LLMTeamAkiyama/cleand_flatlander1024_or_instruct_dedup
收藏Hugging Face2025-08-09 更新2025-09-13 收录
下载链接：
https://hf-mirror.com/datasets/LLMTeamAkiyama/cleand_flatlander1024_or_instruct_dedup
下载链接
链接失效反馈官方服务：
资源简介：
这是一个用于问答任务的数据集，包含问题、思考和答案三个部分。数据集共有2600条数据，平均每个条目的token数为1340，最大token数为3086，总token数为3484377。数据集以JSONL格式存储，经过一系列处理，如重复数据删除、文本长度过滤等，最终以250MB为上限进行分割存储。
This is a question-answering dataset containing three parts: question, thought, and answer. The dataset includes 2,600 entries with an average token count of 1,340 per entry, a maximum token count of 3,086, and a total of 3,484,377 tokens. The dataset is stored in JSONL format and has undergone several processing steps, including duplicate removal and text length filtering, before being split into chunks with a size limit of 250MB.
提供机构：
LLMTeamAkiyama
