Question Answering on PaQa
46.83ROUGE-LRAC_SFT
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| RAC_SFTTraining=Supervised Fine-Tuning2026.01 | 46.83 | 20.17 | 47.97 | 90.85 | 43.36 | 27.62 | |
| RAC_DPOTraining=Direct Preference Optimization2026.01 | 45.26 | 18.32 | 46.4 | 90.41 | 45.75 | 28.54 | |
| Q-Cond2026.01 | 42.46 | 16.62 | 41.58 | 90.12 | — | — | |
| QP-ZeroshotMode=Zero-shot2026.01 | 33.79 | 10.42 | 35.84 | 88.66 | — | — | |
| AT-CoTReasoning=Chain-of-Thought2026.01 | 23.59 | 7.07 | 22.93 | 85.97 | — | — |