how to modify code if I use resnet18 backbone?

[sunmengnan]

There is error here:

return torch.stack(_features, dim=1)

RuntimeError: stack expects each tensor to be equal size, but got [4608, 128] at entry 0 and [4608, 256] at entry 1