-
Notifications
You must be signed in to change notification settings - Fork 28
/
models.py
214 lines (179 loc) · 7.38 KB
/
models.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
import torch
import torch.nn as nn
import torch.nn.functional as F
import mlconfig
import torchvision
mlconfig.register(torchvision.models.resnet50)
mlconfig.register(torch.optim.SGD)
mlconfig.register(torch.optim.Adam)
mlconfig.register(torch.optim.lr_scheduler.MultiStepLR)
mlconfig.register(torch.optim.lr_scheduler.CosineAnnealingLR)
mlconfig.register(torch.optim.lr_scheduler.StepLR)
mlconfig.register(torch.optim.lr_scheduler.ExponentialLR)
class ConvBrunch(nn.Module):
def __init__(self, in_planes, out_planes, kernel_size=3):
super(ConvBrunch, self).__init__()
padding = (kernel_size - 1) // 2
self.out_conv = nn.Sequential(
nn.Conv2d(in_planes, out_planes, kernel_size=kernel_size, padding=padding),
nn.BatchNorm2d(out_planes),
nn.ReLU())
def forward(self, x):
return self.out_conv(x)
@mlconfig.register
class ToyModel(nn.Module):
def __init__(self, type='CIFAR10'):
super(ToyModel, self).__init__()
self.type = type
if type == 'CIFAR10':
self.block1 = nn.Sequential(
ConvBrunch(3, 64, 3),
ConvBrunch(64, 64, 3),
nn.MaxPool2d(kernel_size=2, stride=2))
self.block2 = nn.Sequential(
ConvBrunch(64, 128, 3),
ConvBrunch(128, 128, 3),
nn.MaxPool2d(kernel_size=2, stride=2))
self.block3 = nn.Sequential(
ConvBrunch(128, 196, 3),
ConvBrunch(196, 196, 3),
nn.MaxPool2d(kernel_size=2, stride=2))
# self.global_avg_pool = nn.AdaptiveAvgPool2d(1)
self.fc1 = nn.Sequential(
nn.Linear(4*4*196, 256),
nn.BatchNorm1d(256),
nn.ReLU())
self.fc2 = nn.Linear(256, 10)
self.fc_size = 4*4*196
elif type == 'MNIST':
self.block1 = nn.Sequential(
ConvBrunch(1, 32, 3),
nn.MaxPool2d(kernel_size=2, stride=2))
self.block2 = nn.Sequential(
ConvBrunch(32, 64, 3),
nn.MaxPool2d(kernel_size=2, stride=2))
# self.global_avg_pool = nn.AdaptiveAvgPool2d(1)
self.fc1 = nn.Sequential(
nn.Linear(64*7*7, 128),
nn.BatchNorm1d(128),
nn.ReLU())
self.fc2 = nn.Linear(128, 10)
self.fc_size = 64*7*7
self._reset_prams()
def _reset_prams(self):
for m in self.modules():
if isinstance(m, nn.Conv2d):
nn.init.kaiming_uniform_(m.weight, mode='fan_in', nonlinearity='relu')
elif isinstance(m, nn.Linear):
nn.init.xavier_uniform_(m.weight)
return
def forward(self, x):
x = self.block1(x)
x = self.block2(x)
x = self.block3(x) if self.type == 'CIFAR10' else x
# x = self.global_avg_pool(x)
# x = x.view(x.shape[0], -1)
x = x.view(-1, self.fc_size)
x = self.fc1(x)
x = self.fc2(x)
return x
'''ResNet in PyTorch.
For Pre-activation ResNet, see 'preact_resnet.py'.
