List your company

Confidence Estimation on TRICKME

53.88AccuracyVANILLA-VERB
33.267238.618643.9749.3214Jan 5, 2026

Evaluation results

MethodLinks
VANILLA-VERBBackbone=Qwen2.5-72b2026.0153.8841.6271.32
COT-VERBBackbone=Qwen2.5-72b2026.0153.8843.570.35
SCBackbone=Qwen2.5-72b2026.0153.8836.548.52
P(TRUE)Backbone=Qwen2.5-72b2026.0153.8832.8769.28
P(SUFFICIENT)Backbone=Qwen2.5-72b2026.0153.8832.4171.24
VANILLA-VERBBackbone=Llama3.1-70b2026.0153.7547.4744.49
COT-VERBBackbone=Llama3.1-70b2026.0153.7580.9757.27
SCBackbone=Llama3.1-70b2026.0153.7519.938.26
P(TRUE)Backbone=Llama3.1-70b2026.0153.7535.6264.25
P(SUFFICIENT)Backbone=Llama3.1-70b2026.0153.7523.1671.38
VANILLA-VERBBackbone=Llama3.1-8b2026.0141.9159.2654.55
COT-VERBBackbone=Llama3.1-8b2026.0141.9187.748.63
SCBackbone=Llama3.1-8b2026.0141.9118.9755.37
P(TRUE)Backbone=Llama3.1-8b2026.0141.9155.6152.37
P(SUFFICIENT)Backbone=Llama3.1-8b2026.0141.9133.7458.34
VANILLA-VERBBackbone=Qwen2.5-7b2026.0134.0654.6256.39
COT-VERBBackbone=Qwen2.5-7b2026.0134.0665.3749.89
SCBackbone=Qwen2.5-7b2026.0134.0633.2842.54
P(TRUE)Backbone=Qwen2.5-7b2026.0134.0639.4551
P(SUFFICIENT)Backbone=Qwen2.5-7b2026.0134.0635.2652.11