metadata
dataset_info:
features:
- name: image
dtype: string
- name: width
dtype: int64
- name: height
dtype: int64
- name: conversations
list:
- name: from
dtype: string
- name: value
dtype: string
- name: caption
dtype: string
- name: translated_caption_mk
dtype: string
splits:
- name: train
num_bytes: 219125173
num_examples: 26277
download_size: 111409701
dataset_size: 219125173
configs:
- config_name: default
data_files:
- split: train
path: data/train-*