1.当设置group=1时: conv = nn.Conv2d(in_channels=, out_channels=, kernel_size=, groups=) conv.weight.data.size() 返回: torch.Size([, , , ]) 另一个例子: conv = nn.Conv2d(in_channels=, out_channels=, kernel_size=, groups=) conv.weight.data.size() 返回: torch.Size([,