HaochengWan / PVT

PVT: Point-Voxel Transformer for 3D Deep Learning
https://arxiv.org/abs/2108.06076
MIT License
87 stars 8 forks source link

forward() missing 1 required positional argument: 'rel_pos' #4

Open tomanick opened 2 years ago

tomanick commented 2 years ago

When I wanted to train dataset, I met following problems. image

What is "rel_pos"?

Perhaps shared_transformer.py file should add "rel_pos" parameter in Class SharedTransformer ?

image

GunnerStone commented 2 years ago

Getting the same error. My dataloader is extracting a single data input as size [batch_size, num_channels, num_points] num_channels = 3 # for xyz channels Not sure what rel_pos is or what its supposed to be. It isn't documented well in the example training scripts. The trainloaders in main_part_seg.py also seem to extract their batches similarly [batch_size, xyz_channel, num_points] and pass that directly into the model without any extra argument.

Did you have any luck fixing this? @tomanick