遇见数据集

AirsideLabs/ADSB

收藏
Hugging Face2025-05-20 更新2025-11-01 收录
官方服务:

资源简介:

这个数据集是一个多样化的文本数据集,包含多个配置,每个配置根据不同的NLP任务具有不同的特征。具体包括:chunked配置提供文本块和其元数据;ingested配置包含文本文件的基本信息;lighteval配置为阅读理解任务设计,包含问题、答案和相关的文本块信息;multi_hop_questions配置和single_shot_questions配置分别用于多跳和单次阅读理解任务;summarized配置提供文档的摘要信息。每个配置都包含训练集,且具有不同的数据大小和示例数量。

This dataset is a diverse text dataset containing multiple configurations, each with different features tailored to different NLP tasks. It includes: chunked configuration providing text chunks and their metadata; ingested configuration containing basic information of text files; lighteval configuration designed for reading comprehension tasks with questions, answers, and related text chunks; multi_hop_questions and single_shot_questions configurations for multi-hop and single-shot reading comprehension tasks respectively; summarized configuration offering document summaries. Each configuration contains a training set with varying data sizes and example counts.

提供机构:
AirsideLabs
二维码
社区交流群
二维码
科研交流群
商业服务