PyTorch Extension Library of Optimized Graph Cluster Algorithms
torch.jit.script
conversion of functions on module importbatch_size
argument for fps
, knn
, radius
functions (https://github.com/rusty1s/pytorch_cluster/pull/175)ptr
argument https://github.com/rusty1s/pytorch_cluster/pull/180)Full Changelog: https://github.com/rusty1s/pytorch_cluster/compare/1.6.1...1.6.2
random_walk
(#139)bf16
support for knn
, radius
and graclus
(#144)batch
layout in nearest
(#168)num_nodes
in random_walk
(in case num_nodes=None
) (#112)knn
/radius
calculation for batches with zero-point examplesknn
/radius
calculationfps
can now take in different ratios across different batched point sets, i.e. ratio
can now be a torch.Tensor
radius
computed slightly different results across CPU and CUDA versionsrandom_walk
now supports q != 1
and p != 1
via rejection samplingradius
where the max_num_neighbors
argument has been ignoredThis release fixes some issues in the new knn
and radius
CPU implementations that have led to memory leaks. It is strongly recommended to update the package in case you are currently using torch-cluster==1.5.5
.
torch-cluster
is now fully-jittable thanks to new implementations for knn
and radius
based on nanoflann
rather than scipy
.Fixed a bug in the CUDA version of fps
.