metadata

model-index:
  - name: twitter-roberta-base-hate-latest
    results: []
pipeline_tag: text-classification
language:
  - en

cardiffnlp/twitter-roberta-base-hate-latest

This model is a fine-tuned version of cardiffnlp/twitter-roberta-base-2022-154m for binary hate-speech classification. A combination of 13 different hate-speech datasets in the English language were used to fine-tune the model.

Dataset	Accuracy	Macro-F1	Weighted-F1
hatEval, SemEval-2019 Task 5: Multilingual Detection of Hate Speech Against Immigrants and Women in Twitter	0.5831	0.5646	0.548
ucberkeley-dlab/measuring-hate-speech	0.9273	0.9193	0.928
Detecting East Asian Prejudice on Social Media	0.9231	0.6623	0.9428
Call me sexist, but	0.9686	0.9203	0.9696
Predicting the Type and Target of Offensive Posts in Social Media	0.9164	0.6847	0.9098
HateXplain	0.8653	0.845	0.8662
Large Scale Crowdsourcing and Characterization of Twitter Abusive BehaviorLarge Scale Crowdsourcing and Characterization of Twitter Abusive Behavior	0.7801	0.7446	0.7614
Multilingual and Multi-Aspect Hate Speech Analysis	0.9944	0.4986	0.9972
Hate speech and offensive content identification in indo-european languages	0.8779	0.6904	0.8706
Are You a Racist or Am I Seeing Things?	0.921	0.8935	0.9216
Automated Hate Speech Detection	0.9423	0.9249	0.9429
Hate Towards the Political Opponent	0.8783	0.6595	0.8788
Hateful Symbols or Hateful People?	0.8187	0.7833	0.8323
Overall	0.8766	0.7531	0.8745

Usage

Install tweetnlp via pip.

pip install tweetnlp

Load the model in python.

import tweetnlp
model = tweetnlp.Classifier("cardiffnlp/twitter-roberta-base-hate-latest")
model.predict('I love everybody :)')
>> {'label': 'NOT-HATE'}

Reference paper - Model based on:

@inproceedings{antypas-camacho-collados-2023-robust,
    title = "Robust Hate Speech Detection in Social Media: A Cross-Dataset Empirical Evaluation",
    author = "Antypas, Dimosthenis  and
      Camacho-Collados, Jose",
    booktitle = "The 7th Workshop on Online Abuse and Harms (WOAH)",
    month = jul,
    year = "2023",
    address = "Toronto, Canada",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2023.woah-1.25",
    pages = "231--242",
    abstract = "The automatic detection of hate speech online is an active research area in NLP. Most of the studies to date are based on social media datasets that contribute to the creation of hate speech detection models trained on them. However, data creation processes contain their own biases, and models inherently learn from these dataset-specific biases. In this paper, we perform a large-scale cross-dataset comparison where we fine-tune language models on different hate speech detection datasets. This analysis shows how some datasets are more generalizable than others when used as training data. Crucially, our experiments show how combining hate speech detection datasets can contribute to the development of robust hate speech detection models. This robustness holds even when controlling by data size and compared with the best individual datasets.",
}