Skip to content

pyg-lib 0.4.0: PyTorch 2.2 support, distributed sampling, sparse softmax, edge-level temporal sampling

Latest
Compare
Choose a tag to compare
@rusty1s rusty1s released this 07 Feb 13:09
· 43 commits to master since this release

pyg-lib==0.4.0 brings PyTorch 2.2 support, distributed neighbor sampling, accelerated softmax operations, and edge-level temporal sampling support to PyG πŸŽ‰πŸŽ‰πŸŽ‰

Highlights

PyTorch 2.2 Support

pyg-lib==0.4.0 is fully compatible with PyTorch 2.2 (#294). To install for PyTorch 2.2, simply run

pip install pyg-lib -f https://data.pyg.org/whl/torch-2.2.0+${CUDA}.html

where ${CUDA} should be replaced by either cpu, cu118 or cu121

The following combinations are supported:

PyTorch 2.2 cpu cu118 cu121
Linux βœ… βœ… βœ…
macOS βœ…

Older PyTorch versions like PyTorch 1.12, 1.13, 2.0.0 and 2.1.0 are still supported, and can be installed as described in our README.md.

Distributed Sampling

pyg-lib==0.4.0 integrates all the low-level code for performing distributed neighbor sampling as part of torch_geometric.distributed in PyG 2.5 (#246, #252, #253, #254).

Sparse Softmax Implementation

pyg-lib==0.4.0 supports a fast sparse softmax_csr implementation based on CSR input representation (#264, #282):

from pyg_lib.ops import softmax_csr

src = torch.randn(4, 4)
ptr = torch.tensor([0, 4])
out = softmax_csr(src, ptr)

Edge-level Temporal Sampling

pyg-lib==0.4.0 brings edge-level temporal sampling support to PyG (#280). In particular, neighbor_sample and hetero_neighbor_sample now support the edge_time attribute, which will only samples edges in case they have a lower or equal timestamp than their corresponding seed_time.

Additional Features

Bugfixes

New Contributors

Full Changelog: 0.3.0...0.4.0