metadata
dataset_info:
features:
- name: _id
dtype: string
- name: type
dtype: string
- name: question
dtype: string
- name: answer
dtype: string
- name: image
dtype: image
- name: style
dtype: string
- name: context
struct:
- name: title
list: string
- name: content
list:
list: string
- name: num_contexts
dtype: int32
- name: supporting_facts
struct:
- name: title
list: string
- name: sent_id
list: int32
- name: evidences
struct:
- name: fact
list: string
- name: relation
list: string
- name: entity
list: string
splits:
- name: train
num_bytes: 324232745
num_examples: 1000
download_size: 308116316
dataset_size: 324232745
configs:
- config_name: default
data_files:
- split: train
path: data/train-*