metadata
dataset_info:
features:
- name: dataset_id
dtype: string
- name: image_id
dtype: string
- name: annotator_id
dtype: int64
- name: caption
dtype: string
- name: captions
sequence: string
- name: neg_captions
sequence: string
splits:
- name: train
num_bytes: 734255949
num_examples: 438799
download_size: 194810303
dataset_size: 734255949
configs:
- config_name: default
data_files:
- split: train
path: data/train-*