metadata
dataset_info:
features:
- name: source_id
dtype: string
- name: image
dtype: image
- name: question
dtype: string
- name: answer
dtype: string
- name: sft_prompt
dtype: string
- name: assistant_response
dtype: string
splits:
- name: train
num_bytes: 24089862
num_examples: 451
download_size: 22488113
dataset_size: 24089862
configs:
- config_name: default
data_files:
- split: train
path: data/train-*