multimodal-reasoning-lab/Visual-Jigsaw

该数据集包含了问题文本（Question）、文本推理轨迹（Text Reasoning Trace）、最终答案（Final Answer）以及多张图像，包括问题图像（problem_image_1）和推理图像（reasoning_image_1至reasoning_image_4）。数据集分为训练集，共有21,899个示例，总大小约为12.58GB。数据集适用于需要文本理解和图像处理能力的任务。

The dataset includes question text (Question), text reasoning trace (Text Reasoning Trace), final answer (Final Answer), and multiple images, including problem image (problem_image_1) and reasoning images (reasoning_image_1 to reasoning_image_4). The dataset is split into a training set with a total of 21,899 examples and a total size of approximately 12.58GB. The dataset is suitable for tasks that require text comprehension and image processing capabilities.