10万对中文大模型通用领域复杂指令跟随SFT文本数据集【数据堂】

10万对中文复杂的prompt指令，字数在50~400字之间，每个prompt中的约束条件不低于3个，用于训练提升大模型的指令跟随性。类别覆盖生成类（写新闻稿、采访提纲、文案创作、文稿校对、中英作文、语法学习、研究报告、学习计划、诗歌创作、美食介绍、广告软文、销售话术、公文辅助写作、公文审核、政策文件问答等）、重写类（改写句子、文本纠错、句子合并、简化文案）、摘要类（内容摘要）、提取类（事件要素提取、观点提取、关键词提取、立场抽取、实体抽取）。所有prompt均为人工编写，满足多样性覆盖

100,000 pairs of Chinese complex prompt instructions, with word counts ranging from 50 to 400. Each prompt contains no fewer than 3 constraints, and is designed to train and enhance the instruction-following capabilities of large language models (LLMs). The covered categories include: Generative tasks (news article writing, interview outline creation, copywriting, document proofreading, Chinese-English composition writing, grammar learning, research report writing, study plan formulation, poetry creation, food introduction, advertorial writing, sales pitch crafting, official document auxiliary writing, official document review, policy document Q&A, etc.), Rewriting tasks (sentence rewriting, text error correction, sentence merging, copy simplification), Summarization tasks (content summarization), and Extraction tasks (event element extraction, viewpoint extraction, keyword extraction, stance extraction, entity extraction). All prompts are manually written to ensure comprehensive diversity coverage.

搜集汇总

背景与挑战

背景概述

该数据集包含10万对中文复杂指令，涵盖生成、重写、摘要和提取四大类，每个指令包含至少3个约束条件，专为提升大模型指令跟随能力而设计，所有数据均为人工编写以确保多样性。