I am trying to load a part of a model that was trained in https://github.com/vita-epfl/openpifpaf. The method torch.load()
works when I am loading my own models saved with the extension .pth.tar
. Trying it with a .pkl
file however gives me an error message.
net = torch.load('./models/resnet50block5-pif-paf-edge401-190315-214317.pkl')
net = torch.load('./models/resnet50block5-pif-paf-edge401-190315-214317.pkl')
File "/home/haziq/.local/lib/python3.7/site-packages/torch/serialization.py", line 368, in load
return _load(f, map_location, pickle_module)
File "/home/haziq/.local/lib/python3.7/site-packages/torch/serialization.py", line 542, in _load
result = unpickler.load()
ModuleNotFoundError: No module named 'openpifpaf'
Is it possible to simply load part of the model without requiring the source code?
In case this makes it easier, the model is a bunch of convolutional layers built on top of ResNet. I actually only need the ResNet backbone that has been further trained on the COCO Keypoints dataset.