metadata
dataset_info:
features:
- name: episode_id
dtype: int64
- name: frames
dtype: binary
- name: actions
dtype: int64
- name: health
dtype: int64
- name: step_ids
dtype: int64
splits:
- name: train
num_bytes: 46500834009
num_examples: 2506064
download_size: 46272271549
dataset_size: 46500834009
configs:
- config_name: default
data_files:
- split: train
path: data/train-*