saaduddinM/OXE_bc_z_embeddings

这是一个基于LeRobot的语言表嵌入-only版本的数据集，包含了DINOv3和SigLIP2的图像特征以及EmbeddingGemma的任务文本特征。数据集包含了39350个视频，总共5471693帧，每个视频对应一个任务，共有104个独特任务。数据集以Apache-2.0许可证发布。This is an embedding-only version of the LeRobot Language Table dataset, containing DINOv3 and SigLIP2 image features as well as EmbeddingGemma task-text features. The dataset includes 39,350 videos, totaling 5,471,693 frames, with each video corresponding to a task and a total of 104 unique tasks. The dataset is released under the Apache-2.0 license.