Question Answering on Overall (EM, F1, Trigger rate (T))
41.6EMVerbal-Calibrate
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Verbal-CalibrateScope=global2026.04 | 41.6 | 50.5 | 48.1 | |
| Uncertain-CalibrateScope=local2026.04 | 40.9 | 48.1 | 61.4 | |
| DRAGIN2026.04 | 32.7 | 43.2 | 76.8 | |
| Ret-All2026.04 | 28.4 | 36.4 | 100 | |
| ADARAGUE2026.04 | 27.8 | 35.4 | 43.8 | |
| No-Ret2026.04 | 25.9 | 33.1 | — | |
| Base-Verbal2026.04 | 21.8 | 27.9 | 13.1 | |
| FLARE2026.04 | 20.8 | 30.6 | 99 | |
| Base-UncTok2026.04 | 20.5 | 27.2 | 3.5 | |
| SR-13BBackbone=13B2026.04 | 8.8 | 22.4 | 29.5 | |
| SR-7BBackbone=7B2026.04 | 6.5 | 16.7 | 46.2 |