bdharva/eval_act_so100_benchy_test

这是一个用于机器人任务的 dataset，包含10个剧集，共4301帧，1个任务，20个视频，1个片段，每个片段包含1000帧。数据集的帧率为30fps，支持训练集分割。数据以 Parquet 格式存储，并包含有关机器人动作和状态的多种特征，如主要肩膀、肘部、手腕和夹子的角度。此外，数据集还包含低分辨率和高分辨率的视频流，以及时间戳和帧索引等元数据。

This is a dataset for robotics tasks, containing 10 episodes, 4301 frames, 1 task, 20 videos, 1 chunk with 1000 frames each. The dataset has a frame rate of 30fps and supports train set splitting. The data is stored in Parquet format and includes various features related to robot actions and states, such as angles for the main shoulder, elbow, wrist, and gripper. Additionally, the dataset provides both low-resolution and high-resolution video streams, as well as metadata like timestamp and frame index.