JessicaE/OpenSeeSimE-Structural-Mini

OpenSeeSimE-Structural-Mini是OpenSeeSimE-Structural的一个分层1%子集，用于以降低的计算成本评估视觉语言模型，同时保持模拟类型、问题类型、媒体类型和问题ID的联合分布。数据集包含1,120行数据，来源于五个源类别：Beams、Dog Bone、Hip Implant、Pressure Vessel和Wall Bracket。数据集的组成包括按source_file、media_type和(source_file, question_type)的分布情况。数据集的特征模式与父数据集相同，包括文件名、源文件、问题文本、问题类型、问题ID、答案、答案选项、正确答案索引、图像或视频等。数据集的预期用途包括基准评估、评估管道的烟雾测试以及存储或带宽受限的比较研究。

OpenSeeSimE-Structural-Mini is a stratified 1% subset of OpenSeeSimE-Structural for evaluating vision-language models at a reduced compute footprint while preserving the joint distribution of simulation type, question type, media type, and question id. The dataset contains 1,120 rows sourced from five source classes: Beams, Dog Bone, Hip Implant, Pressure Vessel, and Wall Bracket. The composition of the dataset includes distributions by source_file, media_type, and (source_file, question_type). The feature schema is identical to the parent dataset, including file_name, source_file, question text, question type, question ID, answer, answer choices, correct choice index, image or video, etc. The intended use includes benchmark evaluation, smoke-testing of evaluation pipelines, and comparative studies where storage or bandwidth constraints matter.