File size: 2,537 Bytes
9f13819 |
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 |
import torch.nn as nn
import torch.nn.functional as F
class LambdaLayer(nn.Module):
def __init__(self, lambd):
super(LambdaLayer, self).__init__()
self.lambd = lambd
def forward(self, x):
return self.lambd(x)
class Block(nn.Module):
expansion = 1
def __init__(self, in_planes, planes, conv_layer, stride=1):
super(Block, self).__init__()
self.conv1 = conv_layer(in_planes, planes, kernel_size=3, stride=stride, padding=1, bias=False)
self.bn1 = nn.BatchNorm2d(planes)
self.conv2 = conv_layer(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 != planes:
diff = planes - in_planes
self.shortcut = LambdaLayer(
lambda x: F.pad(x[:, :, ::2, ::2], (0, 0, 0, 0, int(diff * 0.5), int((diff + 1) * 0.5)), "constant", 0))
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 Router(nn.Module):
def __init__(self, block, num_blocks, num_experts=2):
super(Router, self).__init__()
self.in_planes = 16
self.conv_layer = nn.Conv2d
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.layer1 = self._make_layer(block, 16, num_blocks[0], stride=1)
self.layer2 = self._make_layer(block, 32, num_blocks[1], stride=2)
self.layer3 = self._make_layer(block, 64, num_blocks[2], stride=2)
self.fc = nn.Linear(64, num_experts)
self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
def _make_layer(self, block, planes, num_blocks, stride):
planes = planes
strides = [stride] + [1] * (num_blocks - 1)
layers = []
for stride in strides:
layers.append(block(self.in_planes, planes, self.conv_layer, 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.avgpool(out)
out = out.view(out.size(0), -1)
out = self.fc(out)
return out
def build_router(**kwargs):
return Router(Block, [3, 3, 3], **kwargs)
|