Selective Prediction on NyayaBench v2
41.1Guaranteed Test Coverage (alpha=0.20)WSR Betting + LTT
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| WSR Betting + LTTn_cal=134, delta=0.102026.03 | 41.1 | — | — | 5.5 | 18.5 | 26 | |
| Transfer Bettingn_cal=134, delta=0.10, Transfer risk profile=MASSIVE2026.03 | 41.1 | — | — | 6.9 | 18.5 | 26 | |
| Clopper-Pearson + LTTn_cal=134, delta=0.102026.03 | 32.9 | — | — | 5.5 | 18.5 | 23.3 | |
| LTT + Hoeffdingn_cal=134, delta=0.102026.03 | 26 | — | — | — | 3.4 | 18.5 | |
| PAC-Bayes-λ (transfer)n_cal=134, delta=0.10, Transfer risk profile=MASSIVE2026.03 | 26 | 3.4 | 3.4 | 5.5 | 14.4 | 23.3 | |
| LTT + Emp. Bernsteinn_cal=134, delta=0.102026.03 | 23.3 | — | — | — | 3.4 | 14.4 | |
| Hoeffding + Unionn_cal=134, delta=0.102026.03 | 14.4 | — | — | — | — | — | |
| Wasserstein (ε=0.01)n_cal=134, delta=0.10, Wasserstein epsilon=0.012026.03 | 12.3 | — | — | — | — | — | |
| Emp. Bernstein + Unionn_cal=134, delta=0.102026.03 | 3.4 | — | — | — | — | — |