#LLM Hallucinations

MoE Models Tackle LLM Hallucinations

InnerExpert leverages MoE architecture's internal signals for per-token hallucination detection, achieving state-of-the-art results with high efficiency.