Multi-Task Learning for Situated Multi-Domain End-to-End Dialogue Systems
This work addresses the problem of creating more versatile dialogue systems for NLP applications, though it appears incremental as it builds on existing GPT-2 approaches.
The authors tackled the challenge of building task-oriented dialogue systems that handle multiple domains and modalities by applying multi-task learning to a GPT-2 model, achieving better performance across all sub-tasks compared to specialized models.
Task-oriented dialogue systems have been a promising area in the NLP field. Previous work showed the effectiveness of using a single GPT-2 based model to predict belief states and responses via causal language modeling. In this paper, we leverage multi-task learning techniques to train a GPT-2 based model on a more challenging dataset with multiple domains, multiple modalities, and more diversity in output formats. Using only a single model, our method achieves better performance on all sub-tasks, across domains, compared to task and domain-specific models. Furthermore, we evaluated several proposed strategies for GPT-2 based dialogue systems with comprehensive ablation studies, showing that all techniques can further improve the performance.