Document-level Relation Extraction on Re-DocRED Long-Tail Triples 1.0 (test)
75.75PrecisionDREEAM-DOREMI
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DREEAM-DOREMICategory=RoBERTa, Backbone=RoBERTa, Training Strategy=DOREMI iterative training2026.01 | 75.75 | 0.7032 | 44.98 | 56.44 | 0.5486 | |
| DREEAM-DOREMICategory=BERT, Backbone=BERT, Training Strategy=DOREMI iterative training2026.01 | 72.5 | 0.664 | 42.1 | 53.26 | 0.5153 | |
| ATLOP-DOREMICategory=RoBERTa, Backbone=RoBERTa, Training Strategy=DOREMI iterative training2026.01 | 62.32 | 0.5315 | 34.89 | 42.06 | 0.3784 | |
| ATLOP-DOREMICategory=BERT, Backbone=BERT, Training Strategy=DOREMI iterative training2026.01 | 60.23 | 0.5004 | 36.28 | 42.99 | 0.3795 | |
| GPT-4.1Category=LLMs, Zero-shot=true, Temperature=0.752026.01 | 0.93 | — | 1.53 | 1.16 | — | |
| Llama-3.3-70BCategory=LLMs, Zero-shot=true, Temperature=0.72026.01 | 0 | — | 0 | 0 | — |