metadata
dataset_info:
features:
- name: content
dtype: string
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
splits:
- name: train
num_bytes: 334990452695
num_examples: 21383832
download_size: 106884623961
dataset_size: 334990452695
configs:
- config_name: default
data_files:
- split: train
path: data/c/train-*