ResearchBenchmarksFactual Claim Detection on PoliClaim Inconsistent samples S_inc (test)Follow63.6AgreementExperts31.8840.11548.3556.585Feb 16, 2024Evaluation ResultsMethodMethodLinksAgreementAccuracyExpertsSubset Reference=GPT-3...Subset Reference=GPT-3.5 consistency (dagger)2024.0263.691.99ExpertsSubset Reference=GPT-4...Subset Reference=GPT-4 consistency (double dagger)2024.0262.990.79AFaCTABackbone=GPT-4Backbone=GPT-42024.0241.874.64AFaCTABackbone=GPT-3.5Backbone=GPT-3.52024.0233.165.8