its-ai/segmentation_benchmark

这是一个用于AI检测任务的数据集，包含公共和私有数据集。公共数据集提供了文本、分词级别的AI/人类标签以及生成文本所使用的模型名称（如果文本完全由人类生成则为Nan）。私有数据集仅包含文本信息。文本分为AI生成、部分AI生成和完全由人类生成的三种类型。

This is a dataset for AI detection task, including public and private sets. The public set provides text, word-level AI/human labels, and the model name used to generate the text (Nan for fully human texts). The private set contains only text. The text is divided into three types: generated by AI, partially generated by AI, and fully generated by humans.