how does your implementation share the weight？

[pzz2011]

Hi, there, I don't find the any code to evident that the parameter is shared. Maybe becanse I don't I understand how to use the "weight shared function" of pytorch? Can u help me? thanks.

class DRRN(nn.Module):
    def __init__(self):
        super(DRRN, self).__init__()
        self.input = nn.Conv2d(in_channels=1, out_channels=128, kernel_size=3, stride=1, padding=1, bias=False)
        self.conv1 = nn.Conv2d(in_channels=128, out_channels=128, kernel_size=3, stride=1, padding=1, bias=False)
        self.conv2 = nn.Conv2d(in_channels=128, out_channels=128, kernel_size=3, stride=1, padding=1, bias=False)
        self.output = nn.Conv2d(in_channels=128, out_channels=1, kernel_size=3, stride=1, padding=1, bias=False)
        self.relu = nn.ReLU(inplace=True)

        # weights initialization
        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
                m.weight.data.normal_(0, sqrt(2. / n))

    def forward(self, x):
        residual = x
        inputs = self.input(self.relu(x))
        out = inputs
        for _ in range(25):
            out = self.conv2(self.relu(self.conv1(self.relu(out))))
            out = torch.add(out, inputs)

        out = self.output(self.relu(out))
        out = torch.add(out, residual)
return out

[jt827859032]

Hi, @pzz2011 You can regard conv = nn.Conv2D as an instantiation of a convolutional layer. The self.conv1 and self.conv2 are 2 instantiations defined in the __init__ function and I am continuously reusing these two instantiations in the forward implementation (see in the for loop).

for _ in range(25):
    out = self.conv2(self.relu(self.conv1(self.relu(out))))
    out = torch.add(out, inputs)

If you wanna build 3 convolutional layers with different weights, you should define 3 instantiations by utilizing nn.Conv2d:

    conv1 = nn.Conv2d(xxx)
    conv2 = nn.Conv2d(xxx)
    conv3 = nn.Conv2d(xxx)