BriefGPT.xyz
Feb, 2021
AuGPT：使用辅助任务和数据增强进行端到端对话的预训练语言模型
AuGPT: Dialogue with Pre-trained Language Models and Data Augmentation
HTML
PDF
Jonáš Kulhánek, Vojtěch Hudeček, Tomáš Nekvinda, Ondřej Dušek
TL;DR
为了解决注意力语言模型在任务导向对话中的缺陷，这篇论文引入了修改过的训练目标和巨量数据增强技术，研究数据来源的多重组合方式，并通过人工和自动评估证明了方法的高效性，取得了与最先进技术的竞争性表现。
Abstract
attention-based pre-trained language models
such as GPT-2 brought considerable progress to
end-to-end dialogue modelling
. However, they also present considerable risks for task-oriented dialogue, such as lack of
→