metadata
dataset_info:
features:
- name: img_name
dtype: string
- name: location
dtype: string
- name: modality
dtype: string
- name: base_type
dtype: string
- name: answer_type
dtype: string
- name: qid
dtype: int64
- name: content_type
dtype: string
- name: triple
sequence: string
- name: img_id
dtype: int64
- name: q_lang
dtype: string
- name: question
dtype: string
- name: answer
dtype: string
splits:
- name: train
num_bytes: 185630
num_examples: 1000
download_size: 33560
dataset_size: 185630
configs:
- config_name: default
data_files:
- split: train
path: data/train-*