5CD-AI/Viet-OCR-VQA-flash2

该数据集是一个多模态数据集，包含了图像和文本信息。具体来说，每个样本包括一个唯一的标识符(id)，一张图像(image)，一段描述(description)，以及一个包含角色(role)和内容(content)的对话列表(conversations)。数据集分为训练集(train)，样本总数为136,654个。

This dataset is a multimodal dataset containing both images and text information. Specifically, each sample includes a unique identifier (id), an image (image), a description (description), and a list of conversations including role and content. The dataset is split into a training set (train) with a total of 136,654 samples.