-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathvggnet.py
171 lines (119 loc) · 6.52 KB
/
vggnet.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
import torch
import torch.nn as nn
import models.classifier_block as classifier
import models.smoothing_block as smoothing
import models.vggnet_dnn_block as vggnet_dnn
import models.vggnet_mcdo_block as vggnet_mcdo
class VGGNet(nn.Module):
def __init__(self, block, num_blocks,
sblock=smoothing.TanhBlurBlock, num_sblocks=(0, 0, 0, 0, 0),
cblock=classifier.MLPBlock,
num_classes=10, name="vgg", **block_kwargs):
super(VGGNet, self).__init__()
self.name = name
self.layer0 = self._make_layer(block, 3, 64, num_blocks[0], pool=False, **block_kwargs)
self.layer1 = self._make_layer(block, 64, 128, num_blocks[1], pool=True, **block_kwargs)
self.layer2 = self._make_layer(block, 128, 256, num_blocks[2], pool=True, **block_kwargs)
self.layer3 = self._make_layer(block, 256, 512, num_blocks[3], pool=True, **block_kwargs)
self.layer4 = self._make_layer(block, 512, 512, num_blocks[4], pool=True, **block_kwargs)
self.smooth0 = self._make_smooth_layer(sblock, 64, num_sblocks[0], **block_kwargs)
self.smooth1 = self._make_smooth_layer(sblock, 128, num_sblocks[1], **block_kwargs)
self.smooth2 = self._make_smooth_layer(sblock, 256, num_sblocks[2], **block_kwargs)
self.smooth3 = self._make_smooth_layer(sblock, 512, num_sblocks[3], **block_kwargs)
self.smooth4 = self._make_smooth_layer(sblock, 512, num_sblocks[4], **block_kwargs)
self.classifier = []
if cblock is classifier.MLPBlock:
self.classifier.append(nn.MaxPool2d(kernel_size=2, stride=2))
self.classifier.append(nn.AdaptiveAvgPool2d((7, 7)))
self.classifier.append(cblock(7 * 7 * 512, num_classes, **block_kwargs))
else:
self.classifier.append(cblock(512, num_classes, **block_kwargs))
self.classifier = nn.Sequential(*self.classifier)
@staticmethod
def _make_layer(block, in_channels, out_channels, num_blocks, pool, **block_kwargs):
layers, channels = [], in_channels
if pool:
layers.append(nn.MaxPool2d(kernel_size=2, stride=2))
for _ in range(num_blocks):
layers.append(block(channels, out_channels, **block_kwargs))
channels = out_channels
return nn.Sequential(*layers)
@staticmethod
def _make_smooth_layer(sblock, in_filters, num_blocks, **block_kwargs):
layers = []
for _ in range(num_blocks):
layers.append(sblock(in_filters=in_filters, **block_kwargs))
return nn.Sequential(*layers)
def forward(self, x):
x = self.layer0(x)
x = self.smooth0(x)
x = self.layer1(x)
x = self.smooth1(x)
x = self.layer2(x)
x = self.smooth2(x)
x = self.layer3(x)
x = self.smooth3(x)
x = self.layer4(x)
x = self.smooth4(x)
x = self.classifier(x)
return x
# Deterministic
def dnn_11(num_classes=10, name="vgg_dnn_11", **block_kwargs):
return VGGNet(vggnet_dnn.BasicBlock, [1, 1, 2, 2, 2],
num_classes=num_classes, name=name, **block_kwargs)
def dnn_13(num_classes=10, name="vgg_dnn_13", **block_kwargs):
