metadata
dataset_info:
features:
- name: input_mask
sequence: int64
- name: input_ids
sequence: int32
- name: actual_token_len
sequence: int64
splits:
- name: train
num_bytes: 247615553800
num_examples: 80082650
download_size: 5329742623
dataset_size: 247615553800
configs:
- config_name: default
data_files:
- split: train
path: data/train-*