Datasets:
metadata
size_categories:
- 1M<n<10M
task_categories:
- visual-question-answering
pretty_name: UltraPreview
configs:
- config_name: cot
data_files:
- split: train
path: cot/*.parquet
- config_name: obelics
data_files:
- split: train
path: obelics/*.parquet
- config_name: xiaohongshu
data_files:
- split: train
path: xiaohongshu/*.parquet
- config_name: zhihu
data_files:
- split: train
path: zhihu/*.parquet
dataset_info:
- config_name: cot
features:
- name: images
sequence: image
- name: image_ids
sequence: string
- name: texts
list:
- name: from
dtype: string
- name: value
dtype: string
- name: source
dtype: string
- config_name: obelics
features:
- name: images
sequence: image
- name: image_ids
sequence: string
- name: texts
list:
- name: from
dtype: string
- name: value
dtype: string
- name: source
dtype: string
- config_name: xiaohongshu
features:
- name: images
sequence: image
- name: image_ids
sequence: string
- name: texts
list:
- name: from
dtype: string
- name: value
dtype: string
- name: source
dtype: string
- config_name: zhihu
features:
- name: images
sequence: image
- name: image_ids
sequence: string
- name: texts
list:
- name: from
dtype: string
- name: value
dtype: string
- name: source
dtype: string
language:
- en
- zh