metadata
dataset_info:
features:
- name: data
sequence: float32
splits:
- name: train
num_bytes: 19866868920
num_examples: 6458670
download_size: 23885722251
dataset_size: 19866868920
configs:
- config_name: default
data_files:
- split: train
path: data/train-*