List your company
ResearchFormerly WizWand

Multi-hop Reasoning on HotpotQA

67.19AUROCCoT-UQ
47.253252.429157.60562.7809Feb 24, 2025
Updated 8mo ago

Evaluation results

MethodLinks
CoT-UQModel=Llama 2-13B, Strategy=AP, Base Aggregation=Probas-min2025.0267.19
CoT-UQModel=Llama 2-13B, Strategy=AP, Base Aggregation=Probas-mean2025.0266.56
CoT-UQModel=Llama 2-13B, Strategy=AP, Base Aggregation=TOKENSAR2025.0266.29
CoT-UQModel=Llama 3.1-8B, Strategy=AP, Base Aggregation=Probas-min2025.0264.37
CoT-UQModel=Llama 2-13B, Strategy=SE, Base Strategy=Self-Probing, Refinement=KEYKeywords2025.0264.03
CoT-UQModel=Llama 3.1-8B, Strategy=SE, Base Strategy=P(True), Refinement=KEYKeywords2025.0263.1
P(True)Model=Llama 3.1-8B, Strategy=SE2025.0262.39
CoT-UQModel=Llama 3.1-8B, Strategy=AP, Base Aggregation=Probas-mean2025.0262.01
CoT-UQModel=Llama 3.1-8B, Strategy=AP, Base Aggregation=TOKENSAR2025.0261.07
Self-ProbingModel=Llama 2-13B, Strategy=SE2025.0260.63
TOKENSAR + CoT-UQStrategy=AP, Backbone=Mistral 7B2025.0258.59
Probas-minModel=Llama 3.1-8B, Strategy=AP2025.0258.34
Probas-mean + CoT-UQStrategy=AP, Backbone=Mistral 7B2025.0258.24
TOKENSARModel=Llama 2-13B, Strategy=AP2025.0257.33
CoT-UQModel=Llama 3.1-8B, Strategy=SE, Base Strategy=Self-Probing, Refinement=KEYKeywords2025.0257.2
CoT-UQModel=Llama 2-13B, Strategy=SE, Base Strategy=P(True), Refinement=KEYKeywords2025.0257.1
Probas-minModel=Llama 2-13B, Strategy=AP2025.0256.51
Probas-meanModel=Llama 2-13B, Strategy=AP2025.0256.27
TOKENSARStrategy=AP, Backbone=Mistral 7B2025.0255.62
Probas-meanStrategy=AP, Backbone=Mistral 7B2025.0254.63
Self-ProbingModel=Llama 3.1-8B, Strategy=SE2025.0254.33
Probas-meanModel=Llama 3.1-8B, Strategy=AP2025.0253.73
TOKENSARModel=Llama 3.1-8B, Strategy=AP2025.0253.57
P(True)Model=Llama 2-13B, Strategy=SE2025.0251.13
P(True) + CoT-UQStrategy=SE, Backbone=Mistral 7B2025.0250.18
P(True)Strategy=SE, Backbone=Mistral 7B2025.0248.02