Knowledge Boundary Expression on CFLUE (test)
68.7F1 ScoreLSCL
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| LSCLBase Model=Qwen2-7B2026.02 | 68.7 | 86.2 | 67.8 | |
| LSCLBase Model=Qwen-flash2026.02 | 67.8 | 76.9 | 70.6 | |
| LSCLBase Model=Doubao-flash2026.02 | 64.4 | 81.9 | 66.8 | |
| Token probsBase Model=Doubao-flash2026.02 | 57.2 | 84.3 | 65 | |
| Token probsBase Model=Qwen2-7B2026.02 | 52.5 | 76.8 | 53.9 | |
| Prior promptBase Model=Doubao-flash2026.02 | 50.6 | 78.5 | 51.5 | |
| Posterior promptBase Model=Doubao-flash2026.02 | 46.6 | 75.6 | 51.5 | |
| Prior promptBase Model=Qwen-flash2026.02 | 46.2 | 74.4 | 48.7 | |
| Posterior promptBase Model=Qwen-flash2026.02 | 44.9 | 80.1 | 50.2 | |
| Posterior promptBase Model=Qwen2-7B2026.02 | 44.4 | 44.7 | 48.4 | |
| Prior promptBase Model=Qwen2-7B2026.02 | 40 | 65.1 | 50 | |
| Guiding by promptBase Model=Qwen2-7B2026.02 | 36.7 | 50.4 | 38.9 | |
| Token probsBase Model=Qwen-flash2026.02 | 31.4 | 46.7 | 50.3 | |
| Guiding by promptBase Model=Doubao-flash2026.02 | 27.9 | 43.7 | 36.4 | |
| Guiding by promptBase Model=Qwen-flash2026.02 | 6.1 | 9.9 | 33.4 |