metadata
dataset_info:
features:
- name: observation.images.image_1
dtype: video_frame
- name: observation.images.image_2
dtype: video_frame
- name: observation.images.image_3
dtype: video_frame
- name: language_instruction
dtype: string
- name: observation.state
sequence: float32
length: 7
- 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
- name: tokenised_language_instructions
sequence: int64
splits:
- name: train
num_bytes: 6794716
num_examples: 9687
download_size: 862540
dataset_size: 6794716
configs:
- config_name: default
data_files:
- split: train
path: data/train-*