metadata
dataset_info:
features:
- name: images
dtype: image
- name: sentences
list: string
- name: cocoid
dtype: string
- name: paraphrased_captions
list: string
- name: true_triples_list
list:
list:
list: string
- name: object_bbox_map
dtype: string
- name: all_img_objects
list: string
- name: cap_triples
list:
list:
list: string
splits:
- name: train
num_bytes: 18785043099
num_examples: 113287
download_size: 18567634576
dataset_size: 18785043099
configs:
- config_name: default
data_files:
- split: train
path: data/train-*