metadata
dataset_info:
features:
- name: gem_id
dtype: string
- name: id
dtype: string
- name: title
dtype: string
- name: context
dtype: string
- name: question
dtype: string
- name: target
dtype: string
- name: references
list: string
- name: answers
sequence:
- name: text
dtype: string
- name: answer_start
dtype: int32
splits:
- name: train
num_bytes: 4208410.938426248
num_examples: 4000
- name: validation
num_bytes: 420841.0938426248
num_examples: 400
download_size: 2747943
dataset_size: 4629252.0322688725
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*