cmudrc/OpenSeeSimE-Fluid

OpenSeeSimE-Fluid是一个大规模基准数据集，用于评估视觉语言模型在计算流体动力学（CFD）模拟解释任务上的表现。该数据集包含约98,000个问答对，涵盖了参数化变化的流体模拟，包括湍流、热传递和复杂流动模式。数据集旨在提供统计上稳健的评估，支持多种推理能力（如描述、推理、定位和关系理解）和不同问题格式（如二元分类、多项选择和空间定位）。所有数据均来自数值模拟结果，而非视觉解释，确保了标注的一致性和准确性。数据集仅包含测试集，用于评估而非训练。

OpenSeeSimE-Fluid is a large-scale benchmark dataset for evaluating vision-language models on computational fluid dynamics (CFD) simulation interpretation tasks. It contains approximately 98,000 question-answer pairs across parametrically-varied fluid simulations including turbulent flow, heat transfer, and complex flow patterns. The benchmark enables statistically robust evaluation of VLM performance on CFD visualizations, assessment across multiple reasoning capabilities (captioning, reasoning, grounding, relationship understanding), and evaluation using different question formats (binary classification, multiple-choice, spatial grounding). All ground truth is derived from numerical simulation results rather than visual interpretation, ensuring consistency and accuracy. The dataset includes only a test split, intended for evaluation rather than training.