metadata
dataset_info:
features:
- name: id
dtype: string
- name: context
dtype: string
- name: title
dtype: string
- name: labels
list:
- name: answer
dtype: string
- name: answer_start
dtype: int64
- name: id
dtype: string
- name: is_impossible
dtype: bool
- name: original_answer
dtype: string
- name: question
dtype: string
- name: metadata
struct:
- name: doc_id
dtype: string
- name: doc_title
dtype: string
- name: doc_source
dtype: string
- name: doc_published
dtype: int32
- name: created
dtype: string
- name: class
dtype: string
- name: code
dtype: string
splits:
- name: train
num_bytes: 631574568
num_examples: 160000
- name: validation
num_bytes: 80786480
num_examples: 20000
download_size: 337638665
dataset_size: 712361048
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*