vindows/qwen2.5-text-to-sql-benchmark-data

该数据集包含用于微调Qwen2.5模型的Spider基准测试数据和评估结果。主要内容包括Spider基准训练数据（约7000个示例）、开发集（1034个示例）、表结构（tables.json）以及官方评估脚本。此外，还包含Spider评估结果、基准比较文档、模型比较结果以及相关的Python脚本和训练日志。数据集用于评估Qwen2.5模型在Spider基准测试中的性能，包括归一化准确率、组件准确率、精确匹配和平均相似度等指标。

This dataset contains Spider benchmark data and evaluation results for fine-tuned Qwen2.5 models. The main contents include Spider benchmark training data (~7,000 examples), dev set (1,034 examples), table schemas (tables.json), and official evaluation scripts. Additionally, it includes Spider evaluation results, benchmark comparison documentation, model comparison results, and related Python scripts and training logs. The dataset is used to evaluate the performance of Qwen2.5 models on the Spider benchmark, including metrics such as normalized accuracy, component accuracy, exact match, and average similarity.

搜集汇总

背景与挑战

背景概述

该数据集包含Spider基准测试数据和Qwen2.5模型的评估结果，用于文本到SQL任务的性能评估。提供了训练数据、开发集、评估脚本和模型比较工具，支持模型性能的详细分析。