cardiffnlp/twitter-roberta-base-hate-multiclass-latest

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

Classes available

{
  "sexism": 0,
  "racism": 1,
  "disability": 2,
  "sexual_orientation": 3,
  "religion": 4,
  "other": 5,
  "not_hate":6
}

Following metrics are achieved

• Accuracy: 0.9419
• Macro-F1: 0.5752
• Weighted-F1: 0.9390

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('Women are trash 2.')
>> {'label': 'sexism'}
model.predict('@user dear mongoloid respect sentiments & belief refrain totalitarianism. @user')
>> {'label': 'disability'}

Model based on:

@misc{antypas2023robust,
      title={Robust Hate Speech Detection in Social Media: A Cross-Dataset Empirical Evaluation}, 
      author={Dimosthenis Antypas and Jose Camacho-Collados},
      year={2023},
      eprint={2307.01680},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}