Synergizing In-context Learning with Hints for End-to-end Task-oriented Dialog Systems

TL;DR

SyncTOD is proposed that synergizes LLMs with task-specific hints to improve alignment in low-data settings and achieves superior performance compared to LLM-based baselines and SoTA models in low-data settings, while retaining competitive performance in full-data settings.

Abstract

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.

Figures (3)

• Figure 1: SyncTOD predicts useful hints $\hat{H}$ about the expected response. The hints improve exemplar quality via re-ranking and steer the LLM (accessed via API) toward the expected response from within the prompt.
• Figure 2: SyncTOD performance across varying training data sizes.
• Figure 3: Human Evaluation Portal