InternSVG/SArena

SArena是一个统一的SVG任务数据集，支持多模态大型语言模型。它包括四个数据域：图标(icon)、插图(illustration)、化学(chemistry)和动画(animation)。针对这些数据域，它提供了多种任务类型，如图标的理解、编辑和生成任务，插图的生成任务，化学结构的SVG转换任务，以及动画的生成任务。所有数据样本都存储在标准化的.jsonl格式文件中，包含了结构化的输入-输出对。

SArena is a unified SVG task dataset that supports multimodal large language models. It includes four data domains: icon, illustration, chemistry, and animation. For these data domains, it provides various task types such as icon understanding, editing, and generation tasks, illustration generation tasks, SVG conversion tasks for chemical structures, and animation generation tasks. All data samples are stored in standardized .jsonl format files, containing structured input-output pairs.