An Empirical Study of Pre-trained Language Models in Simple Knowledge Graph Question Answering