Graph Property Prediction On Ogbg Code2
Metriken
Ext. data
Number of params
Test F1 score
Validation F1 score
Ergebnisse
Leistungsergebnisse verschiedener Modelle zu diesem Benchmark
Modellname | Ext. data | Number of params | Test F1 score | Validation F1 score | Paper Title | Repository |
|---|---|---|---|---|---|---|
| GCN | No | 11033210 | 0.1507 ± 0.0018 | 0.1399 ± 0.0017 | Semi-Supervised Classification with Graph Convolutional Networks | |
| GAT | No | 11030210 | 0.1569 ± 0.0010 | 0.1442 ± 0.0017 | Graph Attention Networks | |
| EGC-M (No Edge Features) | No | 10986002 | 0.1595 ± 0.0019 | 0.1464 ± 0.0021 | Do We Need Anisotropic Graph Neural Networks? | |
| SAT | No | 15734000 | 0.1937 ± 0.0028 | 0.1773 ± 0.0023 | Structure-Aware Transformer for Graph Representation Learning | |
| DAGNN | No | 35246814 | 0.1751 ± 0.0049 | 0.1607 ± 0.0040 | - | - |
| GMAN+bag of tricks | No | 63684290 | 0.1770 ± 0.0012 | 0.1631 ± 0.0090 | - | - |
| MPNN-Max (No Edge Features) | No | 10971506 | 0.1552 ± 0.0022 | 0.1441 ± 0.0016 | Do We Need Anisotropic Graph Neural Networks? | |
| GPS | No | 12454066 | 0.1894 | 0.1739 ± 0.001 | Recipe for a General, Powerful, Scalable Graph Transformer | |
| DiffPool w/ graphSAGE | No | 10095826 | 0.1401 ± 0.0012 | 0.1405 ± 0.0012 | Hierarchical Graph Representation Learning with Differentiable Pooling | |
| SAT++ with Magnetic Laplacian | No | 14378069 | 0.2222 ± 0.0010 | 0.2044 ± 0.0020 | Transformers Meet Directed Graphs | |
| GraphTrans (GCN-Virtual) | No | 9053246 | 0.1830 ± 0.0024 | 0.1661 ± 0.0012 | - | - |
| PNA (No Edge Features) | No | 10992050 | 0.1570 ± 0.0032 | 0.1453 ± 0.0025 | Do We Need Anisotropic Graph Neural Networks? | |
| GIN+virtual node | No | 13841815 | 0.1581 ± 0.0026 | 0.1439 ± 0.0020 | How Powerful are Graph Neural Networks? | |
| SAT++ with Magnetic Laplacian | No | 14378069 | 0.2222 ± 0.0032 | 0.2044 ± 0.0020 | - | - |
| GIN | No | 12390715 | 0.1495 ± 0.0023 | 0.1376 ± 0.0016 | How Powerful are Graph Neural Networks? | |
| EGC-S (No Edge Features) | No | 11156530 | 0.1528 ± 0.0025 | 0.1427 ± 0.0020 | Do We Need Anisotropic Graph Neural Networks? | |
| DAGformer | No | 14952882 | 0.2018 ± 0.0021 | 0.1846 ± 0.0010 | - | - |
| GraphTrans (GCN) | No | 7563746 | 0.1751 ± 0.0015 | 0.1599 ± 0.0009 | - | - |
| GCN+virtual node | No | 12484310 | 0.1595 ± 0.0018 | 0.1461 ± 0.0013 | Semi-Supervised Classification with Graph Convolutional Networks | |
| DAGNN | - | - | 0.1751 ± 0.0049 | 0.1607 ± 0.0040 | Directed Acyclic Graph Neural Networks |
0 of 21 row(s) selected.