About Resnet downsample part in github code

[haoyu12S]

under the official pytorch code is def _make_layer(self, block, planes, blocks, stride=1): downsample = None if stride != 1 or self.inplanes != planes * block.expansion: downsample = nn.Sequential( conv1x1(self.inplanes, planes * block.expansion, stride), nn.BatchNorm2d(planes * block.expansion), )

But if you check the paper https://arxiv.org/abs/1512.03385 the downsample convolution kernel size is 3x3

what is the difference influence the output? Pytorch change the idea or there is the incorrect codes. Thanks