mateoguaman/language_table_blocktoabsolute_oracle_sim

该数据集是通过[LeRobot](https://github.com/huggingface/lerobot)创建的，是Google Research的[Language Table](https://github.com/google-research/language-table)项目中[language_table_blocktoabsolute_oracle_sim](https://www.tensorflow.org/datasets/catalog/language_table_blocktoabsolute_oracle_sim)数据集的LeRobot v3.0转换版本。原始RLDS数据从`gs://gresearch/robotics/language_table_blocktoabsolute_oracle_sim/0.0.1/`流式传输，转换为LeRobot v2.0格式，保留了所有字段（动作、状态、效应器目标、奖励、完成、指令），然后无损视频拼接为v3.0格式。数据集包含200,000个episodes，15,666,385帧，992个任务，视频分辨率为360x640，帧率为10fps。数据集结构包括观察图像、状态、效应器目标、动作、奖励、完成标志等特征。

This dataset was created using [LeRobot](https://github.com/huggingface/lerobot). It is a LeRobot v3.0 conversion of the [language_table_blocktoabsolute_oracle_sim](https://www.tensorflow.org/datasets/catalog/language_table_blocktoabsolute_oracle_sim) dataset from the [Language Table](https://github.com/google-research/language-table) project (Google Research). The original RLDS data was streamed from `gs://gresearch/robotics/language_table_blocktoabsolute_oracle_sim/0.0.1/`, converted to LeRobot v2.0 format preserving all fields (action, state, effector target, reward, done, instruction), then restructured to v3.0 with lossless video concatenation. The dataset contains 200,000 episodes, 15,666,385 frames, and 992 tasks, with video resolution of 360x640 and frame rate of 10fps. The dataset structure includes features such as observation images, state, effector target, action, reward, done flag, etc.