saaduddinM/OXE_utaustin_mutex_embeddings

这是一个基于LeRobot（语言表）的嵌入仅限版本的数据集，它包含了由DINOv3和SigLIP2图像特征以及EmbeddingGemma任务文本特征组成的固定长度图像嵌入和文本嵌入。这个数据集是原始IPEC-COMMUNITY/utaustin_mutex_lerobot数据集的一个再编码变体，其中原始视频已被图像嵌入替换，并且任务字符串被文本嵌入增强。所有索引、划分和语义都与源数据集保持一致，同时存储和I/O变得更轻量。为了便于上传/下载和从Hub流式传输，小的每个剧集的Parquet文件被合并成大的Parquet碎片。数据集包含1500个剧集，361,883帧，50个唯一任务，分为64个Parquet碎片。

This is an embedding-only release of Language Table (LeRobot), consisting of fixed-length image embeddings and text embeddings composed of DINOv3 and SigLIP2 image features and EmbeddingGemma task-text features. It is a re-encoded variant of the original IPEC-COMMUNITY/utaustin_mutex_lerobot dataset, where raw videos are replaced by image embeddings and task strings are augmented with text embeddings. All indices, splits, and semantics remain consistent with the source dataset while storage and I/O are substantially lighter. The dataset includes 1,500 episodes, 361,883 frames, and 50 unique tasks, divided into 64 Parquet shards.