Hi! Thanks for sharing this code, and great paper :)
Just an FYI, I tried grabbing the code, resolving dependencies, and running the example command python3 main.py --config config/vbunny.yaml.
It ran fine initially, but after many iterations failed with a tensor shape error:
...
Iter: 736
Iter: 737
Iter: 738
Iter: 739
Iter: 740
Traceback (most recent call last):
File "nie/main.py", line 168, in <module>
main(config)
File "nie/main.py", line 96, in main
faces.data[:f] = torch.from_numpy(np.ascontiguousarray(faces_np))
RuntimeError: The expanded size of the tensor (950000) must match the existing size (1094172) at non-singleton dimension 0. Target sizes: [950000, 3]. Tensor sizes: [1094172, 3]
I haven't tried debugging in to the issue at all, but wanted to let you know.
Hi! Thanks for sharing this code, and great paper :)
Just an FYI, I tried grabbing the code, resolving dependencies, and running the example command
python3 main.py --config config/vbunny.yaml
.It ran fine initially, but after many iterations failed with a tensor shape error:
I haven't tried debugging in to the issue at all, but wanted to let you know.