metadata
language:
- en
license: cc-by-sa-3.0
pretty_name: Natural Questions Parsed
dataset_info:
features:
- name: question
dtype: string
- name: long_answer
dtype: string
- name: short_answers
sequence:
sequence: string
- name: yes_no_answer
sequence: int64
- name: candidates
sequence: string
- name: document
dtype: string
- name: id
dtype: string
splits:
- name: train
num_bytes: 21949289652
num_examples: 307373
- name: validation
num_bytes: 533713369
num_examples: 7830
download_size: 11330809696
dataset_size: 22483003021
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*