metadata
dataset_info:
features:
- name: problem
dtype: string
- name: answer
dtype: string
- name: conversations
list:
- name: from
dtype: string
- name: value
dtype: string
- name: images
list: image
- name: _qid
dtype: int32
splits:
- name: train
num_bytes: 3368835060
num_examples: 210000
- name: test
num_bytes: 1443795618
num_examples: 90000
download_size: 4787431789
dataset_size: 4812630678
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*