Datasets:
metadata
dataset_info:
features:
- name: observation.state
sequence: float32
length: 14
- name: action
sequence: float32
length: 14
- name: episode_id
dtype: int64
- name: frame_id
dtype: int64
- name: timestamp
dtype: float32
- name: next.done
dtype: bool
- name: episode_data_id_from
dtype: int64
- name: episode_data_id_to
dtype: int64
- name: observation.images.top
dtype: image
- name: index
dtype: int64
splits:
- name: train
num_bytes: 738057394
num_examples: 25000
download_size: 728970132
dataset_size: 738057394
configs:
- config_name: default
data_files:
- split: train
path: data/train-*