List your company

Uncertainty Estimation on NQ-open

75.88AUROCSE+MCH
54.164859.802465.4471.0776May 12, 2025Jul 7, 2025Sep 2, 2025Oct 29, 2025Dec 25, 2025Feb 20, 2026Apr 18, 2026

Evaluation results

MethodLinks
SE+MCHBase Model=Llama2-chat, Retriever=Contriver, Calibration=MCH2025.0575.8888.0232.5291.5346.9
EigV+MCHBase Model=Llama2-chat, Retriever=Contriver, Calibration=MCH2025.0573.688.6723.9393.8234.88
ECC+MCHBase Model=Llama2-chat, Retriever=Contriver, Calibration=MCH2025.0571.8186.0826.9989.9339.54
PE+MCHBase Model=Llama2-chat, Retriever=Contriver, Calibration=MCH2025.0569.8576.0537.4283.7551.93
TotalBackbone=Mistral-7B2026.0469.8————
SE+NLIBase Model=Llama2-chat, Retriever=Contriver, Calibration=NLI2025.0569.7779.6140.4986.7250
EigV+NLIBase Model=Llama2-chat, Retriever=Contriver, Calibration=NLI2025.0569.4480.9127.6191.7635.27
AleatoricBackbone=Mistral-7B2026.0469————
ECC+NLIBase Model=Llama2-chat, Retriever=Contriver, Calibration=NLI2025.0568.9678.6432.5287.1842.64
SelfCheckGPTBackbone=Mistral-7B2026.0468.8————
SEBase Model=Llama2-chat, Retriever=Contriver, Calibration=None2025.0567.4977.3533.7591.5336.05
Mean Token EntropyBackbone=Mistral-7B2026.0467.4————
ECC+CTIBase Model=Llama2-chat, Retriever=Contriver, Calibration=CTI2025.0567.3775.7323.3187.1837.98
PE+NLIBase Model=Llama2-chat, Retriever=Contriver, Calibration=NLI2025.0567.2166.0247.2477.5755.43
Token EntropyBackbone=Mistral-7B2026.0467.2————
SE+CTIBase Model=Llama2-chat, Retriever=Contriver, Calibration=CTI2025.0567.0977.0225.7689.4740.31
EigV+CTIBase Model=Llama2-chat, Retriever=Contriver, Calibration=CTI2025.0566.8277.3520.2590.1634.5
SemanticEntropyBackbone=Mistral-7B2026.0465.6————
PE+CTIBase Model=Llama2-chat, Retriever=Contriver, Calibration=CTI2025.0565.3861.4944.1776.4353.88
PerplexityBackbone=Mistral-7B2026.0465.2————
Max Token Prob.Backbone=Mistral-7B2026.0465.2————
PEBase Model=Llama2-chat, Retriever=Contriver, Calibration=None2025.0564.8760.1947.8577.3551.16
SC + VCBackbone=Mistral-7B2026.0464.1————
Max Sequence Prob.Backbone=Mistral-7B2026.0464————
EigVBase Model=Llama2-chat, Retriever=Contriver, Calibration=None2025.0563.9465.3712.8888.5624.42
ECCBase Model=Llama2-chat, Retriever=Contriver, Calibration=None2025.0563.5761.499.8283.0618.99
SC Based VCBackbone=Mistral-7B2026.0463————
SC ScoreBackbone=Mistral-7B2026.0461.6————
Self CertaintyBackbone=Mistral-7B2026.0459.8————
Closeness CentralityBackbone=Mistral-7B2026.0459.7————
Kernel Lang. Ent.Backbone=Mistral-7B2026.0458.5————
PTrueBackbone=Mistral-7B2026.0455————