metadata
dataset_info:
features:
- name: image
dtype: image
- name: filepath
dtype: string
- name: sentids
list: int32
- name: filename
dtype: string
- name: imgid
dtype: int32
- name: split
dtype: string
- name: sentences_tokens
list:
list: string
- name: sentences_raw
list: string
- name: sentences_sentid
list: int32
- name: cocoid
dtype: int32
- name: th_sentences_raw
sequence: string
splits:
- name: train
num_bytes: 988476446.9130616
num_examples: 6000
- name: validation
num_bytes: 164040180.6
num_examples: 1000
- name: test
num_bytes: 165530786
num_examples: 1000
download_size: 1303655952
dataset_size: 1318047413.5130615
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*