PyTorch Extension Library of Optimized Autograd Sparse Matrix Operations
SparseTensor
via sparse_mat1 * sparse_mat2
(https://github.com/rusty1s/pytorch_sparse/pull/323)to_symmetric
(https://github.com/rusty1s/pytorch_sparse/pull/327)mps
Apple silicon GPU Acceleration support (https://github.com/rusty1s/pytorch_sparse/pull/335)node_weight
in metis
computation (https://github.com/rusty1s/pytorch_sparse/pull/342)Full Changelog: https://github.com/rusty1s/pytorch_sparse/compare/0.6.17...0.6.18
index_sort
in case pyg-lib
is installed as well (#306)balance_edge
option to the METIS graph partitioning algorithm (#309)SparseTensor.to_torch_sparse_csc_tensor
functionality (#319)spspmm
on newer CUDA versions/GPUstorch.bfloat16
support in spmm
std::unordered_map
with phmap::flat_hash_map
for faster sampling (#266)torch.manual_seed
(#217)SparseTensor
: __eq__
functionalitySparseTensor
: add
functionality of two sparse matrices (#177)SparseTensor
: to_torch_csr_tensor
and from_torch_csr_tensor
functionalitySparseTensor
: Allow indexing via np.array
(#194)SparseTensor
: Skip unnecessary assertions and enable non-blocking data transfers (#195)PyTorch 1.10 is now required.
torch.ops.torch_sparse.hetero_neighbor_sample
torch.ops.torch_sparse.hgt_sample
(thanks to @chantat)set_diag
in case SparseTensor
does not hold any non-zero elementstorch.half
) for all operators in torch-sparse
This release brings PyTorch 1.9.0 and Python 3.9 support to torch-sparse
.
row.max() < sparse_sizes[0]
and col.max() < sparse_sizes[1]
when creating a SparseTensor
in order to avoid unexpected behavior (thanks to @Adam1679)partition
now supports the additional optional argument node_weight
(thanks to @Spazierganger)spmm
now supports torch.half
sample_adj
did not return a sparse matrix with sorted indicesspmm
in case num_edges < num_nodes