fmutairi/medium_satellite_llava_instruct_mix

该数据集是一个多模态数据集，包含图像和文本对话数据。特征包括图像字段和消息字段，其中消息字段是一个列表，每个消息包含内容（包括索引、文本和类型）和角色信息。数据集分为训练集和测试集，训练集有31064个样本，测试集有7766个样本，总下载大小约为5.34GB，数据集大小约为5.32GB。数据用于支持图像与文本交互的任务，如视觉对话或指令跟随。

This dataset is a multimodal dataset containing image and text conversation data. Features include an image field and a messages field, where the messages field is a list with each message containing content (including index, text, and type) and role information. The dataset is split into train and test sets, with 31064 examples in the train set and 7766 examples in the test set. The total download size is approximately 5.34GB, and the dataset size is approximately 5.32GB. It is designed for tasks involving image and text interaction, such as visual dialogue or instruction following.