AasherH/vlm2vla-data
收藏资源简介：
该数据集是一个多模态数据集，包含1295640个训练示例，总大小约为13.2 GB。数据集结构包括messages和images两个主要特征：messages字段包含content（由index、text和type组成）和role，用于表示对话或消息内容；images字段为二进制列表，可能存储图像数据。数据集适用于多模态任务，如结合文本和图像的AI模型训练。数据以train分割形式组织，可通过指定路径访问。
This dataset is a multimodal dataset comprising 1,295,640 training examples, with a total size of approximately 13.2 GB. The dataset structure includes two main features: messages and images. The messages field consists of content (with subfields index, text, and type) and role, representing dialogue or message content; the images field is a list of binary data, likely storing image data. The dataset is suitable for multimodal tasks, such as training AI models that combine text and images. It is organized into a train split and can be accessed via a specified path.
