Attribute Ranking on ARTIFACTBENCH (inductive)
0.469Kendall's TauGPT-5.2
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| GPT-5.2Method Category=LLM-based, Graph Info=Yes2026.05 | 0.469 | 0.527 | 57.3 | 92.8 | |
| NCNCMethod Category=GNN-based, Joint Training=Yes2026.05 | 0.445 | 0.545 | 44.2 | 76 | |
| NCNMethod Category=GNN-based, Joint Training=Yes2026.05 | 0.444 | 0.535 | 44.9 | 75.4 | |
| BUDDYMethod Category=GNN-based, Joint Training=Yes2026.05 | 0.418 | 0.514 | 43 | 76.2 | |
| GATv2ConvMethod Category=GNN-based, Joint Training=Yes2026.05 | 0.418 | 0.5 | 44.2 | 74.3 | |
| NeoGNNMethod Category=GNN-based, Joint Training=Yes2026.05 | 0.405 | 0.5 | 41.8 | 73.1 | |
| GPT-5.2Method Category=LLM-based, Graph Info=No2026.05 | 0.383 | 0.431 | 49.5 | 91.1 | |
| Qwen3-8BMethod Category=LLM-based, Graph Info=Yes2026.05 | 0.205 | 0.232 | 39.5 | 87.6 | |
| Qwen3-8BMethod Category=LLM-based, Graph Info=No2026.05 | 0.186 | 0.209 | 41.3 | 88.3 | |
| Global meanMethod Category=Heuristic2026.05 | 0 | 0 | 4.1 | 76.8 | |
| Model meanMethod Category=Heuristic2026.05 | 0 | 0 | 4.1 | 76.8 | |
| Dataset meanMethod Category=Heuristic2026.05 | 0 | 0 | 4.1 | 76.8 |