metadata
dataset_info:
features:
- name: observation
sequence:
sequence:
sequence: float32
- name: action
sequence: int64
- name: reward
sequence: float32
- name: done
sequence: bool
splits:
- name: train
num_bytes: 176769250
num_examples: 1162000
download_size: 30970817
dataset_size: 176769250
configs:
- config_name: default
data_files:
- split: train
path: data/train-*