from torch import Tensor ... self.conv1 = conv3x3(inplanes, planes, stride) ... Bottleneck in torchvision places the stride for downsampling at 3x3 ...
確定! 回上一頁