metadata
dataset_info:
features:
- name: problem
dtype: string
- name: images
sequence: image
- name: answer
dtype: string
- name: data_source
dtype: string
splits:
- name: train
num_bytes: null
num_examples: 998
configs:
- config_name: default
data_files:
- split: train
path: data/train*