User-Engage Binary Classification on rel-stack (test)
90.47AUROCLinear SAGE
Evaluation results
| Method | Links | |
|---|---|---|
| Linear SAGETraining Protocol=Pretrained and Fine-tuned (P&F)2025.06 | 90.47 | |
| Linear SAGETraining Protocol=Baseline (B)2025.06 | 90.34 | |
| ResNet SAGETraining Protocol=Pretrained and Fine-tuned (P&F)2025.06 | 90.32 | |
| ResNet SAGETraining Protocol=Baseline (B)2025.06 | 90.28 | |
| GriffinTarget DB in pretraining=Yes, Parameter count=22M2025.10 | 89.4 | |
| Linear SAGETraining Protocol=Pretrained (P)2025.06 | 86.99 | |
| RTTarget DB in pretraining=Yes, Parameter count=22M2025.10 | 86.9 | |
| ResNet SAGETraining Protocol=Pretrained (P)2025.06 | 86.55 | |
| Entity MeanTarget DB in pretraining=No, Parameter count=02025.10 | 83.5 | |
| GemmaTarget DB in pretraining=Maybe, Parameter count=27B2025.10 | 78 | |
| GriffinTarget DB in pretraining=No, Parameter count=22M2025.10 | 77.5 | |
| RTTarget DB in pretraining=No, Parameter count=22M2025.10 | 75.7 | |
| RelLLMTarget DB in pretraining=Yes, Parameter count=3B2025.10 | 69.5 | |
| GemmaTarget DB in pretraining=Maybe, Parameter count=12B2025.10 | 67.8 | |
| GemmaTarget DB in pretraining=Maybe, Parameter count=4B2025.10 | 65.9 |