List your company

Question Answering on MedQA 80/20 split (seed 42) (val)

6.9RiskLTT
6.727.9359.1510.365May 18, 2026

Evaluation results

MethodLinks
LTTN (EVAL size)=1,018, Err (Base verifier-fail rate)=31.5%, alpha* (Target risk level)=0.202026.056.9——
ConfFactN (EVAL size)=1,018, Err (Base verifier-fail rate)=31.5%, alpha* (Target risk level)=0.202026.058——
CSAN (EVAL size)=1,018, Err (Base verifier-fail rate)=31.5%, alpha* (Target risk level)=0.202026.0511.439.40
5 online baselinesN (EVAL size)=1,018, Err (Base verifier-fail rate)=31.5%, alpha* (Target risk level)=0.202026.05——100
NEX-ConfN (EVAL size)=1,018, Err (Base verifier-fail rate)=31.5%, alpha* (Target risk level)=0.202026.05——0