ResNet was proposed in “Deep Residual Learning for Image Recognition.”
There is a following paper named “Identity mappings in deep residual networks” which suggests a “pre-act” version of the ResNet, usually known as ResNetV2.
I have yet to find an official implementation of ResNetV2, is there any?
From what I can find in the doc here: https://pytorch.org/docs/stable/torchvision/models.html we have mostly resnet and wide resnet from there.
Maybe @fmassa has a better idea about this?
Sadly, none is preact-ResNet . The network is quite well-known, and outperforms the original ResNet. Would it be a good idea to add into the catalog?
I am not the best person to judge this You can try to open a feature request on the torchvision repo and see what they say!