metadata
language:
- en
dataset_info:
features:
- name: image_id
dtype: int64
- name: caption
dtype: string
- name: 'n'
dtype: int64
- name: n_objects
dtype: int64
- name: n_relations
dtype: int64
- name: n_attributes
dtype: int64
- name: n_connected_components
dtype: int64
- name: hard_negs
dtype: string
- name: x
dtype: float64
- name: 'y'
dtype: float64
- name: width
dtype: float64
- name: height
dtype: float64
- name: Complexity
dtype: string
- name: Negative_type
dtype: string
- name: images
dtype: image
- name: text
dtype: string
- name: hard_texts
sequence: string
splits:
- name: train
num_bytes: 705453788.875
num_examples: 2001
download_size: 681524192
dataset_size: 705453788.875
configs:
- config_name: default
data_files:
- split: train
path: data/train-*