rizwanulrudra/bangla-hate-speech-detector

Bangla Hate Speech Detector

A fine-tuned BanglaBERT model for detecting hate speech in Bengali text, classifying into:

1. Personal (0)
2. Political (1)
3. Religious (2)
4. Geopolitical (3)

Model Details

• Base Model: csebuetnlp/banglabert
• Fine-Tuned On: rezacsedu/Bengali-Hate-Speech-Dataset
• License: MIT

Performance

Metric	Score
Accuracy	0.89
F1 (Macro)	0.87

How to Use

from transformers import pipeline

classifier = pipeline(
    "text-classification", 
    model="rizwanulrudra/bangla-hate-speech-detector"
)

sample_text = "এই সরকার কে যারা নির্লজ্জের মত সাপোর্ট দিয়েছে..."
classifier(sample_text)
# Output: {'label': 'Political', 'score': 0.92}