metadata
license: cc-by-4.0
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
dataset_info:
features:
- name: image
dtype: image
- name: question
dtype: string
- name: A
dtype: string
- name: B
dtype: string
- name: C
dtype: string
- name: D
dtype: string
- name: answer
dtype: string
- name: benchmark
dtype: string
- name: image_id
dtype: string
- name: spurious_attribute
dtype: string
- name: category
dtype: string
- name: description
dtype: string
- name: spurious_group_1
dtype: image
- name: spurious_group_2
dtype: image
- name: spurious_group_3
dtype: image
- name: spurious_group_4
dtype: image
- name: spurious_group_5
dtype: image
- name: spurious_group_6
dtype: image
- name: spurious_group_7
dtype: image
- name: spurious_group_8
dtype: image
- name: spurious_group_9
dtype: image
- name: spurious_group_10
dtype: image
- name: core_group_1
dtype: image
- name: core_group_2
dtype: image
- name: core_group_3
dtype: image
- name: core_group_4
dtype: image
- name: core_group_5
dtype: image
- name: core_group_6
dtype: image
- name: core_group_7
dtype: image
- name: core_group_8
dtype: image
- name: core_group_9
dtype: image
- name: core_group_10
dtype: image
splits:
- name: train
num_bytes: 4296771456
num_examples: 124
download_size: 4269644455
dataset_size: 4296771456
We present SpuriVerse, a multimodal benchmark comprising 124 distinct spurious correlations extracted from real-world datasets, each containing 11 examples --- 1 from the original dataset, referred to as the anchor and 10 synthetic examples, referred as the spurious group --- for a total of 1364 instances.