Synergizing In-context Learning with Hints for End-to-end Task-oriented Dialog Systems
This addresses the problem of response alignment and comprehensibility for users of task-oriented dialog systems in low-data scenarios, representing an incremental improvement.
The paper tackles the problem of large language model-based task-oriented dialog systems generating misaligned and overly comprehensive responses in low-data settings by proposing SyncTOD, which synergizes LLMs with task-specific hints and exemplar selection. The result shows that SyncTOD with ChatGPT achieves superior performance compared to LLM-based baselines and state-of-the-art models in low-data settings while remaining competitive with full data.
End-to-end Task-Oriented Dialog (TOD) systems typically require extensive training datasets to perform well. In contrast, large language model (LLM) based TOD systems can excel even with limited data due to their ability to learn tasks through in-context exemplars. However, these models lack alignment with the style of responses in training data and often generate comprehensive responses, making it difficult for users to grasp the information quickly. In response, we propose SyncTOD that synergizes LLMs with task-specific hints to improve alignment in low-data settings. SyncTOD employs small auxiliary models to provide hints and select exemplars for in-context prompts. With ChatGPT, SyncTOD achieves superior performance compared to LLM-based baselines and SoTA models in low-data settings, while retaining competitive performance in full-data settings.