metadata
dataset_info:
features:
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: labels
sequence: int64
splits:
- name: train
num_bytes: 8575000000
num_examples: 5000000
- name: test
num_bytes: 343000000
num_examples: 200000
download_size: 993317642
dataset_size: 8918000000
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*