flair/nuner_entropy_splits

这是一个包含三个难度级别（简单、中等、困难）的自然语言处理数据集，用于实体识别任务。数据集特征包括分词（tokens）、命名实体标签（ner_tags）和实体跨度（spans）。每个跨度包含起始和结束位置以及标签。数据集总共包含超过三十亿字节，分为易于处理、中等难度和难度较高的三个部分，分别适用于不同阶段的研究和开发。

This is a natural language processing dataset for named entity recognition task, divided into three difficulty levels: easy, medium, and hard. The dataset features include tokenization (tokens), named entity tags (ner_tags), and entity spans. Each span consists of start and end positions along with a label. The dataset totals over three billion bytes and is split into three parts for easy handling, medium difficulty, and higher difficulty, suitable for different stages of research and development.