In the provided implementation of Resnet (https://github.com/pytorch/vision/blob/master/torchvision/models/resnet.py) how the order of layers(Blocks) for gradient flow during backprop is determined? Is it the exact reverse of the order defined in the def forward.
I am asking this question because I want to load pre-trained networks weights and then want to skip the last block during both forward and backward prop.
For doing so, I end up writing if else statement in the def forward of the resent. Now I am sure what is happening during backward prop. (https://github.com/deepaksharma36/Invasive_Plants_Detector/blob/master/python/src/models/resnet_surgery.py)