Question Answering on HEAD-QA (eval)
0Pathwise Violations (PathV)CSA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CSAN (EVAL size)=1,100, Err (Base verifier-fail rate)=26.0%, alpha* (Target risk level)=0.202026.05 | 0 | 62.2 | 16.2 | |
| NEX-ConfN (EVAL size)=1,100, Err (Base verifier-fail rate)=26.0%, alpha* (Target risk level)=0.202026.05 | 0 | — | — | |
| 5 online baselinesN (EVAL size)=1,100, Err (Base verifier-fail rate)=26.0%, alpha* (Target risk level)=0.202026.05 | 100 | — | — |