-
Notifications
You must be signed in to change notification settings - Fork 3
/
our_network.py
164 lines (136 loc) · 6.89 KB
/
our_network.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
import torch
import torch.nn as nn
import torch.nn.functional as F
from quant_module import *
class BasicBlock(nn.Module):
expansion = 1
def __init__(self, lambda_s, w_bits, a_bits, in_planes, planes, stride=1, use_fp=True, activation_quant=True):
super(BasicBlock, self).__init__()
self.conv1 = Conv2d_minmax(lambda_s, w_bits, in_planes, planes, kernel_size=3, stride=stride, padding=1, bias=False, use_fp=use_fp)
self.bn1 = nn.BatchNorm2d(planes)
self.act1 = ReLU_quant(a_bits) if activation_quant else nn.ReLU()
self.conv2 = Conv2d_minmax(lambda_s, w_bits, planes, planes, kernel_size=3, stride=1, padding=1, bias=False, use_fp=use_fp)
self.bn2 = nn.BatchNorm2d(planes)
self.act2 = ReLU_quant(a_bits) if activation_quant else nn.ReLU()
self.downsample = nn.Sequential()
if stride != 1 or in_planes != self.expansion*planes:
self.downsample = nn.Sequential(
Conv2d_minmax(lambda_s, w_bits, in_planes, self.expansion*planes, kernel_size=1, stride=stride, padding=0, bias=False, use_fp=use_fp),
nn.BatchNorm2d(self.expansion*planes)
)
def forward(self, x):
out = self.act1(self.bn1(self.conv1(x)))
out = self.bn2(self.conv2(out))
out += self.downsample(x)
out = self.act2(out)
return out
class ResNet(nn.Module):
def __init__(self, block, num_blocks, num_classes=100, lambda_s=1, w_bits=4, a_bits=4, use_fp=True, activation_quant=True, quant_first_last=False):
super(ResNet, self).__init__()
self.in_planes = 16
self.w_bits = w_bits
self.a_bits = a_bits
self.lambda_s= lambda_s
self.act_quant = activation_quant
if quant_first_last: # Fixed to 8 bit for 3 bits or lower
fl_bits = 8 if w_bits < 4 else w_bits
self.conv1 = Conv2d_minmax(lambda_s, fl_bits, 3, self.in_planes, kernel_size=3, stride=1, padding=1, bias=False, use_fp=use_fp)
else:
self.conv1 = nn.Conv2d(3, self.in_planes, kernel_size=3, stride=1, padding=1, bias=False)
self.bn1 = nn.BatchNorm2d(self.in_planes)
self.act = ReLU_quant(a_bits) if activation_quant else nn.ReLU()
self.layer1 = self._make_layer(block, 16, num_blocks[0], stride=1, use_fp=use_fp)
self.layer2 = self._make_layer(block, 32, num_blocks[1], stride=2, use_fp=use_fp)
self.layer3 = self._make_layer(block, 64, num_blocks[2], stride=2, use_fp=use_fp)
self.avgpool = nn.AvgPool2d(kernel_size=8, stride=1)
if quant_first_last:
self.fc = Linear_minmax(lambda_s, fl_bits, 64 * block.expansion, num_classes, use_fp=use_fp)
else:
self.fc = nn.Linear(64, num_classes)
def _make_layer(self, block, planes, num_blocks, stride, use_fp):
layers = []
layers.append(block(self.lambda_s, self.w_bits, self.a_bits, self.in_planes, planes, stride, use_fp, self.act_quant))
self.in_planes = planes * block.expansion
for _ in range(1, num_blocks):
layers.append(block(self.lambda_s, self.w_bits, self.a_bits, self.in_planes, planes, 1, use_fp, self.act_quant))
return nn.Sequential(*layers)
def forward(self, x):
out = self.act(self.bn1(self.conv1(x)))
out = self.layer1(out)
out = self.layer2(out)
