Evaluating Hallucinations in Chinese Large Language Models