Q-DAPS Estimate question difficulty for LLMs via answer plausibility scoring and entropy-based uncertainty.