Reference:
[1] Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun
Deep Residual Learning for Image Recognition. arXiv:1512.03385
'''
class BasicBlock(nn.Module):
expansion = 1
def __init__(self, in_planes, planes, stride=1):
super(BasicBlock, self).__init__()
self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)
self.bn1 = nn.BatchNorm2d(planes)
self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=1, padding=1, bias=False)
self.bn2 = nn.BatchNorm2d(planes)
self.shortcut = nn.Sequential()
if stride != 1 or in_planes != self.expansion*planes:
self.shortcut = nn.Sequential(
nn.Conv2d(in_planes, self.expansion*planes, kernel_size=1, stride=stride, bias=False),
nn.BatchNorm2d(self.expansion*planes)
)
def forward(self, x):
out = F.relu(self.bn1(self.conv1(x)))
out = self.bn2(self.conv2(out))
out += self.shortcut(x)
out = F.relu(out)
return out
class Bottleneck(nn.Module):
expansion = 4
def __init__(self, in_planes, planes, stride=1):
super(Bottleneck, self).__init__()
self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=1, bias=False)
self.bn1 = nn.BatchNorm2d(planes)
self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)
self.bn2 = nn.BatchNorm2d(planes)
self.conv3 = nn.Conv2d(planes, self.expansion*planes, kernel_size=1, bias=False)
self.bn3 = nn.BatchNorm2d(self.expansion*planes)
self.shortcut = nn.Sequential()
if stride != 1 or in_planes != self.expansion*planes:
self.shortcut = nn.Sequential(
nn.Conv2d(in_planes, self.expansion*planes, kernel_size=1, stride=stride, bias=False),
nn.BatchNorm2d(self.expansion*planes)
)
def forward(self, x):
out = F.relu(self.bn1(self.conv1(x)))
out = F.relu(self.bn2(self.conv2(out)))
out = self.bn3(self.conv3(out))
out += self.shortcut(x)
out = F.relu(out)
return out
class ResNet(nn.Module):
def __init__(self, block, num_blocks, num_classes=10):
super(ResNet, self).__init__()
self.in_planes = 64
self.conv1 = nn.Conv2d(3, 64, kernel_size=3, stride=1, padding=1, bias=False)
self.bn1 = nn.BatchNorm2d(64)
self.layer1 = self._make_layer(block, 64, num_blocks[0], stride=1)
self.layer2 = self._make_layer(block, 128, num_blocks[1], stride=2)
self.layer3 = self._make_layer(block, 256, num_blocks[2], stride=2)
self.layer4 = self._make_layer(block, 512, num_blocks[3], stride=2)
self.linear = nn.Linear(512*block.expansion, num_classes)
self._reset_prams()
def _make_layer(self, block, planes, num_blocks, stride):
strides = [stride] + [1]*(num_blocks-1)
layers = []
for stride in strides:
layers.append(block(self.in_planes, planes, stride))
self.in_planes = planes * block.expansion
return nn.Sequential(*layers)
def forward(self, x):
out = F.relu(self.bn1(self.conv1(x)))
out = self.layer1(out)
out = self.layer2(out)
out = self.layer3(out)
out = self.layer4(out)
out = F.avg_pool2d(out, 4)
out = out.view(out.size(0), -1)
out = self.linear(out)
return out
def _reset_prams(self):
for m in self.modules():
if isinstance(m, nn.Conv2d):
nn.init.kaiming_uniform_(m.weight, mode='fan_in', nonlinearity='relu')
elif isinstance(m, nn.Linear):
nn.init.xavier_uniform_(m.weight)
return
@mlconfig.register
def ResNet18(num_classes=10):
return ResNet(BasicBlock, [2, 2, 2, 2], num_classes=num_classes)
@mlconfig.register
def ResNet34(num_classes=10):
return ResNet(BasicBlock, [3, 4, 6, 3], num_classes=num_classes)
@mlconfig.register
def ResNet50(num_classes=10):
return ResNet(Bottleneck, [3, 4, 6, 3], num_classes=num_classes)
@mlconfig.register
def ResNet101(num_classes=10):
return ResNet(Bottleneck, [3, 4, 23, 3], num_classes=num_classes)
@mlconfig.register
def ResNet152(num_classes=10):
return ResNet(Bottleneck, [3, 8, 36, 3], num_classes=num_classes)