jun-2018/multimodal_doc_data

这是一个包含新闻稿和报告文档的数据集，分为训练集和验证集。数据集中的每个文档都有唯一的ID、标题、文本内容等属性，并提供了文本的OCR识别结果和结构化文本信息。数据集的总大小为329GB，下载大小为155GB。This dataset consists of press releases and reports, divided into training and validation sets. Each document in the dataset has a unique ID, title, text content, and other attributes, and provides OCR recognition results and structured text information. The total size of the dataset is 329GB, with a download size of 155GB.