Probabilistic Model Calibration on TriviaQA, CoQA, and NQ
0.03ECEACSE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ACSEalpha (α)=0.20, Tolerance level=Permissive2026.05 | 0.03 | 0.08 | |
| ACSEalpha (α)=0.10, Tolerance level=Standard2026.05 | 0.04 | 0.09 | |
| CAPalpha (α)=0.20, Tolerance level=Permissive2026.05 | 0.05 | 0.1 | |
| CAPalpha (α)=0.10, Tolerance level=Standard2026.05 | 0.06 | 0.11 | |
| ACSEalpha (α)=0.05, Tolerance level=Strict2026.05 | 0.06 | 0.1 | |
| DDCRP-CPalpha (α)=0.20, Tolerance level=Permissive2026.05 | 0.07 | 0.12 | |
| CAPalpha (α)=0.05, Tolerance level=Strict2026.05 | 0.07 | 0.12 | |
| DDCRP-CPalpha (α)=0.10, Tolerance level=Standard2026.05 | 0.09 | 0.14 | |
| DDCRP-CPalpha (α)=0.05, Tolerance level=Strict2026.05 | 0.1 | 0.15 | |
| SUalpha (α)=0.20, Tolerance level=Permissive2026.05 | 0.14 | 0.22 | |
| SUalpha (α)=0.10, Tolerance level=Standard2026.05 | 0.16 | 0.24 | |
| SUalpha (α)=0.05, Tolerance level=Strict2026.05 | 0.17 | 0.25 |