Chain-of-Skills: A Configurable Model for Open-domain Question Answering

Autor:	Ma, Kaixin, Cheng, Hao, Zhang, Yu, Liu, Xiaodong, Nyberg, Eric, Gao, Jianfeng
Rok vydání:	2023
Předmět:	Computer Science - Computation and Language
Druh dokumentu:	Working Paper
Popis:	The retrieval model is an indispensable component for real-world knowledge-intensive tasks, e.g., open-domain question answering (ODQA). As separate retrieval skills are annotated for different datasets, recent work focuses on customized methods, limiting the model transferability and scalability. In this work, we propose a modular retriever where individual modules correspond to key skills that can be reused across datasets. Our approach supports flexible skill configurations based on the target domain to boost performance. To mitigate task interference, we design a novel modularization parameterization inspired by sparse Transformer. We demonstrate that our model can benefit from self-supervised pretraining on Wikipedia and fine-tuning using multiple ODQA datasets, both in a multi-task fashion. Our approach outperforms recent self-supervised retrievers in zero-shot evaluations and achieves state-of-the-art fine-tuned retrieval performance on NQ, HotpotQA and OTT-QA. Comment: ACL 2023
Databáze:	arXiv
Externí odkaz:	http://arxiv.org/abs/2305.03130 Zobrazit plný text záznamu View this record from Arxiv