metadata
dataset_info:
features:
- name: data_source
dtype: string
- name: prompt
list:
- name: content
dtype: string
- name: role
dtype: string
- name: images
list: image
- name: reward_model
struct:
- name: ground_truth
dtype: string
- name: style
dtype: string
- name: extra_info
struct:
- name: gt_answer
dtype: string
- name: id
dtype: string
- name: mc_options
dtype: 'null'
- name: original_question
dtype: string
splits:
- name: train
num_bytes: 4547551998
num_examples: 7576
download_size: 4411835098
dataset_size: 4547551998
configs:
- config_name: default
data_files:
- split: train
path: data/train-*