mlfoundations-cua-dev/agentnet-gimp-trajectories

这是一个包含多个字段的数据集，其中包括数据集名称、领域、任务ID、指令、消息（包括内容和角色）、图片、步骤动作、原始动作、步骤ID和步骤数量。数据集专为训练而设计，包含训练集分割，并提供了详细的字节数和示例数信息。

This is a dataset with multiple fields, including dataset name, domain, task ID, instruction, messages (including content and role), images, step actions, raw actions, step IDs, and number of steps. The dataset is designed for training, containing a training split and providing detailed information on byte size and number of examples.