Graph Classification on CIFAR10
77.784AccuracyTAU-GRIT
Evaluation Results
| Method | Links | |
|---|---|---|
| TAU-GRITTokenizer=TAU, Backbone=GRIT2025.10 | 77.784 | |
| GRIT2024.02 | 76.468 | |
| GRIT2025.10 | 76.468 | |
| GNN-AK+backbone=PNA*, Sampling=SubgraphDrop2021.10 | 74.79 | |
| EXPHORMER2023.03 | 74.69 | |
| TIGT2024.02 | 73.955 | |
| LRGA + GatedGCNParameter Budget=487K2020.06 | 73.48 | |
| PNA*2021.10 | 73.11 | |
| QUESTAttention=QUEST2026.03 | 72.843 | |
| DGN2021.10 | 72.84 | |
| DGN2023.03 | 72.84 | |
| DGN2024.02 | 72.838 | |
| GNN-AK+backbone=GCN2021.10 | 72.7 | |
| DGN#Param=≈100K2021.08 | 72.7 | |
| PNAedge features=none2022.05 | 72.7 | |
| GPS + GFSAGFSA=true2023.12 | 72.44 | |
| GNN-AK+backbone=GIN, Sampling=SubgraphDrop2021.10 | 72.39 | |
| GraphGPS2023.03 | 72.3 | |
| GPSGFSA=false2023.12 | 72.3 | |
| GPS2024.02 | 72.298 | |
| StandardAttention=Standard2026.03 | 72.298 | |
| GraphGPS2025.10 | 72.298 | |
| GNN-AK+backbone=GIN2021.10 | 72.19 | |
| GIN-AK+2023.03 | 72.19 | |
| GIN-AK+2024.02 | 72.19 | |
| GNN-AK+backbone=GCN, Sampling=SubgraphDrop2021.10 | 71.93 | |
| LRGA + GATParameter Budget=476K2020.06 | 71.57 | |
| GatedGCNParameter Budget=500K2020.06 | 71.33 | |
| LRGA + GatedGCNParameter Budget=93K2020.06 | 70.65 | |
| PNA2021.10 | 70.47 | |
| PNA#Param=≈100K2021.08 | 70.35 | |
| PNA2023.03 | 70.35 | |
| PNA2024.02 | 70.35 | |
| Performer + GRF++Graph Features=GRF++2025.10 | 70.21 | |
| PHM-GNNedge features=none2022.05 | 70.2 | |
| Performer + GRFGraph Features=GRF2025.10 | 70.1 | |
| Performer2025.10 | 69.96 | |
| Graph-ViT + GFSAGFSA=true2023.12 | 69.87 | |
| Graph-ViTGFSA=false2023.12 | 69.67 | |
| GatedGCN2021.10 | 69.37 | |
| FGN2026.06 | 69.3 | |
| GatedGCNedge features=none2022.05 | 69.2 | |
| CRaW12024.02 | 69.013 | |
| CRaWI2023.03 | 69.01 | |
| EGT#Param=≈100K2021.08 | 68.702 | |
| EGT2024.02 | 68.702 | |
| EGT2025.10 | 68.702 | |
| EGT2023.03 | 68.7 | |
| tGNNedge features=none2022.05 | 68.4 | |
| LRGA + GCNParameter Budget=463K2020.06 | 68.27 | |
| LRGA + GATParameter Budget=90K2020.06 | 68 | |
| GNN-AKbackbone=GIN2021.10 | 67.51 | |
| GatedGCN#Param=≈100K2021.08 | 67.312 | |
| GatedGCN2024.02 | 67.312 | |
| GatedGCNParameter Budget=104K2020.06 | 67.31 | |
| GatedGCN2023.03 | 67.31 | |
| GatedGCN2026.06 | 67.3 | |
| LoGoGNN2026.06 | 67.3 | |
| GATParameter Budget=442K2020.06 | 66.11 | |
| Graphormer#Param=≈100K2021.08 | 65.978 | |
| LRGA + GCNParameter Budget=91K2020.06 | 65.8 | |
| SAGEReference=[36]2024.07 | 65.77 | |
| GraphSage#Param=≈100K2021.08 | 65.767 | |
| GraphSAGE2026.06 | 65.7 | |
| GATedge features=none2022.05 | 65.5 | |
| No PromptingQuantization Framework=FP32, Model Architecture=GAT2026.01 | 65.4 | |
| Cy2C-GNNs2024.02 | 64.285 | |
| GAT#Param=≈100K2021.08 | 64.223 | |
| GAT2024.02 | 64.223 | |
| GATParameter Budget=110K2020.06 | 64.22 | |
| GAT2023.03 | 64.22 | |
| GATReference=[36]2024.07 | 64.22 | |
| GAT2026.06 | 64.2 | |
| GPF-LoRAPQuantization Framework=A2Q, Model Architecture=GAT2026.01 | 62.9 | |
| GPF-plusQuantization Framework=A2Q, Model Architecture=GAT2026.01 | 62.3 | |
| No PromptingQuantization Framework=A2Q, Model Architecture=GAT2026.01 | 61.9 | |
| GHC2024.07 | 59.83 | |
| GIN2021.10 | 59.82 | |
| GCN2021.10 | 58.39 | |
| DiffPooledge features=none2022.05 | 57.9 | |
| GPF-LoRAPQuantization Framework=DQ, Model Architecture=GAT2026.01 | 57.5 | |
| GPF-LoRAPQuantization Framework=A2Q, Model Architecture=GIN2026.01 | 57.4 | |
| GPF-plusQuantization Framework=A2Q, Model Architecture=GIN2026.01 | 56.8 | |
| MLP2026.06 | 56.3 | |
| MLPedge features=none2022.05 | 56 | |
| GPF-LoRAPQuantization Framework=QAT, Model Architecture=GIN2026.01 | 55.8 | |
| GCNParameter Budget=101K2020.06 | 55.71 | |
| GCN#Param=≈100K2021.08 | 55.71 | |
| GCN2023.03 | 55.71 | |
| GCN2024.02 | 55.71 | |
| GPF-plusQuantization Framework=QAT, Model Architecture=GIN2026.01 | 55.3 | |
| GIN2023.03 | 55.26 | |
| GIN#Param=≈100K2021.08 | 55.255 | |
| GIN2024.02 | 55.255 | |
| No PromptingQuantization Framework=A2Q, Model Architecture=GIN2026.01 | 54.9 | |
| GCNParameter Budget=504K2020.06 | 54.84 | |
| GCNedge features=none2022.05 | 54.5 | |
| No PromptingQuantization Framework=FP32, Model Architecture=GCN2026.01 | 54.5 | |
| No PromptingQuantization Framework=DQ, Model Architecture=GAT2026.01 | 54.2 | |
| GCNReference=[36]2024.07 | 54.14 |