Technical report on Conversational Question Answering

TL;DR

This paper presents a new conversational question answering system combining multiple techniques, achieving state-of-the-art performance on CoQA without data augmentation.

Contribution

It introduces a novel system integrating rationale tagging, adversarial training, knowledge distillation, and post-processing for improved conversational QA.

Findings

Achieved 90.4 F1 on CoQA test set, surpassing previous single models.

Demonstrated effectiveness of multi-task and adversarial training in conversational QA.

No data augmentation needed for top performance.

Abstract

Conversational Question Answering is a challenging task since it requires understanding of conversational history. In this project, we propose a new system RoBERTa + AT +KD, which involves rationale tagging multi-task, adversarial training, knowledge distillation and a linguistic post-process strategy. Our single model achieves 90.4(F1) on the CoQA test set without data augmentation, outperforming the current state-of-the-art single model by 2.6% F1.