Reubencf/multilingual-image-captions-text

该数据集是原始多语言图像标注集合的重新制作版本，移除了二进制图像数据以提高加载效率。它包含464行由google/gemma-4-31B-it模型生成的英语和多语言标注，涵盖七种语言。内容涉及城市景观、自然场景、室内设计和物品特写等多种视觉主题，适用于文本生成和视觉问答任务。

This dataset provides text-only multilingual image annotations derived from the original multilingual-image-annotations collection, with binary image data removed for efficient loading. It contains 464 rows of English and multilingual captions generated by the google/gemma-4-31B-it model across seven languages. The content covers diverse visual subjects including urban landscapes, nature scenes, interior designs, and object close-ups, suitable for text-generation and visual question answering tasks.