Jishnuuuu/cyberguard-v1

CyberGuard - Cyberbullying Detection Model

Model Description

This model is based on DistilBERT and fine-tuned for detecting cyberbullying and threatening language in text messages. It's designed for the CyberGuard mobile application to protect children from online harassment.

Intended Use

• Real-time message analysis in social media apps
• Cyberbullying detection
• Content moderation
• Safety monitoring for children

How to Use

from transformers import pipeline

classifier = pipeline("text-classification", model="Jishnuuuu/cyberguard-v1")
result = classifier("You are so stupid")
print(result)
# Output: [{'label': 'NEGATIVE', 'score': 0.9998}]

Model Performance

• Base Model: DistilBERT (66M parameters)
• Accuracy: ~90% on sentiment classification
• Speed: ~50ms per inference
• Size: 255MB

Training Data

Based on DistilBERT fine-tuned on SST-2 (Stanford Sentiment Treebank):

• Binary sentiment classification (Positive/Negative)
• Adapted for cyberbullying detection

Limitations

• English language only
• May not catch context-dependent sarcasm
• Best used as part of a comprehensive safety system

Ethical Considerations

This model is designed to protect children's safety while respecting privacy. Message content is analyzed locally and only threat indicators are shared with parents.

License

MIT License - Free for commercial and non-commercial use

Citation

If you use this model, please cite:

@misc{cyberguard2025,
  title={CyberGuard: AI-Powered Cyberbullying Detection},
  author={CyberGuard Team},
  year={2025},
  publisher={Hugging Face},
  howpublished={\url{https://huggingface.co/Jishnuuuu/cyberguard-v1}}
}

Contact

For questions or issues, please contact through the Hugging Face model page.