ResearchBenchmarksKnowledge-intensive Question Answering on MuSiQue (Truthfulness, Hallucination, Accuracy)Follow-0.9Truthfulness Score (T)TruthRL-1.732-1.516-1.3-1.084Sep 30, 2025Evaluation ResultsMethodMethodLinksTruthfulness Score (T)Hallucination Rate (H)Accuracy (A)TruthRLRetrieval Setting=With...Retrieval Setting=With Retrieval, Backbone=Llama3.1-8B-Inst2025.09-0.915.915TruthRLRetrieval Setting=With...Retrieval Setting=Without Retrieval, Backbone=Qwen2.5-7B-Inst2025.09-1.75.33.6