You Only Read Once (YORO): Learning to Internalize Database Knowledge for Text-to-SQL

While significant progress has been made on the text-to-SQL task, recent solutions repeatedly encode the same database schema for every question, resulting in unnecessary high inference cost and often overlooking crucial database knowledge. To address these issues, we propose You Only Read Once (YOR...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2024-09
Main Authors: Kobayashi, Hideo, Lan, Wuwei, Shi, Peng, Chang, Shuaichen, Guo, Jiang, Zhu, Henghui, Wang, Zhiguo, Ng, Patrick
Format: Article
Language:English
Subjects:
Inference
Query languages
Questions
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:While significant progress has been made on the text-to-SQL task, recent solutions repeatedly encode the same database schema for every question, resulting in unnecessary high inference cost and often overlooking crucial database knowledge. To address these issues, we propose You Only Read Once (YORO), a novel paradigm that directly internalizes database knowledge into the parametric knowledge of a text-to-SQL model during training and eliminates the need for schema encoding during inference. YORO significantly reduces the input token length by 66%-98%. Despite its shorter inputs, our empirical results demonstrate YORO's competitive performances with traditional systems on three benchmarks as well as its significant outperformance on large databases. Furthermore, YORO excels in handling questions with challenging value retrievals such as abbreviation.
ISSN:2331-8422