AUROC on MMLU Question Answering
0.636AUROC (MMLU QA)RALC
Evaluation results
| Method | Links | |
|---|---|---|
| RALCSignal=Sem. Uncertainty2026.05 | 0.636 | |
| Hedged QA2026.05 | 0.565 | |
| RALCSignal=Token Probability2026.05 | 0.56 | |
| RALCSignal=Ling. Confidence2026.05 | 0.533 | |
| Original Response2026.05 | 0.513 |