-
Notifications
You must be signed in to change notification settings - Fork 1
/
run_byol.py
152 lines (126 loc) · 5.21 KB
/
run_byol.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
# -*- coding: utf-8 -*-
import os
import sys
import typing
import warnings
import wandb
import numpy as np
import torch
import torch.nn as nn
import torch.multiprocessing as mp
import torch.distributed as dist
from rich.console import Console
from datasets.cifar import CIFAR10, CIFAR10Pair
from datasets.cifar import CIFAR100, CIFAR100Pair
from datasets.stl10 import STL10, STL10Pair
from datasets.imagenet import ImageNet, ImageNetPair
from datasets.transforms import MoCoAugment, FinetuneAugment, TestAugment
from configs.task_configs import BYOLConfig
from tasks.byol import BYOL
from utils.wandb import initialize_wandb
def fix_random_seed(s: int = 0):
np.random.seed(s)
torch.manual_seed(s)
def main(config: BYOLConfig):
"""Main function for single or distributed BYOL training."""
os.environ['CUDA_VISIBLE_DEVICES'] = ','.join([str(gpu) for gpu in config.gpus])
num_gpus_per_node = len(config.gpus)
world_size = config.num_nodes * num_gpus_per_node
distributed = world_size > 1
setattr(config, 'num_gpus_per_node', num_gpus_per_node)
setattr(config, 'world_size', world_size)
setattr(config, 'distributed', distributed)
console = Console()
console.log(config.__dict__)
config.save()
fix_random_seed(config.seed)
if config.distributed:
console.print(f"Distributed training on {world_size} GPUs.")
mp.spawn(
main_worker,
nprocs=config.num_gpus_per_node,
args=(config, )
)
else:
console.print(f"Single GPU training on GPU {config.gpus[0]}.")
main_worker(0, config=config)
def main_worker(local_rank: int, config: object):
"""Single process of BYOL training."""
# Initialize the training process
torch.cuda.set_device(local_rank)
if config.distributed:
dist_rank = config.node_rank * config.num_gpus_per_node local_rank
dist.init_process_group(
backend=config.dist_backend,
init_method=config.dist_url,
world_size=config.world_size,
rank=dist_rank
)
config.batch_size = config.batch_size // config.world_size
config.num_workers = config.num_workers // config.num_gpus_per_node
data_aug_config = dict(size=config.input_size, data=config.data, impl=config.augmentation)
byol_transform = MoCoAugment(**data_aug_config)
finetune_transform = FinetuneAugment(**data_aug_config)
test_transform = TestAugment(**data_aug_config)
# Instantiate datasets used for training, evaluation, and testing.
data_dir = os.path.join(config.data_root, config.data)
if config.data == 'cifar10':
train_set = CIFAR10Pair(data_dir,
train=True,
transform=byol_transform)
finetune_set = CIFAR10(data_dir, train=True, transform=finetune_transform)
test_set = CIFAR10(data_dir, train=False, transform=test_transform)
elif config.data == 'cifar100':
train_set = CIFAR100Pair(data_dir,
train=True,
transform=byol_transform)
finetune_set = CIFAR100(data_dir, train=True, transform=finetune_transform)
test_set = CIFAR100(data_dir, train=False, transform=test_transform)
elif config.data == 'stl10':
train_set = STL10Pair(data_dir,
split='train unlabeled',
transform=byol_transform)
finetune_set = STL10(data_dir, split='train', transform=finetune_transform)
test_set = STL10(data_dir, split='test', transform=test_transform)
elif config.data == 'imagenet':
train_set = ImageNetPair(data_dir,
split='train',
transform=byol_transform)
finetune_set, test_set = ImageNet.split_into_two_subsets(
data_dir, split='val', transforms=[finetune_transform, test_transform]
)
else:
raise NotImplementedError(
f"Invalid data argument: {config.data}. "
f"Supports only one of the following: 'cifar10', 'cifar100', 'stl10', 'imagenet'."
)
# A wandb instance: https://wandb.ai
if local_rank == 0:
initialize_wandb(config)
# Instantiate BYOL trainer
trainer = BYOL(config=config, local_rank=local_rank)
# Start training
elapsed_sec = trainer.run(
dataset=train_set,
finetune_set=finetune_set,
test_set=test_set,
save_every=config.save_every,
eval_every=config.eval_every,
)
if trainer.logger is not None:
elapsed_mins = elapsed_sec / 60
elapsed_hours = elapsed_mins / 60
trainer.logger.info(f'Total training time: {elapsed_mins:,.2f} minutes ({elapsed_hours:,.2f} hours).')
trainer.logger.handlers.clear()
wandb.finish()
if __name__ == '__main__':
warnings.filterwarnings('ignore')
torch.backends.cudnn.deterministic = False
torch.backends.cudnn.benchmark = True
config = BYOLConfig.parse_arguments()
try:
main(config)
except KeyboardInterrupt:
wandb.finish()
os.system("kill $(ps aux | grep multiprocessing.spawn | grep -v grep | awk '{print $2}') ")
sys.exit(0)