Context-faithful Question Answering on NQ
90.39ConRPromptCD
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PromptCDModel=QWEN2.5-7B-INSTRUCT2026.02 | 90.39 | 14.39 | 13.74 | |
| PromptCDModel=MISTRALV0.3-7B-INSTRUCT2026.02 | 86.19 | 3.99 | 4.43 | |
| OpinModel=QWEN2.5-7B-INSTRUCT2026.02 | 82.59 | 21.79 | 20.88 | |
| AttrModel=QWEN2.5-7B-INSTRUCT2026.02 | 82.39 | 17.59 | 17.59 | |
| PromptCDModel=LLAMA2-7B-CHAT2026.02 | 75.99 | 6.39 | 7.76 | |
| OpinModel=LLAMA2-7B-CHAT2026.02 | 74.59 | 15.39 | 17.11 | |
| AttrModel=LLAMA3-8B-INSTRUCT2026.02 | 74.39 | 41.39 | 35.75 | |
| OpinModel=MISTRALV0.3-7B-INSTRUCT2026.02 | 73.39 | 16.99 | 18.81 | |
| VanillaModel=QWEN2.5-7B-INSTRUCT2026.02 | 73.39 | 32.39 | 31.27 | |
| PromptCDModel=LLAMA3-8B-INSTRUCT2026.02 | 72.69 | 6.59 | 8.33 | |
| AttrModel=MISTRALV0.3-7B-INSTRUCT2026.02 | 71.59 | 25.59 | 26.34 | |
| OpinModel=LLAMA3-8B-INSTRUCT2026.02 | 69.99 | 20.99 | 23.07 | |
| AttrModel=LLAMA2-7B-CHAT2026.02 | 64.39 | 21.99 | 25.46 | |
| VanillaModel=MISTRALV0.3-7B-INSTRUCT2026.02 | 46.19 | 58.59 | 55.92 | |
| VanillaModel=LLAMA3-8B-INSTRUCT2026.02 | 43.92 | 34.05 | 43.47 | |
| VanillaModel=LLAMA2-7B-CHAT2026.02 | 43.35 | 43.75 | 50.24 |