data-maki/sql-grimoire

Grimoire of SQL是一个专门用于训练和评估文本到SQL模型的数据集。它整合并改进了多个现有数据集（如Spider、BirdBench和Gretel），通过纠正错误、优化自然语言查询和验证SQL查询的可执行性来提高数据质量。数据集特别设计用于支持如GPT-4及其变体的高质量微调，确保强大的SQL生成能力。数据集包含约104,000个示例，每个示例包括数据集来源、数据库ID、修正后的用户查询、原始查询、修正后的SQL查询、上下文信息、示例行和难度评级。数据集强调数据质量，通过SQL修正、查询优化和一致性增强来提高数据集的可靠性。

Grimoire of SQL is a comprehensive dataset tailored for training and evaluating text-to-SQL models. It consolidates and enhances multiple existing datasets, including Spider, BirdBench, and Gretel, by correcting errors, refining natural language queries, and validating SQL queries for runnability. The dataset is specifically designed to support high-quality fine-tuning of models like GPT-4 and its variants, ensuring robust SQL generation capabilities. The dataset contains approximately 104,000 examples, each including the dataset source, database ID, corrected user query, original query, corrected SQL query, context information, example rows, and difficulty rating. The dataset emphasizes data quality, improving reliability through SQL corrections, query refinements, and consistency enhancements.