metadata
dataset_info:
features:
- name: id
dtype: int64
- name: num_objects
dtype: int64
- name: objects
sequence: string
- name: relations
sequence: string
- name: chain_included_llm
dtype: bool
- name: file_name
dtype: string
- name: caption
dtype: string
- name: caption_source
dtype: string
- name: image
dtype: image
splits:
- name: train
num_bytes: 1804110488.121
num_examples: 38393
download_size: 1733026747
dataset_size: 1804110488.121
configs:
- config_name: default
data_files:
- split: train
path: data/train-*