Fruit Shop Dataset

该数据集是一个模拟水果店管理的综合数据集，涵盖了诸如采购、销售、调价以及商品退货等操作。此外，该数据集包含50个带有标注标准答案的问题，其中15个为简单问题，35个为难题，旨在测试多跳推理能力。规模上，数据集包含70条记录，大约3.3千个标记。任务旨在评估大型语言模型在复杂推理场景中的性能，使用SQL数据库作为符号记忆。

This is a comprehensive dataset simulating fruit shop management, covering operations such as procurement, sales, price adjustment, and product returns. Additionally, this dataset contains 50 questions with annotated standard answers, 15 of which are simple questions and 35 are complex ones, aimed at testing multi-hop reasoning capabilities. In terms of scale, the dataset includes 70 records and approximately 3.3 thousand tokens. The task aims to evaluate the performance of Large Language Models (LLMs) in complex reasoning scenarios, using SQL databases as symbolic memory.