twitter-roberta-large-sensitive-multilabel

twitter-roberta-base-sensitive-binary

This is a RoBERTa-large model trained on 154M tweets until the end of December 2022 and finetuned for detecting sensitive content (multilabel classification) on the X-Sensitive dataset. The original Twitter-based RoBERTa model can be found here.

A sensitive content binary model can be found here.

Labels

"id2label": {
  "0": "conflictual",
  "1": "profanity",
  "2": "sex",
  "3": "drugs",
  "4": "selfharm",
  "5": "spam",
  "6": "not-sensitive"
}

Full classification example

from transformers import pipeline
    
pipe = pipeline(model='cardiffnlp/twitter-roberta-large-sensitive-multilabel')
text = "Call me today to earn some money mofos!"

pipe(text)

Output:

[[{'label': 'conflictual', 'score': 0.03700090944766998},
  {'label': 'profanity', 'score': 0.9770461916923523},
  {'label': 'sex', 'score': 0.01981434039771557},
  {'label': 'drugs', 'score': 0.017757439985871315},
  {'label': 'selfharm', 'score': 0.008804548531770706},
  {'label': 'spam', 'score': 0.07784222811460495},
  {'label': 'not-sensitive', 'score': 0.010364986956119537}]]

BibTeX entry and citation info

@article{antypas2024sensitive,
  title={Sensitive Content Classification in Social Media: A Holistic Resource and Evaluation},
  author={Antypas, Dimosthenis and Sen, Indira and Perez-Almendros, Carla and Camacho-Collados, Jose and Barbieri, Francesco},
  journal={arXiv preprint arXiv:2411.19832},
  year={2024}
}