Edge Classification on WN18RR (test)
93.52AccuracyUniGraphLM
Evaluation results
| Method | Links | |
|---|---|---|
| UniGraphLMLearning Setting=multi-domain multi-task, Fine-tuning=None (joint instruction tuning only)2026.05 | 93.52 | |
| TEA-GLMLearning Setting=multi-domain multi-task, Fine-tuning=None (joint instruction tuning only)2026.05 | 92.85 | |
| Vicuna-7BLearning Setting=multi-domain multi-task, Fine-tuning=None (joint instruction tuning only)2026.05 | 34.84 | |
| LLaMA-2-7BLearning Setting=multi-domain multi-task, Fine-tuning=None (joint instruction tuning only)2026.05 | 29.16 | |
| UniGraphLMSetting=Cross-task, Zero-shot transfer=true, Trained on=Arxiv2026.05 | 24.12 | |
| TEA-GLMSetting=Cross-task, Zero-shot transfer=true, Trained on=Arxiv2026.05 | 22.65 | |
| GOFALearning Setting=multi-domain multi-task, Fine-tuning=None (joint instruction tuning only)2026.05 | 21.95 | |
| GOFASetting=Cross-task, Zero-shot transfer=true, Trained on=Arxiv2026.05 | 18.79 |