FAMES/model.py at master · SJTU-ECTL/FAMES · GitHub

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
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.nn.init as init
from torch.autograd import Variable
from AM.AM_layer import AMConv2d
from torch.autograd import Function

class BasicBlock(nn.Module):
    expansion = 1

    def __init__(self, in_planes, planes,lut_list, stride=1):
        super(BasicBlock, self).__init__()
        # self.conv1 = nn.Conv2d(in_planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)
        self.conv1 = AMConv2d(in_planes, planes, kernel_size=3, stride=stride, padding=1, bias=False,file_path=lut_list[0])
        self.bn1 = nn.BatchNorm2d(planes)
        # self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=1, padding=1, bias=False)
        self.conv2 = AMConv2d(planes, planes, kernel_size=3, stride=1, padding=1, bias=False, file_path=lut_list[1])
        self.bn2 = nn.BatchNorm2d(planes)
        self.shortcut = nn.Sequential()
        if stride != 1 or in_planes != planes:
                self.shortcut = nn.Sequential(
                     # nn.Conv2d(in_planes, self.expansion * planes, kernel_size=1, stride=stride, bias=False),
                     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.bn1(self.conv1(x)))
        out = self.bn2(self.conv2(out))
        out += self.shortcut(x)
        out = F.relu(out)
        return out


class ResNet(nn.Module):
    def __init__(self, block, num_blocks,mul_list, num_classes=10):
        super(ResNet, self).__init__()
        self.in_planes = 16

        #self.conv1 = nn.Conv2d(3, 16, kernel_size=3, stride=1, padding=1, bias=False)
        step=0
        self.conv1 = AMConv2d(3, 16, kernel_size=3, stride=1, padding=1,bias=False, file_path=mul_list[step])
        step=step+1
        self.bn1 = nn.BatchNorm2d(16)
        self.layer1 = self._make_layer(block, 16, num_blocks[0], stride=1,lut_list=mul_list[step:step+2*num_blocks[0]])
        step=step+2*num_blocks[0]
        self.layer2 = self._make_layer(block, 32, num_blocks[1], stride=2,lut_list=mul_list[step:step+2*num_blocks[1]])
        step=step+2*num_blocks[1]
        self.layer3 = self._make_layer(block, 64, num_blocks[2], stride=2,lut_list=mul_list[step:step+2*num_blocks[2]])
        step=step+2*num_blocks[2]
        # self.linear = nn.Linear(64, num_classes)
        self.linear = nn.Linear(64, num_classes)

    def _make_layer(self, block, planes, num_blocks,lut_list, stride):
        strides = [stride] + [1]*(num_blocks-1)
        layers = []
        for idx,stride in enumerate(strides):
            layers.append(block(self.in_planes, planes,lut_list[2*idx:2*idx+2], stride))
            self.in_planes = planes * block.expansion

        return nn.Sequential(*layers)

    def forward(self, x):
        # out = F.relu(self.bn1(self.conv1(x)))
        out = F.relu(self.bn1(self.conv1(x)))
        out = self.layer1(out)
        out = self.layer2(out)
        out = self.layer3(out)
        out = F.avg_pool2d(out, 8)
        out = out.view(out.size(0), -1)
        out = self.linear(out)
        return out

def resnet14(mul):
        return ResNet(BasicBlock, [2, 2, 2],mul)

def resnet8(mul):
        return ResNet(BasicBlock, [1, 1, 1],mul)

def resnet20(mul):
        return ResNet(BasicBlock, [3, 3, 3],mul)