metadata
dataset_info:
features:
- name: source
dtype: string
- name: input_ids
list: int32
- name: attention_mask
list: int8
- name: labels
list: int64
- name: length
dtype: int64
splits:
- name: train
num_bytes: 3550065181
num_examples: 129214
download_size: 4363704412
dataset_size: 3550065181
configs:
- config_name: default
data_files:
- split: train
path: data/train-*