def __init__(self, inplanes, planes, stride=1, downsample=None):
super(BasicBlock, self).__init__()
self.conv1 = conv3x3(inplanes, planes, stride)
self.bn1 = layer.BatchNorm2d(planes)
self.conv2 = conv3x3(planes, planes)
self.bn2 = layer.BatchNorm2d(planes)
self.relu1 = layer.ReLU()
self.add = layer.Add()
self.relu2 = layer.ReLU()
self.downsample = downsample
self.stride = stride
def _make_layer(self, block, planes, blocks, stride=1):
downsample = None
if stride != 1 or self.inplanes != planes * block.expansion:
conv = layer.Conv2d(
self.inplanes,
planes * block.expansion,
1,
stride=stride,
bias=False,
)
bn = layer.BatchNorm2d(planes * block.expansion)
def _downsample(x):
return bn(conv(x))
downsample = _downsample
layers = []
layers.append(block(self.inplanes, planes, stride, downsample))
self.inplanes = planes * block.expansion
for i in range(1, blocks):
layers.append(block(self.inplanes, planes))
def forward(x):
for layer in layers:
x = layer(x)
return x
return forward, layers
def __init__(self, block, layers, num_classes=10, num_channels=3):
self.inplanes = 64
super(ResNet, self).__init__()
self.num_classes = num_classes
self.input_size = 224
self.dimension = 4
self.conv1 = layer.Conv2d(num_channels,
64,
7,
stride=2,
padding=3,
bias=False)
self.bn1 = layer.BatchNorm2d(64)
self.relu = layer.ReLU()
self.maxpool = layer.MaxPool2d(kernel_size=3, stride=2, padding=1)
self.layer1, layers1 = self._make_layer(block, 64, layers[0])
self.layer2, layers2 = self._make_layer(block, 128, layers[1], stride=2)
self.layer3, layers3 = self._make_layer(block, 256, layers[2], stride=2)
self.layer4, layers4 = self._make_layer(block, 512, layers[3], stride=2)
self.avgpool = layer.AvgPool2d(7, stride=1)
self.flatten = layer.Flatten()
self.fc = layer.Linear(num_classes)
self.softmax_cross_entropy = layer.SoftMaxCrossEntropy()
self.register_layers(*layers1, *layers2, *layers3, *layers4)
def __init__(self, inplanes, planes, stride=1, downsample=None):
super(Bottleneck, self).__init__()
self.conv1 = layer.Conv2d(inplanes, planes, 1, bias=False)
self.bn1 = layer.BatchNorm2d(planes)
self.relu1 = layer.ReLU()
self.conv2 = layer.Conv2d(planes,
planes,
3,
stride=stride,
padding=1,
bias=False)
self.bn2 = layer.BatchNorm2d(planes)
self.relu2 = layer.ReLU()
self.conv3 = layer.Conv2d(planes,
planes * self.expansion,
1,
bias=False)
self.bn3 = layer.BatchNorm2d(planes * self.expansion)
self.add = layer.Add()
self.relu3 = layer.ReLU()
self.downsample = downsample
self.stride = stride
def __init__(self):
super(MyModel, self).__init__()
self.conv1 = layer.Conv2d(2, 2)
self.bn1 = layer.BatchNorm2d(2)
self.doublelinear1 = DoubleLinear(2, 4, 2)
self.optimizer = opt.SGD()
def __init__(self,
in_filters,
out_filters,
reps,
strides=1,
padding=0,
start_with_relu=True,
grow_first=True):
super(Block, self).__init__()
if out_filters != in_filters or strides != 1:
self.skip = layer.Conv2d(in_filters,
out_filters,
1,
stride=strides,
padding=padding,
bias=False)
self.skipbn = layer.BatchNorm2d(out_filters)
else:
self.skip = None
self.layers = []
filters = in_filters
if grow_first:
self.layers.append(layer.ReLU())
self.layers.append(
layer.SeparableConv2d(in_filters,
out_filters,
3,
stride=1,
padding=1,
bias=False))
self.layers.append(layer.BatchNorm2d(out_filters))
filters = out_filters
for i in range(reps - 1):
self.layers.append(layer.ReLU())
self.layers.append(
layer.SeparableConv2d(filters,
filters,
3,
stride=1,
padding=1,
bias=False))
self.layers.append(layer.BatchNorm2d(filters))
if not grow_first:
self.layers.append(layer.ReLU())
self.layers.append(
layer.SeparableConv2d(in_filters,
out_filters,
3,
stride=1,
padding=1,
bias=False))
self.layers.append(layer.BatchNorm2d(out_filters))
if not start_with_relu:
self.layers = self.layers[1:]
else:
self.layers[0] = layer.ReLU()
if strides != 1:
self.layers.append(layer.MaxPool2d(3, strides, padding + 1))
self.register_layers(*self.layers)
self.add = layer.Add()
def __init__(self, num_classes=10, num_channels=3):
""" Constructor
Args:
num_classes: number of classes
"""
super(Xception, self).__init__()
self.num_classes = num_classes
self.input_size = 299
self.dimension = 4
self.conv1 = layer.Conv2d(num_channels, 32, 3, 2, 0, bias=False)
self.bn1 = layer.BatchNorm2d(32)
self.relu1 = layer.ReLU()
self.conv2 = layer.Conv2d(32, 64, 3, 1, 1, bias=False)
self.bn2 = layer.BatchNorm2d(64)
self.relu2 = layer.ReLU()
# do relu here
self.block1 = Block(64,
128,
2,
2,
padding=0,
start_with_relu=False,
grow_first=True)
self.block2 = Block(128,
256,
2,
2,
padding=0,
start_with_relu=True,
grow_first=True)
self.block3 = Block(256,
728,
2,
2,
padding=0,
start_with_relu=True,
grow_first=True)
self.block4 = Block(728,
728,
3,
1,
start_with_relu=True,
grow_first=True)
self.block5 = Block(728,
728,
3,
1,
start_with_relu=True,
grow_first=True)
self.block6 = Block(728,
728,
3,
1,
start_with_relu=True,
grow_first=True)
self.block7 = Block(728,
728,
3,
1,
start_with_relu=True,
grow_first=True)
self.block8 = Block(728,
728,
3,
1,
start_with_relu=True,
grow_first=True)
self.block9 = Block(728,
728,
3,
1,
start_with_relu=True,
grow_first=True)
self.block10 = Block(728,
728,
3,
1,
start_with_relu=True,
grow_first=True)
self.block11 = Block(728,
728,
3,
1,
start_with_relu=True,
grow_first=True)
self.block12 = Block(728,
1024,
2,
2,
start_with_relu=True,
grow_first=False)
self.conv3 = layer.SeparableConv2d(1024, 1536, 3, 1, 1)
self.bn3 = layer.BatchNorm2d(1536)
self.relu3 = layer.ReLU()
# do relu here
self.conv4 = layer.SeparableConv2d(1536, 2048, 3, 1, 1)
self.bn4 = layer.BatchNorm2d(2048)
self.relu4 = layer.ReLU()
self.globalpooling = layer.MaxPool2d(10, 1)
self.flatten = layer.Flatten()
self.fc = layer.Linear(num_classes)
self.softmax_cross_entropy = layer.SoftMaxCrossEntropy()