metadata
dataset_info:
features:
- name: observation.images.image
dtype: video_frame
- name: observation.images.wrist_image
dtype: video_frame
- name: language_instruction
dtype: string
- name: observation.state
sequence: float32
length: 8
- name: action
sequence: float32
length: 7
- name: timestamp
dtype: float32
- name: episode_index
dtype: int64
- name: frame_index
dtype: int64
- name: next.reward
dtype: float32
- name: next.done
dtype: bool
- name: index
dtype: int64
splits:
- name: train
num_bytes: 105226148
num_examples: 358234
download_size: 34959271
dataset_size: 105226148
configs:
- config_name: default
data_files:
- split: train
path: data/train-*