-
Notifications
You must be signed in to change notification settings - Fork 0
/
model_batch64.py
282 lines (241 loc) · 11 KB
/
model_batch64.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
"""
Implementation of AlexNet, from paper
"ImageNet Classification with Deep Convolutional Neural Networks" by Alex Krizhevsky et al.
See: https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf
"""
import os
import torch
import torch.nn as nn
import torch.optim as optim
import torch.nn.functional as F
from torch.utils import data
from torch.utils.data.distributed import DistributedSampler
import torchvision.datasets as datasets
import torchvision.transforms as transforms
# from tensorboardX import SummaryWriter
import torch.multiprocessing as mp
# import horovod.torch as hvd
# 新增多节点运行DDP
import torch.distributed as dist
from torch.nn.parallel import DistributedDataParallel as DDP
import argparse
import time
# define model parameters
NUM_EPOCHS = 1
BATCH_SIZE = 64
MOMENTUM = 0.9
LR_DECAY = 0.0005
LR_INIT = 0.01
IMAGE_DIM = 227 # pixels
NUM_CLASSES = 1000 # 1000 classes for imagenet 2012 dataset
DEVICE_IDS = [0] # GPUs to use
# modify this to point to your data directory
# INPUT_ROOT_DIR = 'alexnet_data_in'
TRAIN_IMG_DIR = '/work1/aicao/teapot/data/ImageNet227/train/'
OUTPUT_DIR = 'alexnet_data_out'
LOG_DIR = OUTPUT_DIR + '/tblogs' # tensorboard logs
CHECKPOINT_DIR = OUTPUT_DIR + '/models' # model checkpoints
# make checkpoint path directory
os.makedirs(CHECKPOINT_DIR, exist_ok=True)
class AlexNet(nn.Module):
"""
Neural network model consisting of layers propsed by AlexNet paper.
"""
def __init__(self, num_classes=1000):
"""
Define and allocate layers for this neural net.
Args:
num_classes (int): number of classes to predict with this model
"""
super().__init__()
# input size should be : (b x 3 x 227 x 227)
# The image in the original paper states that width and height are 224 pixels, but
# the dimensions after first convolution layer do not lead to 55 x 55.
self.net = nn.Sequential(
nn.Conv2d(in_channels=3, out_channels=96, kernel_size=11, stride=4), # (b x 96 x 55 x 55)
nn.ReLU(),
# nn.LocalResponseNorm(size=5, alpha=0.0001, beta=0.75, k=2), # section 3.3
nn.MaxPool2d(kernel_size=3, stride=2), # (b x 96 x 27 x 27)
nn.Conv2d(96, 256, 5, padding=2), # (b x 256 x 27 x 27)
nn.ReLU(),
# nn.LocalResponseNorm(size=5, alpha=0.0001, beta=0.75, k=2), # 去除normalization测试时间
nn.MaxPool2d(kernel_size=3, stride=2), # (b x 256 x 13 x 13)
nn.Conv2d(256, 384, 3, padding=1), # (b x 384 x 13 x 13)
nn.ReLU(),
nn.Conv2d(384, 384, 3, padding=1), # (b x 384 x 13 x 13)
nn.ReLU(),
nn.Conv2d(384, 256, 3, padding=1), # (b x 256 x 13 x 13)
nn.ReLU(),
nn.MaxPool2d(kernel_size=3, stride=2), # (b x 256 x 6 x 6)
)
# classifier is just a name for linear layers
self.classifier = nn.Sequential(
nn.Dropout(p=0.5, inplace=True), # dropout probability
nn.Linear(in_features=(256 * 6 * 6), out_features=4096),
nn.ReLU(),
nn.Dropout(p=0.5, inplace=True),
nn.Linear(in_features=4096, out_features=4096),
nn.ReLU(),
nn.Linear(in_features=4096, out_features=num_classes),
)
self.init_bias() # initialize bias and weights
def init_bias(self):
for layer in self.net:
if isinstance(layer, nn.Conv2d):
nn.init.normal_(layer.weight, mean=0, std=0.01) # Gaussian distribution
nn.init.constant_(layer.bias, 0)
# original paper = 1 for Conv2d layers 2nd, 4th, and 5th conv layers
# nn.init.constant_(self.net[4].bias, 1)
# nn.init.constant_(self.net[10].bias, 1)
# nn.init.constant_(self.net[12].bias, 1)
nn.init.constant_(self.net[3].bias, 1)
nn.init.constant_(self.net[8].bias, 1)
nn.init.constant_(self.net[10].bias, 1)
def forward(self, x):
"""
Pass the input through the net.
Args:
x (Tensor): input tensor
Returns:
output (Tensor): output tensor
"""
x = self.net(x)
x = x.view(-1, 256 * 6 * 6) # reduce the dimensions for linear layer input
return self.classifier(x)
def parse_opt():
parser = argparse.ArgumentParser(description='PyTorch DDP ImageNet2012 slurm training')
parser.add_argument("--master_addr", required=True, type=str, default='localhost', help="Address of master, will default to localhost.")
parser.add_argument('--master_port', required=True, type=str, default='29500', help="Port that master is listening on, will default to 29500.")
