ServiceNow/BigDocs-7.5M
收藏Hugging Face2025-06-20 更新2025-08-09 收录
下载链接：
https://hf-mirror.com/datasets/ServiceNow/BigDocs-7.5M
下载链接
链接失效反馈官方服务：
资源简介：
BigDocs-7.5M 是一个用于训练多模态模型在文档和代码任务上的开放和许可数据集。该数据集包含了多个子数据集，每个子数据集都有详细的配置信息，包括数据文件路径、特征描述、数据大小和示例数量。数据加载指南解释了如何处理缺少 image 列的数据集，并提供了一个 Python 脚本来替换图像。数据集使用 CC-BY-4.0 许可证，并且记录了每个数据集的依赖关系。
BigDocs-7.5M is an open and permissively-licensed dataset for training multimodal models on document and code tasks. The dataset consists of multiple sub-datasets, each with detailed configuration information including data file paths, feature descriptions, dataset size, and number of examples. The guide on data loading explains how to handle datasets with missing image columns by using img_id and provides a Python script for substituting images. The dataset is licensed under CC-BY-4.0 and documents dependencies for each dataset.
提供机构：
ServiceNow
