Question Answering on 2Wiki Noisy
24.17F1 ScoreOpenDecoder
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| OpenDecoderBackbone=Qwen-2.5-3B-Instruct2026.01 | 24.17 | 22.13 | |
| RbFTBackbone=Qwen-2.5-3B-Instruct2026.01 | 23.6 | 22 | |
| Vanilla SFTBackbone=Qwen-2.5-3B-Instruct2026.01 | 22.07 | 20.4 | |
| AstuteRAGBackbone=Qwen-2.5-3B-Instruct2026.01 | 20.84 | 18.6 | |
| InstructRAGBackbone=Qwen-2.5-3B-Instruct2026.01 | 15.3 | 9 | |
| Vanilla RAGBackbone=Qwen-2.5-3B-Instruct2026.01 | 11.97 | 20.38 | |
| RobustRAGBackbone=Qwen-2.5-3B-Instruct2026.01 | 5.3 | 8.2 |