out = self.layer3(out)
out = self.avgpool(out)
out = out.view(out.size(0), -1)
out = self.fc(out)
return out
def ResNet32(w_bits, a_bits, lambda_s, use_fp=True, activation_quant=False, quant_first_last=False):
return ResNet(BasicBlock, [5,5,5], w_bits=w_bits, a_bits=a_bits, lambda_s=lambda_s, use_fp=use_fp, activation_quant=activation_quant, quant_first_last=quant_first_last)
cfg = {
'A': [64, 'M', 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],
'B': [64, 64, 'M', 128, 128, 'M', 256, 256, 'M', 512, 512, 'M', 512, 512, 'M'],
'D': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 'M', 512, 512, 512, 'M', 512, 512, 512, 'M'],
'E': [64, 64, 'M', 128, 128, 'M', 256, 256, 256, 256, 'M', 512, 512, 512, 512, 'M',
512, 512, 512, 512, 'M'],
}
class VGG(nn.Module):
'''
VGG model
'''
def __init__(self, features, q_cfg):
super(VGG, self).__init__()
self.features = features
self.q_cfg = q_cfg
self.quant_first_last = q_cfg['quant_fl']
self.use_fp = q_cfg['use_fp']
fl_bit = q_cfg['w_bits'] if q_cfg['w_bits'] > 3 else 8
fl_lambda_s = q_cfg['lambda_s']
if self.quant_first_last:
self.last_layer = Linear_minmax(fl_lambda_s, fl_bit, 4096, 100, use_fp=self.use_fp)
else:
self.last_layer = nn.Linear(4096, 100)
self.act = ReLU_quant(self.q_cfg['a_bits']) if self.q_cfg['activation_quant'] else nn.ReLU()
self.classifier = nn.Sequential(
Linear_minmax(self.q_cfg['lambda_s'], self.q_cfg['w_bits'], 512, 4096, use_fp=self.use_fp, bias=True),
self.act, # ReLUQuant
nn.Dropout(),
Linear_minmax(self.q_cfg['lambda_s'], self.q_cfg['w_bits'], 4096, 4096, use_fp=self.use_fp, bias=True),
self.act, #ReLUQuant x
nn.Dropout(),
self.last_layer,
)
def forward(self, x):
x = self.features(x)
x = x.view(x.size(0), -1)
x = self.classifier(x)
return x
def make_layers(cfg, q_cfg, batch_norm=False):
layers = []
in_channels = 3
activation = ReLU_quant(q_cfg['a_bits']) if q_cfg['activation_quant'] else nn.ReLU()
fl_bit = q_cfg['w_bits'] if q_cfg['w_bits'] > 3 else 8
fl_lambda_s = q_cfg['lambda_s']
if q_cfg['quant_fl']:
first_layer = Conv2d_minmax(fl_lambda_s, fl_bit, in_channels, cfg[0], kernel_size=3, padding=1, use_fp=q_cfg['use_fp'], bias=False)
else:
first_layer = nn.Conv2d(in_channels, cfg[0], kernel_size=3, padding=1, bias=False)
if batch_norm:
layers += [first_layer, nn.BatchNorm2d(cfg[0]), activation]
else:
layers += [first_layer, activation]
in_channels = cfg[0]
for v in cfg[1:]:
if v == 'M':
layers += [nn.MaxPool2d(kernel_size=2, stride=2)]
else:
conv2d = Conv2d_minmax(q_cfg['lambda_s'], q_cfg['w_bits'], in_channels, v, kernel_size=3, padding=1, use_fp=q_cfg['use_fp'], bias=False)
if batch_norm:
layers += [conv2d, nn.BatchNorm2d(v), activation]
else:
layers += [conv2d, activation]
in_channels = v
return nn.Sequential(*layers)
def Vgg16_bn(w_bits, a_bits, lambda_s, use_fp=True, activation_quant=True, quant_first_last=False):
"""VGG 16-layer model (configuration "D") with batch normalization"""
q_cfg = {'w_bits': w_bits, "a_bits": a_bits, "lambda_s": lambda_s, "use_fp": use_fp,\
"activation_quant": activation_quant, "quant_fl":quant_first_last}
return VGG(make_layers(cfg['D'], q_cfg, batch_norm=True), q_cfg)