metadata
dataset_info:
features:
- name: id
dtype: int64
- name: image
dtype: string
- name: question
dtype: string
- name: type
dtype: string
- name: answer
struct:
- name: truth
sequence: string
- name: hallu
sequence: string
splits:
- name: train
num_bytes: 1788943
num_examples: 15220
download_size: 220938
dataset_size: 1788943
configs:
- config_name: default
data_files:
- split: train
path: data/train-*