Myanmar Hate Speech Detection Model
Model Description
This model is a Myanmar (Burmese) Hate Speech Detection system fine-tuned from bert-base-multilingual-cased for binary text classification.
The model classifies text into:
- 0 → Normal Speech
- 1 → Hate Speech
It is designed for academic and research purposes.
Base Model
- bert-base-multilingual-cased
- Transformer architecture
- Fine-tuned using Hugging Face Transformers
Dataset
The model was trained using:
- Gojokun00/myanmar-speech-hate-and-normal
The dataset contains Myanmar text labeled as hate or non-hate speech.
Training Details
- Epochs: 3
- Batch size: 8
- Learning rate: 2e-5
- Optimizer: AdamW
- Evaluation strategy: Validation split