面向任务的端到端对话系统的任务优化适配器（英文版）

英文标题：Task-Optimized Adapters for an End-to-End Task-Oriented Dialogue System中文摘要：该研究提出了一种基于 Task-Optimized Adapters 和强化学习的 End-to-end TOD 系统，能够独立学习每个任务，并在 MultiWOZ 基准测试上表现出优越的性能，特别是在 2.2 数据集上的 DST 任务达到了最先进的水平。英文摘要：Task-Oriented Dialogue (TOD) systems are designed to carry out specific tasksby tracking dialogue states and generating appropriate responses to help usersachieve defined goals. Recently, end-to-end dialogue models pre-trained basedon large datasets have shown promising performance in