Paragraph-based Transformer Pre-training for Multi-Sentence Inference
This work addresses the need for better multi-sentence modeling in NLP inference tasks, offering an incremental improvement for researchers and practitioners in question answering and fact-checking.
The paper tackles the problem of poor performance of pre-trained transformers on multi-candidate inference tasks like answer sentence selection and fact verification by proposing a new paragraph-level pre-training objective, resulting in superior performance over traditional methods on three AS2 and one fact verification datasets.
Inference tasks such as answer sentence selection (AS2) or fact verification are typically solved by fine-tuning transformer-based models as individual sentence-pair classifiers. Recent studies show that these tasks benefit from modeling dependencies across multiple candidate sentences jointly. In this paper, we first show that popular pre-trained transformers perform poorly when used for fine-tuning on multi-candidate inference tasks. We then propose a new pre-training objective that models the paragraph-level semantics across multiple input sentences. Our evaluation on three AS2 and one fact verification datasets demonstrates the superiority of our pre-training technique over the traditional ones for transformers used as joint models for multi-candidate inference tasks, as well as when used as cross-encoders for sentence-pair formulations of these tasks. Our code and pre-trained models are released at https://github.com/amazon-research/wqa-multi-sentence-inference .