metadata
dataset_info:
features:
- name: chosen
dtype: string
- name: img_path
dtype: string
- name: prompt
dtype: string
- name: rejected
dtype: string
- name: question
dtype: string
- name: image
dtype: image
- name: generated_context
dtype: string
- name: right_context
dtype: string
- name: wrong_context
dtype: string
splits:
- name: train
num_bytes: 352458105.125
num_examples: 6135
download_size: 336574426
dataset_size: 352458105.125
configs:
- config_name: default
data_files:
- split: train
path: data/train-*