Spider-Realistic
收藏arXiv2025-09-30 收录
下载链接：
https://zenodo.org/record/5205322#.ytts_o5kgab
下载链接
链接失效反馈官方服务：
资源简介：
该数据集名为Spider，包含了1034条英文表述及其对应的SQL查询语句，覆盖了20个不同的数据库架构。它旨在评估文本到SQL模型，允许研究人员提交模型预测的查询语句。此外，Spider数据集根据黄金SQL查询语句的复杂性，被分为四个难度级别：简单、中等、困难和超难。该数据集的规模为1034条表述，所涉及的任务是文本到SQL的转换。
The dataset named Spider contains 1034 pairs of English natural language utterances and their corresponding SQL query statements, spanning 20 distinct database schemas. It is developed to evaluate text-to-SQL models, and allows researchers to submit query statements predicted by their models. Additionally, the Spider dataset is divided into four difficulty tiers based on the complexity of its gold-standard SQL queries: simple, medium, hard, and extra hard. Comprising 1034 such utterance-SQL pairs, the core task of this dataset is text-to-SQL conversion.
提供机构：
Spider
