Add README for Hugging Face Space
Browse files
README.md
CHANGED
|
@@ -10,4 +10,62 @@ pinned: false
|
|
| 10 |
license: mit
|
| 11 |
---
|
| 12 |
|
| 13 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
license: mit
|
| 11 |
---
|
| 12 |
|
| 13 |
+
---
|
| 14 |
+
title: Tawalt
|
| 15 |
+
emoji: π
|
| 16 |
+
colorFrom: blue
|
| 17 |
+
colorTo: indigo
|
| 18 |
+
sdk: gradio
|
| 19 |
+
sdk_version: 6.9.0
|
| 20 |
+
app_file: app.py
|
| 21 |
+
pinned: true
|
| 22 |
+
---
|
| 23 |
+
|
| 24 |
+
# Tawalt π β The Language AI Forgot
|
| 25 |
+
|
| 26 |
+
> The first conversational NLP pipeline built natively for Tarifit Berber.
|
| 27 |
+
|
| 28 |
+
## What is Tawalt?
|
| 29 |
+
|
| 30 |
+
Tarifit is a Berber language spoken by 5M+ people in Northern Morocco.
|
| 31 |
+
No NLP model. No dataset. No chatbot. Nothing β until now.
|
| 32 |
+
|
| 33 |
+
Tawalt is a real-time chat pipeline that understands and responds in Tarifit.
|
| 34 |
+
|
| 35 |
+
## Features
|
| 36 |
+
|
| 37 |
+
- π¬ Responds in Tarifit Berber
|
| 38 |
+
- π§ Intent detection β greetings, questions, farewells, apologies
|
| 39 |
+
- βοΈ Auto-corrects typos in Tarifit text
|
| 40 |
+
- β‘ Real-time message queue processing
|
| 41 |
+
|
| 42 |
+
## Try it
|
| 43 |
+
|
| 44 |
+
Type any of these to start:
|
| 45 |
+
|
| 46 |
+
| Tarifit | Meaning |
|
| 47 |
+
|---------|---------|
|
| 48 |
+
| azul | hello |
|
| 49 |
+
| sba7 lkhir | good morning |
|
| 50 |
+
| manec t-llid? | how are you? |
|
| 51 |
+
| sam7ahi | sorry |
|
| 52 |
+
| ar nemzar | goodbye |
|
| 53 |
+
|
| 54 |
+
## Tech Stack
|
| 55 |
+
|
| 56 |
+
- Python 3
|
| 57 |
+
- spaCy 3.8
|
| 58 |
+
- rapidfuzz
|
| 59 |
+
- Gradio
|
| 60 |
+
|
| 61 |
+
## Links
|
| 62 |
+
|
| 63 |
+
- π Notebook: [Google Colab](https://github.com/jamalinu/tawalt)
|
| 64 |
+
- π» Code: [GitHub](https://github.com/jamalinu/tawalt)
|
| 65 |
+
|
| 66 |
+
## Status
|
| 67 |
+
|
| 68 |
+
`v0.1 β Active development`
|
| 69 |
+
|
| 70 |
+
750 million people speak low-resource languages invisible to modern AI.
|
| 71 |
+
Tarifit is just the beginning.
|