metadata
dataset_info:
features:
- name: input_ids
list: int32
- name: labels
list: int64
- name: num_mask_tokens
dtype: int64
- name: original
list: int64
splits:
- name: train
num_bytes: 19474815000
num_examples: 2870669
- name: test
num_bytes: 198018600
num_examples: 28999
download_size: 3268994346
dataset_size: 19672833600
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*