Document-level Information Extraction on BETTER
36.74F1 ScoreTHINKTWICE Qwen 3 (oracle)
Evaluation Results
| Method | Links | |
|---|---|---|
| THINKTWICE Qwen 3 (oracle)Selector=oracle, Backbone=Qwen 3, Shots=Zero-shot2026.01 | 36.74 | |
| THINKTWICE Llama R1 (oracle)Selector=oracle, Backbone=Llama R1, Shots=Zero-shot2026.01 | 34.08 | |
| THINKTWICE Qwen 3Selector=F1 Voting, Backbone=Qwen 3, Shots=Zero-shot2026.01 | 20.02 | |
| THINKTWICE Qwen 3Selector=Majority, Backbone=Qwen 3, Shots=Zero-shot2026.01 | 17.38 | |
| THINKTWICE Llama R1Selector=F1 Voting, Backbone=Llama R1, Shots=Zero-shot2026.01 | 17.1 | |
| Greedy Qwen 3Selector=X, Backbone=Qwen 3, Shots=Zero-shot2026.01 | 16.12 | |
| Greedy Llama R1Selector=X, Backbone=Llama R1, Shots=Zero-shot2026.01 | 14.78 | |
| THINKTWICE Llama R1Selector=Majority, Backbone=Llama R1, Shots=Zero-shot2026.01 | 3.12 |