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

Uncertainty Estimation on CoQA

0.857AUROC

Logit Magnitude

Updated 26d ago

Evaluation Results

Method
Logit Magnitude 2026.05	0.857	0.685	78.5	24.4
Logit Magnitude 2026.05	0.855	0.721	78.1	13.5
Logit Magnitude 2026.05	0.83	0.699	75.6	13.8
LogTokU 2026.05	0.822	0.695	73.6	13.8
Total 2026.04	0.819	-	-	-
LogTokU 2026.05	0.819	0.704	73.5	13.5
LogTokU 2026.05	0.816	0.663	73.5	24.4
MetaUE 2026.05	0.807	0.692	73.9	0
MetaUE 2026.05	0.794	0.676	72.2	0
Logit Magnitude 2026.05	0.793	0.604	72.8	19.4
Logit Magnitude 2026.05	0.791	0.603	72.7	16.3
Logit Magnitude 2026.05	0.791	0.656	72.3	18.6
LogTokU 2026.05	0.781	0.652	70.4	18.6
LogTokU 2026.05	0.773	0.516	69.6	23.1
LogTokU 2026.05	0.771	0.587	69.4	19.4
MetaUE 2026.05	0.765	0.638	70.1	0
Logit Magnitude 2026.05	0.756	0.504	70.3	23.1
Semantic Entropy 2026.05	0.754	0.575	70.2	195.9
LogTokU 2026.05	0.751	0.499	67.5	18.9
Logit Magnitude 2026.05	0.733	0.488	68.2	17.4
Logit Magnitude 2026.05	0.731	0.482	67.7	18.9
Logit Magnitude 2026.05	0.73	0.486	67.5	15
MetaUE 2026.05	0.724	0.564	67.3	0
LogTokU 2026.05	0.723	0.484	68.1	17.4
Self-Certainty 2026.05	0.717	0.474	65.8	23.1
Self-Consistency 2026.05	0.714	0.561	65.1	195.9
Self-Certainty 2026.05	0.71	0.627	64.9	13.5
MetaUE 2026.05	0.709	0.475	65.6	0
Hidden-State Score 2026.05	0.704	0.543	65.8	19.4
Self-Certainty 2026.05	0.703	0.471	64.7	18.9
MetaUE 2026.05	0.701	0.601	65.2	0
Sequence Entropy 2026.05	0.68	0.542	63.1	19.4
MetaUE 2026.05	0.679	0.453	63.5	0
Semantic Entropy 2026.05	0.676	0.443	63.8	173
Self-Certainty 2026.05	0.675	0.599	62.5	13.8
Self-Consistency 2026.05	0.667	0.445	62	173
Self-Certainty 2026.05	0.66	0.445	66.1	17.4
MetaUE 2026.05	0.656	0.43	62.4	0
Hidden-State Score 2026.05	0.654	0.445	66.5	17.4
Self-Certainty 2026.05	0.651	0.52	61.4	19.4
Sequence Entropy 2026.05	0.649	0.436	60.5	17.4
Kernel Lang. Ent. 2026.04	0.642	-	-	-
Closeness Centrality 2026.04	0.612	-	-	-
Self-Certainty 2026.05	0.608	0.544	57.8	18.6
Self-Certainty 2026.05	0.591	0.528	56.3	24.4
SemanticEntropy 2026.04	0.561	-	-	-
Max Sequence Prob. 2026.04	0.561	-	-	-
PTrue 2026.04	0.555	-	-	-
SC Score 2026.04	0.553	-	-	-
Perplexity 2026.04	0.513	-	-	-
Max Token Prob. 2026.04	0.51	-	-	-
Self Certainty 2026.04	0.51	-	-	-
Token Entropy 2026.04	0.502	-	-	-
Mean Token Entropy 2026.04	0.498	-	-	-
SC Based VC 2026.04	0.483	-	-	-
Aleatoric 2026.04	0.458	-	-	-
SelfCheckGPT 2026.04	0.454	-	-	-
SC + VC 2026.04	0.411	-	-	-