DBGorilla

该数据集通过DBGorilla排行榜展示了在不同评估指标下，模型性能的明显层次结构，涵盖了8个大型语言模型在不同查询复杂度上的表现。此外，数据集还包括了诸如精确匹配分数、抽象语法树分数等性能指标，并对涉及地面真实查询的API组件进行了详细分解。该数据集的测试查询规模为315个，其任务是评估大型语言模型在数据库查询方面的性能。

This dataset reveals the clear hierarchical structure of model performance across diverse evaluation metrics via the DBGorilla Leaderboard, covering the performance of 8 large language models (LLMs) at varying query complexity levels. Additionally, the dataset encompasses performance metrics such as Exact Match score and Abstract Syntax Tree (AST) score, and conducts detailed decomposition of API components related to ground-truth queries. This dataset consists of 315 test queries, with its core task being to evaluate the performance of large language models in database query scenarios.