Simplifying Paragraph-Level Question Generation via Transformer Language Models

Question Generation (QG) is an important task in Natural Language Processing (NLP) that involves generating questions automatically when given a context paragraph. While many techniques exist for the of QG, they employ complex model architectures, extensive features, and additional mechanisms to boost performance. In this work, we show transformer-based finetuning can be used create robust question generation systems using only single pretrained language model, without use mechanisms, answer metadata, features. Our best outperforms previous more RNN-based Seq2Seq models, with 8.62 14.27 increase METEOR ROUGE_L scores, respectively. We it also performs on par models answer-awareness other special despite being single-model system. analyze how various factors affect model’s performance, such as input data formatting, length paragraphs, answer-awareness. Lastly, look into failure modes identify possible reasons why fails.