List your company
ResearchFormerly WizWand

First-order logical inference on FOLIO-wiki (testable)

87.45AccuracyCR
31.165245.777660.3975.0024Aug 8, 2023
Updated 8mo ago

Evaluation results

MethodLinks
CRModel=GPT-4, n=22023.0887.45
CoT-SCModel=GPT-4, k=162023.0885.02
CoTModel=GPT-42023.0884.46
DirectModel=GPT-42023.0880.52
CRModel=GPT-3.5-turbo, n=22023.0873.03
CRModel=LLaMA-65B, n=22023.0872.1
CoT-SCModel=LLaMA-65B, k=162023.0870.79
DirectModel=LLaMA-65B2023.0867.42
CoTModel=LLaMA-65B2023.0867.42
CoTModel=GPT-3.5-turbo2023.0864.61
CoT-SCModel=GPT-3.5-turbo, k=162023.0863.33
DirectModel=GPT-3.5-turbo2023.0862.92
CRModel=LLaMA 13B, n=22023.0853.37
CoT-SCModel=LLaMA 13B, k=162023.0852.43
CoTModel=LLaMA 13B2023.0849.06
DirectModel=LLaMA 13B2023.0844.75
[Random]2023.0833.33