Graph Regression On Zinc Full

Metrics

Test MAE

Results

Performance results of various models on this benchmark

		Paper Title
GCN	0.152±0.023	Semi-Supervised Classification with Graph Convolutional Networks
GraphSAGE	0.126±0.003	Inductive Representation Learning on Large Graphs
GATv2	0.079±0.004	How Attentive are Graph Attention Networks?
GAT	0.078±0.006	Graph Attention Networks
GIN	0.068±0.004	How Powerful are Graph Neural Networks?
PNA	0.057±0.007	Principal Neighbourhood Aggregation for Graph Nets
TokenGT	0.047±0.010	Pure Transformers are Powerful Graph Learners
δ-2-LGNN	0.045±0.006	Weisfeiler and Leman go sparse: Towards scalable higher-order graph embeddings
δ-2-GNN	0.042±0.003	Weisfeiler and Leman go sparse: Towards scalable higher-order graph embeddings
Graphormer	0.036±0.002	Do Transformers Really Perform Bad for Graph Representation?
ESA (Edge set attention, no positional encodings)	0.027±0.001	An end-to-end attention-based approach for learning on graphs
GraphGPS	0.024±0.007	Recipe for a General, Powerful, Scalable Graph Transformer
SignNet	0.024±0.003	Sign and Basis Invariant Networks for Spectral Graph Representation Learning
GRIT	0.023	Graph Inductive Biases in Transformers without Message Passing
ESA + RWSE (Edge set attention, Random Walk Structural Encoding)	0.017±0.001	An end-to-end attention-based approach for learning on graphs
ESA + RWSE (Edge set attention, Random Walk Structural Encoding, tuned)	0.0154±0.0001	An end-to-end attention-based approach for learning on graphs
TIGT	0.014	Topology-Informed Graph Transformer
ESA + RWSE + CY2C (Edge set attention, Random Walk Structural Encoding, clique adjacency, tuned)	0.0122±0.0004	An end-to-end attention-based approach for learning on graphs
ESA + rings + NodeRWSE + EdgeRWSE	0.0109±0.0002	An end-to-end attention-based approach for learning on graphs

0 of 19 row(s) selected.