CurHarsh/sft_robo_rlbench_state

这个数据集是一个包含文本消息和图像的多模态数据集。文本消息由内容（content）和角色（role）两部分组成，而图像则包括来自不同摄像头的视角，如正面、左肩上方、右肩上方、头顶和手腕。数据集的总大小约为7.43GB，其中训练集包含了14,223个示例。

This dataset is a multimodal dataset containing text messages and images. The text messages consist of content and role, while the images include perspectives from different cameras such as front, left shoulder, right shoulder, overhead, and wrist. The total size of the dataset is approximately 7.43GB, with the training set containing 14,223 examples.