parser.add_argument('--batch-size', type=int, default=16, help='total batch size for all GPUs, -1 for autobatch')
parser.add_argument('--dist_url', type=str, help='distributed backend init_method')
parser.add_argument('--world_size', type=int, default=1, help='world size of multi gpu/machine training')
parser.add_argument('--local_rank', type=int, default=0, help='process rank')
opt = parser.parse_args()
return opt
if __name__ == '__main__':
# time_start = time.time()
# print the seed value
seed = torch.initial_seed()
# print('Used seed : {}'.format(seed))
# tbwriter = SummaryWriter(log_dir=LOG_DIR)
# print('TensorboardX summary writer created')
opt = parse_opt()
torch.distributed.init_process_group(backend="nccl", init_method=opt.dist_url, world_size=opt.world_size, rank=opt.local_rank)
# create model
# local_rank = opt.local_rank
# torch.cuda.set_device(local_rank)
# device = torch.device(f'cuda:{opt.local_rank}')
device = torch.device("cuda:0")
alexnet = AlexNet(num_classes=NUM_CLASSES).to(device)
# 构造DDP模型
# alexnet = alexnet.cuda() # 提前把模型加载到GPU
alexnet = DDP(alexnet, device_ids=DEVICE_IDS, output_device=opt.local_rank).to(device)
# print(alexnet)
# print('AlexNet created')
# alexnet.apply(weights_init)
# create dataset and data loader
# data_time_1 = time.time()
dataset = datasets.ImageFolder(os.path.join(TRAIN_IMG_DIR), transforms.Compose([
# transforms.RandomResizedCrop(IMAGE_DIM, scale=(0.9, 1.0), ratio=(0.9, 1.1)),
transforms.CenterCrop(IMAGE_DIM),
# transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
]))
# print('Dataset created')
# 新增DistributedSampler
train_sampler = DistributedSampler(dataset)
dataloader = data.DataLoader(
dataset=dataset,
# shuffle=True,
pin_memory=True,
# num_workers=8,
drop_last=True,
batch_size=BATCH_SIZE,
sampler=train_sampler)
# print('Dataloader created')
# data_time_2 = time.time()
# data_time = data_time_2 - data_time_1
# print(data_time)
# create optimizer
# the one that WORKS
optimizer = optim.Adam(params=alexnet.parameters(), lr=0.0001)
# optimizer = optim.Adam(params=alexnet.parameters(), lr=LR_INIT, weight_decay=LR_DECAY)
### BELOW is the setting proposed by the original paper - which doesn't train....
# optimizer = optim.SGD(
# params=alexnet.parameters(),
# lr=LR_INIT,
# momentum=MOMENTUM,
# weight_decay=LR_DECAY)
# print('Optimizer created')
# multiply LR by 1 / 10 after every 30 epochs
lr_scheduler = optim.lr_scheduler.StepLR(optimizer, step_size=30, gamma=0.1) # 学习率调整
# print('LR Scheduler created')
# start training!!
# print('Starting training...')
total_steps = 0
time_start = time.time()
for epoch in range(NUM_EPOCHS):
# lr_scheduler.step()
for imgs, classes in dataloader:
imgs, classes = imgs.to(device), classes.to(device)
# print(classes)
# calculate the loss
output = alexnet(imgs)
# print(output)
loss = F.cross_entropy(output, classes)
# update the parameters
optimizer.zero_grad()
loss.backward()
optimizer.step()
lr_scheduler.step()
# # log the information and add to tensorboard
# if total_steps % 10 == 0:
# with torch.no_grad():
# _, preds = torch.max(output, 1)
# accuracy = torch.sum(preds == classes)
# # print('Epoch: {} \tStep: {} \tLoss: {:.12f} \tAcc: {}'
# # .format(epoch + 1, total_steps, loss.item(), accuracy.item()))
# # tbwriter.add_scalar('loss', loss.item(), total_steps)
# # tbwriter.add_scalar('accuracy', accuracy.item(), total_steps)
# # print out gradient values and parameter average values
# if total_steps % 100 == 0:
# with torch.no_grad():
# # print and save the grad of the parameters
# # also print and save parameter values
# # print('*' * 10)
# for name, parameter in alexnet.named_parameters():
# if parameter.grad is not None:
# avg_grad = torch.mean(parameter.grad)
# # print('\t{} - grad_avg: {}'.format(name, avg_grad))
# # tbwriter.add_scalar('grad_avg/{}'.format(name), avg_grad.item(), total_steps)
# # tbwriter.add_histogram('grad/{}'.format(name),
# # parameter.grad.cpu().numpy(), total_steps)
# if parameter.data is not None:
# avg_weight = torch.mean(parameter.data)
# # s('\t{} - param_avg: {}'.format(name, avg_weight))
# # tbwriter.add_histogram('weight/{}'.format(name),
# # parameter.data.cpu().numpy(), total_steps)
# # tbwriter.add_scalar('weight_avg/{}'.format(name), avg_weight.item(), total_steps)
total_steps += 1
time_end = time.time()
time_total = time_end - time_start
print('total time: ', time_total)
print('imgs per second: ', BATCH_SIZE * total_steps / time_total)
# # save checkpoints
# checkpoint_path = os.path.join(CHECKPOINT_DIR, 'alexnet_states_e{}.pkl'.format(epoch + 1))
# state = {
# 'epoch': epoch,
# 'total_steps': total_steps,
# 'optimizer': optimizer.state_dict(),
# 'model': alexnet.state_dict(),
# 'seed': seed,
# }
# torch.save(state, checkpoint_path)
# # save model
# if dist.get_rank() == 0:
# torch.save(alexnet.module.state_dict(), "%d.ckpt" % epoch)
# time_end = time.time()
# time_total = time_end - time_start
# print('total time: ', time_total)
# print('imgs per second: ', BATCH_SIZE * total_steps / time_total)