metadata
dataset_info:
features:
- name: input_text
dtype: string
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 8766686434
num_examples: 2050000
- name: validation
num_bytes: 60369417
num_examples: 14078
download_size: 2251755758
dataset_size: 8827055851
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*