metadata
dataset_info:
features:
- name: text
dtype: string
- name: topic
dtype: string
- name: title
dtype: string
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: special_tokens_mask
sequence: int8
- name: chunk
dtype: int64
splits:
- name: train
num_bytes: 2630020554
num_examples: 35620
download_size: 1018008571
dataset_size: 2630020554
configs:
- config_name: default
data_files:
- split: train
path: data/train-*