yunfei
yunfei
In 2023, looks like that this bug is still existing..., seeming that it is time to find another project aforementioned by EliotChenKJ.
Thanks. I solved it by using another function to replace distCUDA2. def distCUDA2(points): points_np = points.detach().cpu().float().numpy() dists, inds = KDTree(points_np).query(points_np, k=4) meanDists = (dists[:, 1:] ** 2).mean(1) return torch.tensor(meanDists, dtype=points.dtype,...
Maybe ml-hugs is not suitable for a custom dataset, due to the relatively complex implementation of dataset preprocessing.