Question Answering on SQuAD 2.0 (AUROC)
58.8AUROCRALC
Evaluation Results
| Method | Links | |
|---|---|---|
| RALCSignal=Sem. Uncertainty2026.05 | 58.8 | |
| RALCSignal=Token Probability2026.05 | 55.7 | |
| Hedged QA2026.05 | 52.1 | |
| Original Response2026.05 | 49.8 | |
| RALCSignal=Ling. Confidence2026.05 | 48.8 |