romsi-hate-speech 1.0.0

Detect Romanized Sinhala hate speech using XLM-RoBERTa.

🇱🇰 Romanized Sinhala Hate Speech Detection

This project aims to detect hate speech in Romanized Sinhala text using transformer-based deep learning models (e.g., XLM-RoBERTa). It includes both the model training pipeline and inference (prediction) functionality.

---

🔡 Supported Input

The model expects Romanized Sinhala text, such as:

• "umbalata lokuwatma pissu"
• "mokada oya karanne"

The input is automatically tokenized and passed to the model for classification.

---

🧪 Model Output

• 0 → Not Hate Speech
• 1 → Hate Speech

---

📜 License

This project is for academic and research use. Please contact the author before using it for commercial purposes.

---

✍️ Author

Sakun Chamikara
MSc Research Project – Romanized Sinhala Hate Speech Detection
Sri Lanka