[2407.04121] Hallucination Detection: Robustly Discerning Reliable Answers in Large Language Models