metadata
dataset_info:
features:
- name: id
dtype: string
- name: category
dtype: string
- name: level
dtype: string
- name: instruction
dtype: string
- name: image
dtype: image
- name: total_id
dtype: string
splits:
- name: train
num_bytes: 62135591532.425
num_examples: 101115
download_size: 4809827770
dataset_size: 62135591532.425
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
이미지, 쿼리 (instruction, image)
image는 PIL 이미지객체
Dataset({
features: ['id', 'category', 'level', 'instruction', 'image', 'total_id'],
num_rows: 101115
})