metadata
dataset_info:
features:
- name: title
dtype: string
- name: news_category
dtype: string
- name: source
dtype: string
- name: context
dtype: string
- name: question
dtype: string
- name: question_type
dtype: int64
- name: is_impossible
dtype: bool
- name: answer_text
dtype: string
- name: answer_start
dtype: int64
- name: negative_samples
sequence: string
- name: search_result
sequence: string
- name: answer
dtype: string
- name: extracted_ref_numbers
sequence: int64
splits:
- name: train
num_bytes: 4900918
num_examples: 196
download_size: 2824532
dataset_size: 4900918
configs:
- config_name: default
data_files:
- split: train
path: data/train-*