Mathematical Reasoning on GSM8K (eval)
0.3RiskLTT
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| LTTN (EVAL size)=1,055, Err (Base verifier-fail rate)=5.0%, alpha* (Target risk level)=0.052026.05 | 0.3 | — | — | |
| ConfFactN (EVAL size)=1,055, Err (Base verifier-fail rate)=5.0%, alpha* (Target risk level)=0.052026.05 | 0.3 | — | — | |
| CSAN (EVAL size)=1,055, Err (Base verifier-fail rate)=5.0%, alpha* (Target risk level)=0.052026.05 | 2.6 | 71.7 | 0 | |
| 5 online baselinesN (EVAL size)=1,055, Err (Base verifier-fail rate)=5.0%, alpha* (Target risk level)=0.052026.05 | — | — | 100 | |
| NEX-ConfN (EVAL size)=1,055, Err (Base verifier-fail rate)=5.0%, alpha* (Target risk level)=0.052026.05 | — | — | 10 |