stanford-oval/churro-dataset
收藏官方服务：
资源简介：
CHURRO数据集是一个包含大约10万页，跨越22个世纪和46种语言集群的历史文档数据集。它用于训练历史文档转录的开放权重视觉语言模型，包含了手写和印刷文本。
CHURRO-DS is a dataset consisting of approximately 100K pages from 155 historical collections spanning 22 centuries and 46 language clusters. It is used for training open-weight vision-language models for historical document transcription, including both handwritten and printed text.
提供机构：
stanford-oval
