Share your thoughts, 1 month free Claude Pro on usSee more

Uncertainty Estimation on CommonsenseQA

0.74AUROC

Geometry-Calibrated Conformal Abstention

Updated 1mo ago

Evaluation Results

Method	Links
Geometry-Calibrated Conformal Abstention 2026.04		0.74	0.9
Proposed 2026.04		0.71	0.9
Perplexity 2026.04		0.69	0.87
SAR 2026.04		0.69	0.88
Focus 2026.04		0.69	0.86
Eigen 2026.04		0.69	0.88
Eigen 2026.04		0.68	0.87
Perplexity 2026.04		0.68	0.88
SAR 2026.04		0.68	0.88
Semantic 2026.04		0.68	0.88
Focus 2026.04		0.67	0.86
Semantic 2026.04		0.66	0.87
ATRMD 2026.04		0.63	0.85
Attention 2026.04		0.54	0.81
ATRMD 2026.04		0.53	0.8
P(True) 2026.04		0.53	0.82
Attention 2026.04		0.5	0.77
P(True) 2026.04		0.47	0.76