AuGPT: Auxiliary Tasks and Data Augmentation for End-To-End Dialogue with Pre-Trained Language Models
This work aims to improve the robustness and performance of end-to-end dialogue systems for users by addressing limitations in pre-trained language models.
This paper addresses the issues of knowledge grounding and diversity in end-to-end dialogue modeling using pre-trained language models like GPT-2. The authors introduce modified training objectives and massive data augmentation via back-translation, resulting in substantial outperformance of the baseline on MultiWOZ data and competitive performance with state-of-the-art in both automatic and human evaluations.
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 knowledge grounding or diversity. To address these issues, we introduce modified training objectives for language model finetuning, and we employ massive data augmentation via back-translation to increase the diversity of the training data. We further examine the possibilities of combining data from multiples sources to improve performance on the target dataset. We carefully evaluate our contributions with both human and automatic methods. Our model substantially outperforms the baseline on the MultiWOZ data and shows competitive performance with state of the art in both automatic and human evaluation.