1.5.0
This release is a big one thanks to many wonderful contributors. You guys are awesome!
Breaking Changes and Highlights
NeighborSamplergot completely revamped: it's now much faster, allows for parallel sampling, and allows to easily apply skip-connections or self-loops. Seeexamples/reddit.pyor the newly introduced OGB examples (examples/ogbn_products_sage.pyandexamples/ogbn_products_gat.py). The latter also sets a new SOTA on the OGB leaderboards (reaching 0.7945 ± 0.0059 test accuracy)SAGEConvnow usesconcat=Trueby default, and there is no option to disable it anymoreNode2Vecgot enhanced by a parallel sampling mechanism, and as a result, its API slightly changedMetaPath2Vec: The first model in PyG that is able to operate on heteregenous graphsGNNExplainer: Generating explanations for graph neural networksGraphSAINT: A graph sampling based inductive learning methodSchNetmodel for learning on molecular graphs, comes with pre-trained weights for each target of theQM9dataset (thanks to @Nyuten)
Additional Features
ASAPooling: Adaptive structure aware pooling for learning hierarchical graph representations (thanks to @ekagra-ranjan)ARGVAnode clustering example, seeexamples/argva_node_clustering.py(thanks to @gsoosk)MFConv: Molecular fingerprint graph convolution operator (thanks to @rhsimplex)GIN-E-Convthat extends theGINConvto also account for edge featuresDimeNet: Directional message passing for molecular graphsSIGN: Scalable inception graph neural networksGravNetConv(thanks to @jkiesele)
Datasets
Minor changes
GATConvcan now return attention weights via thereturn_attention_weightsargument (thanks to @douglasrizzo)InMemoryDatasetnow has acopymethod that converts sliced datasets back into a contiguous memory layoutPlanetoidgot enhanced by the ability to let users choose between different splitting methods (thanks to @dongkwan-kim)k_hop_subgraph: Computes thek-hop subgraph around a subset of nodesgeodesic_distance: Geodesic distances can now be computed in parallel (thanks to @jannessm)tree_decomposition: The tree decompostion algorithm for generating junction trees from moleculesSortPoolbenchmark script now uses 1-D convolutions after pooling, leading to better performance (thanks to @muhanzhang)
Bugfixes
- Fixed a bug in
write_off - Fixed a bug in the processing of the
GEDDatasetdataset to_networkxconversion can now also properly handle non-tensor attributes- Fixed a bug in
read_obj(thanks to @mwussow)