search

UMich-CURLY / SE3ET

SE3ET: SE(3)-Equivariant Transformer for Low-Overlap Point Cloud Registration
9 stars 0 forks source link

Issues with loading weights #1

Open ishaanmht opened 4 weeks ago

ishaanmht commented 4 weeks ago

Hello,

I would like to congratulate authors for such an amazing work. I was trying to run your demo code but I have been facing some issues with using the pretrained weights that you have provided. Would you have any suggestions:

main()
  File "demo.py", line 60, in main
    state_dict = torch.load(args.weights)
  File "/home/glenn/anaconda3/envs/se3et/lib/python3.8/site-packages/torch/serialization.py", line 592, in load
    return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args)
  File "/home/glenn/anaconda3/envs/se3et/lib/python3.8/site-packages/torch/serialization.py", line 851, in _load
    result = unpickler.load()
ModuleNotFoundError: No module named 'torch._tensor'
chienerh commented 3 weeks ago

Hi Ishaan, Thank you for your interest in the work. Could you provide more information on how you run the code? For example, which experiment folder are you running, and which pre-trained weight are you loading? I'll try to repeat the steps to see if it is a PyTorch version issue. Thank you!