def conv3x3mtl(in_planes, out_planes, stride=1):
"""3x3 convolution with padding"""
return Conv2dMtl(in_planes,
out_planes,
kernel_size=3,
stride=stride,
padding=1,
bias=False)
def __init__(self, block, layers, num_classes=10):
self.inplanes = 16
super(ResNetMtl, self).__init__()
self.conv1 = Conv2dMtl(3, 16, kernel_size=3, stride=1, padding=1,
bias=False)
self.bn1 = nn.BatchNorm2d(16)
self.relu = nn.ReLU(inplace=True)
self.layer1 = self._make_layer(block, 16, layers[0])
self.layer2 = self._make_layer(block, 32, layers[1], stride=2)
self.layer3 = self._make_layer(block, 64, layers[2], stride=2, last_phase=True)
self.avgpool = nn.AvgPool2d(8, stride=1)
self.fc = modified_linear.CosineLinear(64 * block.expansion, num_classes)
for m in self.modules():
if isinstance(m, Conv2dMtl):
nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')
elif isinstance(m, nn.BatchNorm2d):
nn.init.constant_(m.weight, 1)
nn.init.constant_(m.bias, 0)
def _make_layer(self, block, planes, blocks, stride=1, last_phase=False):
downsample = None
if stride != 1 or self.inplanes != planes * block.expansion:
downsample = nn.Sequential(
Conv2dMtl(self.inplanes, planes * block.expansion,
kernel_size=1, stride=stride, bias=False),
nn.BatchNorm2d(planes * block.expansion),
)
layers = []
layers.append(block(self.inplanes, planes, stride, downsample))
self.inplanes = planes * block.expansion
if last_phase:
for i in range(1, blocks-1):
layers.append(block(self.inplanes, planes))
layers.append(block(self.inplanes, planes, last=True))
else:
for i in range(1, blocks):
layers.append(block(self.inplanes, planes))
return nn.Sequential(*layers)
def conv3x3mtl(in_planes, out_planes, stride=1):
return Conv2dMtl(in_planes, out_planes, kernel_size=3, stride=stride,
padding=1, bias=False)