README.md

metadata

license: cc-by-4.0
language:
  - en
pretty_name: PhotoChat++
size_categories:
  - n<1K
multilinguality:
  - monolingual
annotation_creators:
  - crowd-sourced
tags:
  - multi-modal dialogue
source_datasets:
  - PhotoChat
task_ids:
  - conversational
task_categories:
  - text-to-image
  - image-to-text
splits:
  - name: train
    num_examples: 968
dataset_size: 968

Dataset Card for PhotoChat++

🚨 Disclaimer: All models and datasets are intended for research purposes only.

Dataset Description

• Repository: Code
• Paper: Large Language Models can Share Images, Too!
• Point of Contact: Young-Jun Lee

Dataset Summary

PhotoChat++ is a publicly available multi-modal dialogue dataset, an extended version of PhotoChat. PhotoChat++ contains six intent labels, a triggering sentence, an image description, and salient information (e.g., “words” or “phrases”) to invoke the image-sharing behavior. The purpose of this dataset is to thoroughly assess the image-sharing capability of LLMs based on humans' internal operating systems.

Languages

English

Dataset Structure

field	type	description
dialogue_id	str	the identifier for the dialogue, containing the original dialogue identifier from PhotoChat
dialogue	list of dict	the dialogue, where each dict entry includes {message, share_photo, user_id} (from PhotoChat)
photo_id	str	the identifier for the photo (from PhotoChat)
photo_url	str	the URL for the photo (from PhotoChat)
photo_description	str	the description of the photo (from PhotoChat)
intents	list of str	all intents annotated from crowd-sourcing
trigger_sentences	list of str	all triggering sentences that invoke the image-sharing behavior, annotated from crowd-sourcing
image_descriptions	list of str	all image descriptions annotated from crowd-sourcing, which are different from the photo_description field
salient_information	list of str	all salient information (e.g., word or phrase) annotated from crowd-sourcing

Dataset Creation

We create PhotoChat++ dataset via crowd-sourcing.

Further Details, Social Impacts, and Limitations

Please refer to our paper.

Limitations

Please refer to the Limitation section in our paper.

Recommendations

Since PhotoChat++ is constructed via crowd-sourcing based on dialogues from the PhotoChat dataset, which is licensed under the CC BY 4.0 International license, PhotoChat++ is shared under the same CC BY 4.0 International license. Therefore, following this license, it is possible to use the PhotoChat++ dataset for commercial purposes. However, we strongly recommend using our dataset for academic and research purposes.

Acknowledgements

This work was supported by a grant from the KAIST-KT joint research project through AI Tech Lab, Institute of Convergence Technology, funded by KT [Project No. G01230605, Development of Task-oriented Persona-based Dialogue Generation Combining Multi-modal Interaction and Knowledge Modeling].

Citation

Please cite our work if you find the resources in this repository useful:

@article{lee2023large,
  title={Large Language Models can Share Images, Too!},
  author={Lee, Young-Jun and Lee, Dokyong and Sung, Joo Won and Hyeon, Jonghwan and Choi, Ho-Jin},
  journal={arXiv preprint arXiv:2310.14804},
  year={2023}
}