Question Answering on TrivialQA Extreme
40.41F1 ScoreOpenDecoder
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| OpenDecoderBackbone=Qwen-2.5-3B-Instruct2026.01 | 40.41 | 38.27 | |
| RbFTBackbone=Qwen-2.5-3B-Instruct2026.01 | 38.18 | 33.5 | |
| AstuteRAGBackbone=Qwen-2.5-3B-Instruct2026.01 | 35.03 | 27.1 | |
| Vanilla SFTBackbone=Qwen-2.5-3B-Instruct2026.01 | 34.76 | 33.4 | |
| InstructRAGBackbone=Qwen-2.5-3B-Instruct2026.01 | 21.51 | 24.8 | |
| Vanilla RAGBackbone=Qwen-2.5-3B-Instruct2026.01 | 11.96 | 18 | |
| RobustRAGBackbone=Qwen-2.5-3B-Instruct2026.01 | 7.39 | 7.13 |