Unreferrenced function in Pytorch official code

John_Deterious · March 11, 2020, 6:29am

I’m reading ResNet code from official website, however, I see this norm_layer used without ever defined:

pytorch/vision/blob/2611f5cc79230f3729393279093775d08bc52718/torchvision/models/resnet.py#L50


             base_width=64, dilation=1, norm_layer=None):
    super(BasicBlock, self).__init__()
    if norm_layer is None:
        norm_layer = nn.BatchNorm2d
    if groups != 1 or base_width != 64:
        raise ValueError('BasicBlock only supports groups=1 and base_width=64')
    if dilation > 1:
        raise NotImplementedError("Dilation > 1 not supported in BasicBlock")
    # Both self.conv1 and self.downsample layers downsample the input when stride != 1
    self.conv1 = conv3x3(inplanes, planes, stride)
    self.bn1 = norm_layer(planes)
    self.relu = nn.ReLU(inplace=True)
    self.conv2 = conv3x3(planes, planes)
    self.bn2 = norm_layer(planes)
    self.downsample = downsample
    self.stride = stride


def forward(self, x):
    identity = x


    out = self.conv1(x)

IliasPap · March 11, 2020, 8:09am

It is defined in line 43 otherwise it is None

ptrblck · March 11, 2020, 8:10am

norm_layer is defines as nn.BatchNorm2d by default here and passed to the blocks here.