LehongWu/opsd-v3_8-collect_6tasks_013789_v3_1_0416-gemini3flash_medium-repeat8_0416_all500trajs

该数据集包含36,677个训练示例，主要用于多模态任务，涉及图像和文本数据。特征包括图像列表、学生提示、教师提示、完成文本以及额外信息（如目标、任务特定提示、先前指令、专家推理、专家指令、轨迹键、步骤ID和消息行索引）。数据集可能用于教育或对话生成场景，但具体描述未在README中提供。

This dataset consists of 36,677 training examples designed for multimodal tasks, involving both image and text data. Features include a list of images, student prompts, teacher prompts, completion text, and extra information (such as goal, task-specific prompt, previous instruction, expert reasoning, expert instruction, trajectory key, step ID, and message row index). It may be used in educational or dialogue generation contexts, but a detailed description is not provided in the README.