metadata
dataset_info:
features:
- name: question_id
dtype: int64
- name: image
dtype: image
- name: mask
dtype: image
- name: object
dtype: string
- name: verb
dtype: string
- name: noun
dtype: string
- name: instruction
dtype: string
- name: point_2d
list:
list: int64
- name: point_2d_norm
list:
list: int64
- name: width
dtype: int64
- name: height
dtype: int64
splits:
- name: train
num_bytes: 216785664
num_examples: 1543
download_size: 214228821
dataset_size: 216785664
configs:
- config_name: default
data_files:
- split: train
path: data/train-*