pytorch_cluster
pytorch_cluster copied to clipboard
rusty1s
PyTorch Extension Library of Optimized Graph Cluster Algorithms
Let's create 3 points in 1D space: ```python import torch as pt from torch_cluster import radius_graph x = pt.tensor([[0.], [1.], [2.]], device='cuda') ``` If `max_num_neighbors` is 2 or more, it...
Radius CPU computation with KDTree.query() when number of neighbors is higher than max_num_neighbors
Hello, In the radius search CPU, you are using: ``` tree = scipy.spatial.cKDTree(x.detach().numpy()) _, col = tree.query( y.detach().numpy(), k=max_num_neighbors, distance_upper_bound=r + 1e-8) ``` It seems that there is no guaranty...
I'm training a PointNet++, based on `https://github.com/pyg-team/pytorch_geometric/blob/master/examples/pointnet2_classification.py`. I'm getting a similar error as [this issue](https://github.com/pyg-team/pytorch_geometric/issues/850): ```python File "(..)/train_pointnet2.py", line 33, in forward idx = fps(pos, batch, ratio=self.ratio) File "(..)/python3.7/site-packages/torch_cluster/fps.py", line...
Hi , thank you for sharing the library. I have used it and it's really good and very fast. could you please share how we can use the fps and...
`torch_cluster.radius` is not compatible with the CUDA graph (https://pytorch.org/docs/stable/notes/cuda.html#cuda-graphs) ```python from torch_cluster import radius device = pt.device('cuda') x = pt.tensor([0.0], device=device) y = pt.tensor([1.0], device=device) graph = pt.cuda.CUDAGraph() with pt.cuda.graph(graph):...
When testing with the following codes ``` torch.manual_seed(1234) start = torch.tensor([0, 1]) cumdeg = torch.tensor([0, 3, 7]) neighbor_sampler(start, cumdeg, size=2) ``` we cannot get consistent bebaviors with multiple runs as...
Hey @rusty1s, I am using voxel_grid, and on my 8000 point cloud, it creates a 5001 indexes with the highest being around 400k. As a result, the pool_pos used to...
Is there any way to call random_walk function with custom transition probabilities?
Hi, Thank you for the great library. Is there a plan to return the radius of the sampled points as well as the points in the function `torch_cluster.fps`? I mean,...
Hey @rusty1s , I would be great to be able to specify a min_num_neighnors. If the sample < min_num_neighnors, points will be duplicated to have the correct number of points....