small_models.py

from models import *


class SmallResNet(nn.Module):

    def __init__(self, block, layers, num_classes=10, zero_init_residual=False,
                 groups=1, width_per_group=64, replace_stride_with_dilation=None,
                 norm_layer=None, norm_moment=0.9, split=False):
        super(SmallResNet, self).__init__()
        if norm_layer is None:
            norm_layer = nn.BatchNorm2d
        self._norm_layer = norm_layer
        self.split = split

        self.inplanes = 64
        self.dilation = 1
        if replace_stride_with_dilation is None:
            # each element in the tuple indicates if we should replace
            # the 2x2 stride with a dilated convolution instead
            replace_stride_with_dilation = [False]
        if len(replace_stride_with_dilation) != 1:
            raise ValueError("replace_stride_with_dilation should be None "
                             "or a 1-element tuple, got {}".format(replace_stride_with_dilation))
        self.groups = groups
        self.base_width = width_per_group
        # using cifar so I change the first 4 layer!
        self.conv1 = nn.Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
        self.bn1 = norm_layer(self.inplanes, momentum=norm_moment)
        self.relu = nn.ReLU(inplace=True)
        # self.conv1 = nn.Conv2d(3, self.inplanes, kernel_size=7, stride=2, padding=3,
        #                        bias=False)
        # self.bn1 = norm_layer(self.inplanes, momentum=norm_moment)
        # self.relu = nn.ReLU(inplace=True)
        # self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
        self.layer1 = self._make_layer(block, 64, layers[0])
        self.layer2 = self._make_layer(block, 128, layers[1], stride=2,
                                       dilate=replace_stride_with_dilation[0])
        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
        # self.fc = nn.Linear(512 * block.expansion, num_classes)

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')
            elif isinstance(m, (nn.BatchNorm2d, nn.GroupNorm)):
                nn.init.constant_(m.weight, 1)
                nn.init.constant_(m.bias, 0)

        # Zero-initialize the last BN in each residual branch,
        # so that the residual branch starts with zeros, and each residual block behaves like an identity.
        # This improves the model by 0.2~0.3% according to https://arxiv.org/abs/1706.02677
        if zero_init_residual:
            for m in self.modules():
                if isinstance(m, Bottleneck):
                    nn.init.constant_(m.bn3.weight, 0)
                elif isinstance(m, BasicBlock):
                    nn.init.constant_(m.bn2.weight, 0)

    def _make_layer(self, block, planes, blocks, stride=1, dilate=False):
        norm_layer = self._norm_layer
        downsample = None
        previous_dilation = self.dilation
        if dilate:
            self.dilation *= stride
            stride = 1
        if stride != 1 or self.inplanes != planes * block.expansion:
            downsample = nn.Sequential(
                conv1x1(self.inplanes, planes * block.expansion, stride),
                norm_layer(planes * block.expansion, momentum=0.9),
            )

        layers = []
        layers.append(block(self.inplanes, planes, stride, downsample, self.groups,
                            self.base_width, previous_dilation, norm_layer))
        self.inplanes = planes * block.expansion
        for _ in range(1, blocks):
            layers.append(block(self.inplanes, planes, groups=self.groups,
                                base_width=self.base_width, dilation=self.dilation,
                                norm_layer=norm_layer))

        return nn.Sequential(*layers)

    def _forward_impl(self, x):
        # See note [TorchScript super()]
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)
        # x = self.maxpool(x)

        if self.split:
            f = self.avgpool(x)
            f = f.view(f.size(0), -1)

        x = self.layer1(x)
        x = self.layer2(x)

        x = self.avgpool(x)
        x = x.view(x.size(0), -1)
        # x = self.fc(x)

        if self.split:
            return f, x
        else:
            return x

    def forward(self, x):
        return self._forward_impl(x)


def _small_resnet(arch, block, layers, **kwargs):
    model = SmallResNet(block, layers, **kwargs)
    return model


def SmallResnet(**kwargs):
    return _small_resnet('smallresnet', BasicBlock, [1, 1], **kwargs)