Entity Classification on RELBENCH rel-stack user-badge (test)
0.9AUROCGelGT
Evaluation Results
| Method | Links | |
|---|---|---|
| GelGT2026.05 | 0.9 | |
| Rel-LLMEvaluation Protocol=Fine-tuned2025.06 | 0.8964 | |
| Rel-LLM2026.05 | 0.89 | |
| Nodeformer2026.05 | 0.89 | |
| RDLModel Type=GNN + Tabular2025.06 | 0.8886 | |
| ICL + MLPEvaluation Protocol=MLP on LLM representations2025.06 | 0.8819 | |
| GriffinTarget DB in pretraining=Yes, Parameter count=22M2025.10 | 0.823 | |
| RTTarget DB in pretraining=Yes, Parameter count=22M2025.10 | 0.811 | |
| RTTarget DB in pretraining=No, Parameter count=22M2025.10 | 0.801 | |
| GemmaTarget DB in pretraining=Maybe, Parameter count=27B2025.10 | 0.8 | |
| GemmaTarget DB in pretraining=Maybe, Parameter count=12B2025.10 | 0.798 | |
| GemmaTarget DB in pretraining=Maybe, Parameter count=4B2025.10 | 0.791 | |
| GriffinTarget DB in pretraining=No, Parameter count=22M2025.10 | 0.735 | |
| ICLEvaluation Protocol=In-context learning2025.06 | 0.7113 | |
| Entity MeanTarget DB in pretraining=No, Parameter count=02025.10 | 0.662 | |
| LightGBMModel Type=Gradient Boosted Trees2025.06 | 0.6343 | |
| Rel-ZeroEvaluation Protocol=Zero-shot2025.06 | 0.6212 | |
| RelLLMTarget DB in pretraining=Yes, Parameter count=3B2025.10 | 0.621 |