Distortion computation issue during training

[voltej]

I got an error in util/util.py:get_local_tris: Exception has occurred: RuntimeError stack expects each tensor to be equal size, but got [3697, 5] at entry 0 and [3697, 4] at entry 1 File "/home/user/projects/meshing/dawand/util/util.py", line 324, in get_local_tris local_tris = torch.stack((x, y), dim=-1).reshape(len(angle), 3, 2) File "/home/user/projects/meshing/dawand/util/losses.py", line 51, in arap local_tris = get_local_tris(vertices, faces, device=device) File "/home/user/projects/meshing/dawand/models/dawand.py", line 759, in optimize_parameters arap_distortion_s2 = self.distortion_fcn(local_tris_s2.float(), faces, uv_res_s2.float(), File "/home/user/projects/meshing/dawand/train.py", line 204, in <module> loss, loss_dict = model.optimize_parameters(epoch) RuntimeError: stack expects each tensor to be equal size, but got [3697, 5] at entry 0 and [3697, 4] at entry 1

the runs smoohtly in dawand.py:759 local_tris_s2 = get_local_tris(vertices, faces, device=self.device)

but the output local_tris_s2 is then through dawand.py:759 and losses:51 again used as input to get_local_tris and then the error occurs.

Should this somewhat work?