README.md

meme-cap

This is a repo for the paper: MemeCap: A Dataset for Captioning and Interpreting Memes

Data

Train+Val and Test

We used 10% of training data as validation data.

Images are available here: link

Model

We used OpenFlamingo-9B and MiniGPT4 models for our zero-shot and few-shot experiments.

Check out their code base for the model implementation.

Paper

If the code is helpful for your project, please cite our paper (Bibtex below).

@misc{hwang2023memecap,
      title={MemeCap: A Dataset for Captioning and Interpreting Memes}, 
      author={EunJeong Hwang and Vered Shwartz},
      year={2023},
      eprint={2305.13703},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}