A few question about sharing weights of convolution kernels and dilation convlution in pytorch

1,Could you tell me how can I achieve multiple convolution kernels sharing the same weight? If I want to make 4 conv to share 1 conv weight.
for example, I want to keep conv1 and conv2,3,4 using the same weights(using the weight of conv1)
self.conv1 = nn.Conv2d(3, 3, 3)
self.conv2 = nn.Conv2d(3, 3, 3)
self.conv3 = nn.Conv2d(3, 3, 3)
self.conv4 = nn.Conv2d(3, 3, 3)

2,May I ask whether the implementation of dilated convolution is to generate a convolution kernel with a dilation rate = r as in the original paper or to periodically sample the feature map at equal intervals, then convolve with a common convolution kernel, and then resample the image Spell back the original feature map size(As mentioned in “Smoothed Dilated Convolutions for Improved Dense Prediction”?)?

Thank you in advance!