metadata
dataset_info:
features:
- name: image_id
dtype: string
- name: image_url
dtype: string
- name: image
dtype: image
- name: options
list:
- name: label
dtype: string
- name: text
dtype: string
- name: answer
dtype: string
- name: prompt_1
dtype: string
- name: prompt_2
dtype: string
- name: prompt_3
dtype: string
- name: task
dtype: string
splits:
- name: deep_semantics
num_bytes: 51241466
num_examples: 1001
- name: description
num_bytes: 50878157
num_examples: 1001
- name: title
num_bytes: 47561935
num_examples: 1001
download_size: 141590822
dataset_size: 149681558
configs:
- config_name: default
data_files:
- split: deep_semantics
path: data/deep_semantics-*
- split: description
path: data/description-*
- split: title
path: data/title-*
citation_bibtex: |-
@article{yang2024can,
title={Can Large Multimodal Models Uncover Deep Semantics Behind Images?},
author={Yang, Yixin and Li, Zheng and Dong, Qingxiu and Xia, Heming and Sui, Zhifang},
journal={arXiv preprint arXiv:2402.11281},
year={2024}
}
Citation
If you use this dataset, please cite the following paper:
@article{yang2024can,
title={Can Large Multimodal Models Uncover Deep Semantics Behind Images?},
author={Yang, Yixin and Li, Zheng and Dong, Qingxiu and Xia, Heming and Sui, Zhifang},
journal={arXiv preprint arXiv:2402.11281},
year={2024}
}
This is just an huggingface version of the exact same dataset for easier integration to some testing pipelines.