LLM 对于面向任务的对话系统是否足够？（英文版）

英文标题：Are LLMs All You Need for Task-Oriented Dialogue?中文摘要：本研究旨在研究大型语言模型在多轮任务和与外部数据库交互方面的能力，发现在显式信仰状态跟踪方面，它们表现不如专门的任务特定模型，但是如果给出正确的插槽值，它们表现出将对话引导到成功结局的能力，并且在有真实信仰状态分布或域内示例的情况下，这种能力得到了改进。英文摘要：Instructions-tuned Large Language Models (LLMs) gained recently hugepopularity thanks to their ability to interact with users through conversation.In this work we aim to evaluate their ability to complete multi-turn tasks andinteract with external databases in the context of established task-orienteddialogue