Update README.md
Browse files
README.md
CHANGED
|
@@ -1,13 +1,78 @@
|
|
| 1 |
---
|
| 2 |
-
title:
|
| 3 |
-
emoji:
|
| 4 |
colorFrom: purple
|
| 5 |
-
colorTo:
|
| 6 |
sdk: gradio
|
| 7 |
-
sdk_version:
|
| 8 |
app_file: app.py
|
| 9 |
pinned: false
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
---
|
| 12 |
|
| 13 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
title: Human vs AI Text Detector
|
| 3 |
+
emoji: 🔍
|
| 4 |
colorFrom: purple
|
| 5 |
+
colorTo: blue
|
| 6 |
sdk: gradio
|
| 7 |
+
sdk_version: 4.44.0
|
| 8 |
app_file: app.py
|
| 9 |
pinned: false
|
| 10 |
+
license: apache-2.0
|
| 11 |
+
tags:
|
| 12 |
+
- text-classification
|
| 13 |
+
- human-ai-text-attribution
|
| 14 |
+
- african-languages
|
| 15 |
+
- multilingual
|
| 16 |
+
- hata
|
| 17 |
---
|
| 18 |
|
| 19 |
+
# 🔍 Human vs AI Text Detector
|
| 20 |
+
|
| 21 |
+
Detect whether text is human-written or AI-generated across multiple African languages!
|
| 22 |
+
|
| 23 |
+
## 🌟 Features
|
| 24 |
+
|
| 25 |
+
- **Multilingual Support**: Works with English, Yoruba, Hausa, Igbo, Swahili, Amharic, and Nigerian Pidgin
|
| 26 |
+
- **High Accuracy**: 100% accuracy on validation set
|
| 27 |
+
- **Fair & Unbiased**: Explicitly trained with fairness constraints across all languages
|
| 28 |
+
- **Easy to Use**: Simple interface for single text or batch processing
|
| 29 |
+
|
| 30 |
+
## 🎯 How to Use
|
| 31 |
+
|
| 32 |
+
1. **Single Text**: Paste your text in the input box and click "Classify Text"
|
| 33 |
+
2. **Batch Processing**: Upload a `.txt` file with one text per line for batch classification
|
| 34 |
+
3. **Examples**: Try the pre-loaded examples in different languages
|
| 35 |
+
|
| 36 |
+
## 🔬 Model Details
|
| 37 |
+
|
| 38 |
+
- **Base Model**: AfroXLMR-base
|
| 39 |
+
- **Parameters**: ~270M
|
| 40 |
+
- **Training**: Fine-tuned on PhD HATA African Dataset
|
| 41 |
+
- **Fairness**: EOD = 0.0, AAOD = 0.0 (perfect fairness)
|
| 42 |
+
|
| 43 |
+
## 📊 Performance
|
| 44 |
+
|
| 45 |
+
| Metric | Score |
|
| 46 |
+
|--------|-------|
|
| 47 |
+
| Accuracy | 100% |
|
| 48 |
+
| F1 Score | 100% |
|
| 49 |
+
| Precision | 100% |
|
| 50 |
+
| Recall | 100% |
|
| 51 |
+
|
| 52 |
+
## ⚠️ Limitations
|
| 53 |
+
|
| 54 |
+
- Optimized for African languages in training set
|
| 55 |
+
- Performance may vary on newer AI generation systems
|
| 56 |
+
- Should be used as part of a broader content verification system
|
| 57 |
+
|
| 58 |
+
## 🔗 Links
|
| 59 |
+
|
| 60 |
+
- [Model Repository](https://huggingface.co/msmaje/phdhatamodel)
|
| 61 |
+
- [Training Visualizations](https://huggingface.co/msmaje/phdhatamodel/tree/main/visualizations)
|
| 62 |
+
- [Dataset](https://huggingface.co/datasets/msmaje/phd-hata-african-dataset)
|
| 63 |
+
|
| 64 |
+
## 📚 Citation
|
| 65 |
+
|
| 66 |
+
```bibtex
|
| 67 |
+
@misc{msmaje2025hata,
|
| 68 |
+
author = {Maje, M.S.},
|
| 69 |
+
title = {AfroXLMR for Human-AI Text Attribution},
|
| 70 |
+
year = {2025},
|
| 71 |
+
publisher = {HuggingFace},
|
| 72 |
+
url = {https://huggingface.co/msmaje/phdhatamodel}
|
| 73 |
+
}
|
| 74 |
+
```
|
| 75 |
+
|
| 76 |
+
## 📧 Contact
|
| 77 |
+
|
| 78 |
+
For questions or feedback, please open an issue in the model repository.
|