상세정보
학술대회
Improving End-to-End Task-Oriented Dialogue System with A Simple Auxiliary Task
- 저자
- 이요한
- 발행일
- 202111
- 출처
- Findings of the Association for Computational Linguistics: EMNLP 2021, pp.1296-1303
- 협약과제
- 21HS2800, 준지도학습형 언어지능 원천기술 및 이에 기반한 외국인 지원용 한국어 튜터링 서비스 개발, 이윤근
- 초록
- The paradigm of leveraging large pretrained language models has made significant progress on benchmarks on task-oriented dialogue (TOD) systems. In this paper, we combine this paradigm with multi-task learning framework for end-toend TOD modeling by adopting span prediction as an auxiliary task. In end-toend setting, our model achieves new stateof-the-art results with combined scores of 108.3 and 107.5 on MultiWOZ 2.0 and MultiWOZ 2.1, respectively. Furthermore, we demonstrate that multi-task learning improves not only the performance of model but its generalization capability through domain adaptation experiments in the few-shot setting. The code is available at github.com/bepoetree/MTTOD.
- KSP 제안 키워드
- End to End(E2E), Generalization capability, Language model, Learning framework, Performance of model, Task-oriented, dialogue system, domain adaptation, multi-task learning
본 저작물은 크리에이티브 커먼즈 저작자 표시 (CC BY) 조건에 따라 이용할 수 있습니다.
