ResearchBenchmarksHallucination monitoring on Modern History domainFollow84AUROCHalMit46.5656.286675.72Jul 21, 2025Evaluation ResultsMethodMethodLinksAUROCAUC-PRF1 ScoreAccuracyHalMitBackbone=Llama2Backbone=Llama22025.0784807789HalMitBackbone=Llama3.1Backbone=Llama3.12025.0784846789SCGBackbone=Llama2Backbone=Llama22025.0778798182SCGBackbone=Llama3.1Backbone=Llama3.12025.0778776076PPBackbone=Llama2Backbone=Llama22025.0774767379ICLBackbone=Llama2Backbone=Llama22025.0769676467ICLBackbone=Llama3.1Backbone=Llama3.12025.0759555461PPBackbone=Llama3.1Backbone=Llama3.12025.0748454656