Graph Property Prediction On Ogbg Molpcba

评估指标

Ext. data
Number of params
Test AP
Validation AP

评测结果

各个模型在此基准测试上的表现结果

Paper TitleRepository
HyperFusionNo108870850.3204 ± 0.00010.3353 ± 0.0002--
HyperFusinoNo108870850.3204 ± 0.00010.3353 ± 0.0002--
TGT-Ag+TGT-At-DPYes470000000.3167 ± 0.0031-Triplet Interaction Improves Graph Transformers: Accurate Molecular Graph Learning with Triplet Graph Transformers
HIG(pre-trained on PCQM4M)Yes1195296650.3167 ± 0.00340.3252 ± 0.0043--
Graphormer-1195296640.3140 ± 0.00320.3227 ± 0.0024Do Transformers Really Perform Bad for Graph Representation?
Graphormer (pre-trained on PCQM4M)Yes1195296640.3140 ± 0.00320.3227 ± 0.0024Do Transformers Really Perform Bad for Graph Representation?
GatedGCN-HSG--0.3129±0.0020-Next Level Message-Passing with Hierarchical Support Graphs
PDFNo38420480.3031 ± 0.00260.3115 ± 0.0020Towards Better Graph Representation Learning with Parameterized Decomposition & Filtering
PASNo55609600.3012 ± 0.00390.3151 ± 0.0047--
Nested GIN+virtual node (ensemble)No441874800.3007 ± 0.00370.3059 ± 0.0056Nested Graph Neural Networks
Nested GIN+virtual node (ens)--0.3007 ± 0.00370.3059 ± 0.0056Nested Graph Neural Networks
GINE+botNo55116800.2994 ± 0.00190.3094 ± 0.0023--
CRaWlNo61157280.2986 ± 0.00250.3075 ± 0.0020Walking Out of the Weisfeiler Leman Hierarchy: Graph Learning Beyond Message Passing
GatedGCN+No60168600.2981 ± 0.00240.3011 ± 0.0037Can Classic GNNs Be Strong Baselines for Graph-level Tasks? Simple Architectures Meet Excellence
GINE+ w/ APPNPNo61470290.2979 ± 0.00300.3126 ± 0.0023Graph convolutions that can finally model local structure
EGT--0.2961 ± 0.0024-Global Self-Attention as a Replacement for Graph Convolution
PHC-GNNNo16903280.2947 ± 0.00260.3068 ± 0.0025Parameterized Hypercomplex Graph Neural Networks for Graph Classification
GIN-AKNo30810290.2930 ± 0.00440.3047 ± 0.0007From Stars to Subgraphs: Uplifting Any GNN with Local Structure Awareness
GINE+ w/ virtual nodesNo61470290.2917 ± 0.00150.3065 ± 0.0030Graph convolutions that can finally model local structure
GPSNo97444960.29070.3015 ± 0.0038Recipe for a General, Powerful, Scalable Graph Transformer
0 of 36 row(s) selected.
Graph Property Prediction On Ogbg Molpcba | SOTA | HyperAI超神经