metadata
dataset_info:
features:
- name: id
dtype: string
- name: task_name
dtype: string
- name: task_description
dtype: string
- name: global_media
sequence:
image:
decode: false
- name: example_text
dtype: string
- name: example_media
sequence:
image:
decode: false
- name: query_text
dtype: string
- name: query_media
sequence:
image:
decode: false
- name: answer
dtype: string
- name: eval_context
dtype: string
splits:
- name: single_image
num_bytes: 2799903936
num_examples: 4108
- name: core
num_bytes: 35226745585
num_examples: 6531
- name: open
num_bytes: 11195891410
num_examples: 1158
download_size: 33312758764
dataset_size: 49222540931
configs:
- config_name: default
data_files:
- split: single_image
path: data/single_image-*
- split: core
path: data/core-*
- split: open
path: data/open-*