Node Classification on PATTERN (test)
87.281Test AccuracyGRIT
Evaluation Results
| Method | Links | |
|---|---|---|
| GRITencoding=HDSE2023.08 | 87.281 | |
| GRIT# Parameters=~ 500K2023.05 | 87.196 | |
| GRIT2023.08 | 87.196 | |
| GRITparameter budget=≈ 500K2023.12 | 87.196 | |
| GRIT2024.06 | 87.196 | |
| GRIT2026.05 | 87.196 | |
| CSA2023.04 | 87.011 | |
| NeuralWalker2024.06 | 86.977 | |
| SATencoding=HDSE2023.08 | 86.933 | |
| GIN-AK+2022.05 | 86.85 | |
| GIN-AK+# Parameters=~ 500K2023.05 | 86.85 | |
| GIN-AK+2023.08 | 86.85 | |
| GIN-AK+2026.05 | 86.85 | |
| K-Subgraph SAT2022.05 | 86.848 | |
| SAT2023.04 | 86.848 | |
| K-Subgraph SAT# Parameters=~ 500K2023.05 | 86.848 | |
| SAT2023.08 | 86.848 | |
| SATparameter budget=≈ 500K2023.12 | 86.848 | |
| Graphormer2024.06 | 86.848 | |
| EGT2022.05 | 86.821 | |
| EGT2023.04 | 86.821 | |
| EGT# Parameters=~ 500K2023.05 | 86.821 | |
| Graphormer-GD2023.08 | 86.821 | |
| EGTparameter budget=≈ 500K2023.12 | 86.821 | |
| EGT2026.05 | 86.821 | |
| GNASDepth=4, Parameters=0.35M, Search Cost=2.45 hr, Training Cost=2.15 hr2021.03 | 86.8 | |
| GPS-GCCMBackbone=GPS2026.05 | 86.772 | |
| GREDparameter budget=≈ 500K2023.12 | 86.759 | |
| Graph ViT/MLP-Mixer2023.08 | 86.742 | |
| GraphGPSencoding=HDSE2023.08 | 86.737 | |
| GTencoding=HDSE2023.08 | 86.713 | |
| Exphormer2024.06 | 86.7 | |
| GPS-PCLBackbone=GPS2026.05 | 86.695 | |
| TransformerComplexity=O(N^2), Positional Encoding=Baseline2025.09 | 86.69 | |
| GPS2022.05 | 86.685 | |
| GPS2023.04 | 86.685 | |
| GPS# Parameters=~ 500K2023.05 | 86.685 | |
| GraphGPS2023.08 | 86.685 | |
| GPSparameter budget=≈ 500K2023.12 | 86.685 | |
| GPS2024.06 | 86.685 | |
| DGN2022.05 | 86.68 | |
| DGN# Parameters=~ 500K2023.05 | 86.68 | |
| DGN2026.05 | 86.68 | |
| ViT-PSsamples=10, pretrain=ImageNet-21k2023.06 | 86.65 | |
| GPS2026.05 | 86.648 | |
| WIREModel Architecture=Performer, Complexity=O(N)2025.09 | 86.63 | |
| SAN2022.05 | 86.581 | |
| SAN2023.04 | 86.581 | |
| SAN# Parameters=~ 500K2023.05 | 86.581 | |
| SAN2023.08 | 86.581 | |
| SANparameter budget=≈ 500K2023.12 | 86.581 | |
| SAN2026.05 | 86.581 | |
| ViT-PSsamples=1, pretrain=ImageNet-21k2023.06 | 86.573 | |
| ViT-Canonicalpretrain=ImageNet-21k2023.06 | 86.534 | |
| RingGNNlayers=22023.06 | 86.245 | |
| ViT-PSsamples=10, pretrain=none2023.06 | 85.989 | |
| ViT-PSsamples=1, pretrain=none2023.06 | 85.868 | |
| ViT-Canonicalpretrain=none2023.06 | 85.825 | |
| SINC-GCN2025.11 | 85.79 | |
| SIR-GCN2025.11 | 85.75 | |
| PerformerComplexity=O(N), Positional Encoding=Baseline2025.09 | 85.71 | |
| GPS-LGDBackbone=GPS2026.05 | 85.662 | |
| PPGNlayers=32023.06 | 85.661 | |
| GCNlayers=162023.06 | 85.614 | |
| GCNparameter budget=≈ 500K2023.12 | 85.614 | |
| GINDepth=4, Parameters=0.10M, Training Cost=0.40 hr2021.03 | 85.59 | |
| GINparameter budget=≈ 500K2023.12 | 85.59 | |
| GIN2025.11 | 85.59 | |
| GatedGCN2022.05 | 85.568 | |
| GatedGCN# Parameters=~ 500K2023.05 | 85.568 | |
| GatedGCNlayers=162023.06 | 85.568 | |
| GatedGCN2023.08 | 85.568 | |
| GatedGCNparameter budget=≈ 500K2023.12 | 85.568 | |
| GatedGCN2024.06 | 85.568 | |
| GatedGCN2026.05 | 85.568 | |
| GCN2025.11 | 85.5 | |
| MoNetDepth=4, Parameters=0.10M, Training Cost=0.90 hr2021.03 | 85.48 | |
| GIN2022.05 | 85.387 | |
| GIN# Parameters=~ 500K2023.05 | 85.387 | |
| GINlayers=162023.06 | 85.387 | |
| GIN2023.08 | 85.387 | |
| GIN2024.06 | 85.387 | |
| GIN2026.05 | 85.387 | |
| SGFormer2023.08 | 85.287 | |
| GraphNASDepth=4, Parameters=0.48M, Search Cost=120 hr, Training Cost=8.25 hr2021.03 | 85.21 | |
| GT2023.08 | 84.808 | |
| ViT-GAsamples=10, pretrain=ImageNet-21k2023.06 | 84.641 | |
| GatedGCNDepth=4, Parameters=0.10M, Training Cost=3.09 hr2021.03 | 84.48 | |
| ViT-GAsamples=10, pretrain=none2023.06 | 83.22 | |
| ViT-GAsamples=1, pretrain=ImageNet-21k2023.06 | 81.933 | |
| ViT-FApretrain=ImageNet-21k2023.06 | 80.015 | |
| GAT2022.05 | 78.271 | |
| GAT# Parameters=~ 500K2023.05 | 78.271 | |
| GATparameter budget=≈ 500K2023.12 | 78.271 | |
| GAT2024.06 | 78.271 | |
| GAT2026.05 | 78.271 | |
| GATlayers=162023.06 | 78.221 | |
| ViT-GAsamples=1, pretrain=none2023.06 | 76.956 | |
| GATDepth=4, Parameters=0.11M, Training Cost=0.57 hr2021.03 | 75.82 | |
| GCN2022.05 | 71.892 |