metadata
dataset_info:
features:
- name: image
dtype: image
- name: question
dtype: string
- name: answer
dtype: string
- name: fixations
sequence:
sequence: float64
splits:
- name: train
num_bytes: 495274238
num_examples: 698
download_size: 492834126
dataset_size: 495274238
configs:
- config_name: default
data_files:
- split: train
path: data/train-*