List your company

Uncertainty Estimation on NaturalQA

77AUROCSAR
48.9256.2163.570.79Feb 19, 2024Jul 1, 2024Nov 12, 2024Mar 25, 2025Aug 6, 2025Dec 17, 2025Apr 30, 2026

Evaluation results

MethodLinks
SARBackbone=LLaMA-3.2-3B-Instruct2026.047774
Geometry-Calibrated Conformal AbstentionBackbone=LLaMA-3.2-3B-Instruct2026.047775
SE + MARSBackbone=Mistral-7b2024.0275.5—
SEBackbone=Mistral-7b2024.0275.12—
SE + MARSBackbone=Llama2-13b2024.0273.89—
SEBackbone=Llama2-13b2024.0273.56—
SE + MARSBackbone=Llama2-7b2024.0272.75—
SEBackbone=Llama2-7b2024.0272.47—
Confidence + MARSBackbone=Mistral-7b2024.0271.36—
SE + MARSBackbone=Falcon-7b2024.0271.24—
PerplexityBackbone=LLaMA-3.2-3B-Instruct2026.047167
Confidence + MARSBackbone=Llama2-13b2024.0270.88—
Entropy + MARSBackbone=Mistral-7b2024.0270.71—
Entropy + MARSBackbone=Llama2-13b2024.0270.63—
SEBackbone=Falcon-7b2024.0270.41—
Confidence + MARSBackbone=Llama2-7b2024.0269.81—
ConfidenceBackbone=Mistral-7b2024.0269.54—
SE + MARSBackbone=Llama2-7b-chat2024.0269.43—
Entropy + MARSBackbone=Llama2-7b2024.0269.32—
SEBackbone=Llama2-7b-chat2024.0268.66—
ConfidenceBackbone=Llama2-7b2024.0268.56—
ConfidenceBackbone=Llama2-13b2024.0268.56—
EntropyBackbone=Llama2-13b2024.0268.34—
Confidence + MARSBackbone=Falcon-7b2024.0268.3—
EntropyBackbone=Mistral-7b2024.0268.05—
Confidence + MARSBackbone=Llama2-7b-chat2024.0267.86—
Entropy + MARSBackbone=Falcon-7b2024.0267.51—
Entropy + MARSBackbone=Llama2-7b-chat2024.0267.41—
EntropyBackbone=Llama2-7b2024.0267.08—
ConfidenceBackbone=Llama2-7b-chat2024.0265.98—
EntropyBackbone=Llama2-7b-chat2024.0265.23—
SemanticBackbone=LLaMA-3.2-3B-Instruct2026.046462
ConfidenceBackbone=Falcon-7b2024.0263.78—
EntropyBackbone=Falcon-7b2024.0263.28—
EigenBackbone=LLaMA-3.2-3B-Instruct2026.046362
FocusBackbone=LLaMA-3.2-3B-Instruct2026.046264
ATRMDBackbone=LLaMA-3.2-3B-Instruct2026.045958
P(True)Backbone=LLaMA-3.2-3B-Instruct2026.045555
AttentionBackbone=LLaMA-3.2-3B-Instruct2026.045053