List your company

Confidence Estimation on 20Q

33.87AccuracyVANILLA-VERB
24.593227.001629.4131.8184Jan 5, 2026

Evaluation results

MethodLinks
VANILLA-VERBBackbone=Llama3.1-70b2026.0133.8759.6317.6
COT-VERBBackbone=Llama3.1-70b2026.0133.8758.3934.49
SCBackbone=Llama3.1-70b2026.0133.8732.9928.98
P(TRUE)Backbone=Llama3.1-70b2026.0133.8767.8240.82
P(SUFFICIENT)Backbone=Llama3.1-70b2026.0133.8713.0548.43
VANILLA-VERBBackbone=Qwen2.5-72b2026.0132.3647.9267.97
COT-VERBBackbone=Qwen2.5-72b2026.0132.3651.4372.33
SCBackbone=Qwen2.5-72b2026.0132.3645.6928.9
P(TRUE)Backbone=Qwen2.5-72b2026.0132.3642.1268.88
P(SUFFICIENT)Backbone=Qwen2.5-72b2026.0132.3645.8666.81
VANILLA-VERBBackbone=Qwen2.5-7b2026.0125.2258.0561.13
COT-VERBBackbone=Qwen2.5-7b2026.0125.2264.9352.6
SCBackbone=Qwen2.5-7b2026.0125.2245.4413.85
P(TRUE)Backbone=Qwen2.5-7b2026.0125.2246.6847.16
P(SUFFICIENT)Backbone=Qwen2.5-7b2026.0125.2236.6455.24
VANILLA-VERBBackbone=Llama3.1-8b2026.0124.9567.82-6.36
COT-VERBBackbone=Llama3.1-8b2026.0124.9563.7546.97
SCBackbone=Llama3.1-8b2026.0124.9518.0536.73
P(TRUE)Backbone=Llama3.1-8b2026.0124.9569.0242.1
P(SUFFICIENT)Backbone=Llama3.1-8b2026.0124.9541.0838.57