metadata
dataset_info:
features:
- name: text
dtype: string
- name: corrupted_text
dtype: string
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: labels
sequence: int64
splits:
- name: split_1
num_bytes: 13513091981
num_examples: 500000
- name: split_2
num_bytes: 13513166739
num_examples: 500000
download_size: 324054720
dataset_size: 27026258720
configs:
- config_name: default
data_files:
- split: split_1
path: data/split_1-*
- split: split_2
path: data/split_2-*