = DDP(trainer.classifier, device_ids=[gpu])
cudnn.benchmark = True
# resume
if args.resume:
resume(args, trainer)
# ------------------------ start training ------------------------------------------------------------ #
if args.rank == 0:
logger.info('Start training from iteration {},'
' and the total training iterations is {}'.format(trainer.ITERS_PER_EPOCH * args.start_epoch + 1,
trainer.total_iters))
trainer.prefetcher = DataPrefetcher(trainer.data_loader, args.single_aug)
for epoch in range(args.start_epoch, args.total_epochs):
# set epoch
trainer.epoch = epoch
if trainer.prefetcher.next_input is None:
if args.world_size > 1:
trainer.data_loader.sampler.set_epoch(epoch)
trainer.prefetcher = DataPrefetcher(trainer.data_loader, args.single_aug)
trainer.train(args, logger, writer)
# save models
if args.rank == 0:
state_dict = get_state_dict(trainer)
save_checkpoint(state_dict, False, file_name, 'last_epoch')
if args.rank == 0:
logger.info("Pre-training of experiment: {} is done.".format(args.experiment_name))
writer.close()
def main():
args = default_argument_parser().parse_args()
# setup randomization
if args.seed is not None:
random.seed(args.seed)
torch.manual_seed(args.seed)
cudnn.deterministic = True
warnings.warn(
"You have chosen to seed training. This will turn on the CUDNN deterministic setting, "
"which can slow down your training considerably! You may see unexpected behavior when restarting "
"from checkpoints."
)
# multi-processing
args.multiprocessing_distributed = args.num_machines > 1
print('Total number of using machines: {}'.format(args.num_machines))
if args.machine_rank == 0:
master_ip = subprocess.check_output(['hostname', '--fqdn']).decode("utf-8")
master_ip = str(master_ip).strip()
args.dist_url = 'tcp:https://{}:23456'.format(master_ip)
print('dist_url on Machine 0:', args.dist_url)
ngpus_per_node = torch.cuda.device_count()
if ngpus_per_node > 1:
args.world_size = ngpus_per_node * args.num_machines
mp.spawn(main_worker, nprocs=ngpus_per_node, args=(ngpus_per_node, args))
else:
args.world_size = 1
main_worker(0, ngpus_per_node, args)
if __name__ == "__main__":
main()