anonymous-Data-Preparation-Bench/Data-Prep-Bench
收藏资源简介：
该数据集是一个用于大型语言模型(LLM)监督微调(SFT)和评估的综合资源，覆盖六个领域：金融、医学、法律、数学、科学和通用领域。一个关键特点是，我们使用了12种不同的数据生成方法（包括基于代理的方法、DataFlow系列、纯LLM生成和SKILL方法），并利用多个前沿模型（如GPT-5, Claude Opus 4.6, Gemini 3.0 Pro等）处理原始语料库，生成高质量的问答对。此外，该仓库提供了用于模型评估的标准化基准文件。
This dataset is a comprehensive resource for supervised fine-tuning (SFT) and evaluation of Large Language Models (LLMs), covering six domains: finance, medicine, law, mathematics, science, and general-purpose domains. A key feature is that we adopted 12 distinct data generation methods, including agent-based approaches, the DataFlow series, pure LLM generation, and the SKILL method, and utilized multiple cutting-edge models such as GPT-5, Claude Opus 4.6, Gemini 3.0 Pro to process the original corpus and generate high-quality question-answer pairs. Additionally, this repository provides standardized benchmark files for model evaluation.
