cs5242-hateful-memes/hateful-memes-data

Hateful Memes (CS5242 submission mirror)

Mirror of the Facebook Hateful Memes Challenge dataset (Kiela et al., 2020) used for reproducibility of our CS5242 (NUS) submission.

Contents

• —img/ — 10,000 PNG images of memes
• —train.jsonl (8,500), dev_seen.jsonl (500), dev_unseen.jsonl (540), test_seen.jsonl (1,000), test_unseen.jsonl (2,000)

Provenance

This mirror merges two existing mirrors of the original Meta release:

• —Label files and most images from neuralcatcher/hateful_memes.
• —~1,941 images missing from that mirror were recovered from the Kaggle mirror parthplc/facebook-hateful-meme-dataset (matching filenames).

The label files (.jsonl) are byte-identical to the neuralcatcher snapshot at commit d201c488dc7024623d1ecbcc987b3f132c4c2e12.

License and use

The dataset is released under the Meta Hateful Memes license (see LICENSE.txt). By downloading, you agree to use it only for non-commercial research and not to attempt to identify any individuals.

Citation

@inproceedings{Kiela:2020hatefulmemes,
 author = {Kiela, Douwe and Firooz, Hamed and Mohan, Aravind and Goswami, Vedanuj and Singh, Amanpreet and Ringshia, Pratik and Testuggine, Davide},
 booktitle = {NeurIPS},
 title = {The Hateful Memes Challenge: Detecting Hate Speech in Multimodal Memes},
 year = {2020}
}