Error Detection on MAQA* High-Ambiguity Subset H[p*] >= 1.5 (test)
65AUROCVar. Credal
Evaluation Results
| Method | Links | |
|---|---|---|
| Var. Credal2026.02 | 65 | |
| Deep Ensembles2026.02 | 53 | |
| Semantic Entropy2026.02 | 52 | |
| MC Dropout2026.02 | 52 | |
| P(True)2026.02 | 51 |
Loading the SOTA2 catalog…