return VGGNet(vggnet_dnn.BasicBlock, [2, 2, 2, 2, 2],
num_classes=num_classes, name=name, **block_kwargs)
def dnn_16(num_classes=10, name="vgg_dnn_16", **block_kwargs):
return VGGNet(vggnet_dnn.BasicBlock, [2, 2, 3, 3, 3],
num_classes=num_classes, name=name, **block_kwargs)
def dnn_19(num_classes=10, name="vgg_dnn_19", **block_kwargs):
return VGGNet(vggnet_dnn.BasicBlock, [2, 2, 4, 4, 4],
num_classes=num_classes, name=name, **block_kwargs)
# MC dropout
def mcdo_11(num_classes=10, name="vgg_mcdo_11", **block_kwargs):
return VGGNet(vggnet_mcdo.BasicBlock, [1, 1, 2, 2, 2],
num_classes=num_classes, name=name, **block_kwargs)
def mcdo_13(num_classes=10, name="vgg_mcdo_13", **block_kwargs):
return VGGNet(vggnet_mcdo.BasicBlock, [2, 2, 2, 2, 2],
num_classes=num_classes, name=name, **block_kwargs)
def mcdo_16(num_classes=10, name="vgg_mcdo_16", **block_kwargs):
return VGGNet(vggnet_mcdo.BasicBlock, [2, 2, 3, 3, 3],
num_classes=num_classes, name=name, **block_kwargs)
def mcdo_19(num_classes=10, name="vgg_mcdo_19", **block_kwargs):
return VGGNet(vggnet_mcdo.BasicBlock, [2, 2, 4, 4, 4],
num_classes=num_classes, name=name, **block_kwargs)
# Deterministic + Smoothing
def dnn_smooth_11(num_classes=10, name="vgg_dnn_smoothing_11", **block_kwargs):
return VGGNet(vggnet_dnn.BasicBlock, [1, 1, 2, 2, 2],
num_sblocks=[1, 1, 1, 1, 1],
num_classes=num_classes, name=name, **block_kwargs)
def dnn_smooth_13(num_classes=10, name="vgg_dnn_smoothing_13", **block_kwargs):
return VGGNet(vggnet_dnn.BasicBlock, [2, 2, 2, 2, 2],
num_sblocks=[1, 1, 1, 1, 1],
num_classes=num_classes, name=name, **block_kwargs)
def dnn_smooth_16(num_classes=10, name="vgg_dnn_smoothing_16", **block_kwargs):
return VGGNet(vggnet_dnn.BasicBlock, [2, 2, 3, 3, 3],
num_sblocks=[1, 1, 1, 1, 1],
num_classes=num_classes, name=name, **block_kwargs)
def dnn_smooth_19(num_classes=10, name="vgg_dnn_smoothing_19", **block_kwargs):
return VGGNet(vggnet_dnn.BasicBlock, [2, 2, 4, 4, 4],
num_sblocks=[1, 1, 1, 1, 1],
num_classes=num_classes, name=name, **block_kwargs)
# MC dropout + Smoothing
def mcdo_smooth_11(num_classes=10, name="vgg_mcdo_smoothing_11", **block_kwargs):
return VGGNet(vggnet_mcdo.BasicBlock, [1, 1, 2, 2, 2],
num_sblocks=[1, 1, 1, 1, 1],
num_classes=num_classes, name=name, **block_kwargs)
def mcdo_smooth_13(num_classes=10, name="vgg_mcdo_smoothing_13", **block_kwargs):
return VGGNet(vggnet_mcdo.BasicBlock, [2, 2, 2, 2, 2],
num_sblocks=[1, 1, 1, 1, 1],
num_classes=num_classes, name=name, **block_kwargs)
def mcdo_smooth_16(num_classes=10, name="vgg_mcdo_smoothing_16", **block_kwargs):
return VGGNet(vggnet_mcdo.BasicBlock, [2, 2, 3, 3, 3],
num_sblocks=[1, 1, 1, 1, 1],
num_classes=num_classes, name=name, **block_kwargs)
def mcdo_smooth_19(num_classes=10, name="vgg_mcdo_smoothing_19", **block_kwargs):
return VGGNet(vggnet_mcdo.BasicBlock, [2, 2, 4, 4, 4],
num_sblocks=[1, 1, 1, 1, 1],
num_classes=num_classes, name=name, **block_kwargs)