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: 4632508808
num_examples: 250938
download_size: 4609632689
dataset_size: 4632508808
configs:
- config_name: default
data_files:
- split: train
path: data/train-*