Create README.me
Browse files
README.me
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
title: Amharic POS Tagger
|
| 3 |
+
emoji: 🇪🇹
|
| 4 |
+
colorFrom: blue
|
| 5 |
+
colorTo: green
|
| 6 |
+
sdk: gradio
|
| 7 |
+
sdk_version: 4.0.0
|
| 8 |
+
app_file: app.py
|
| 9 |
+
pinned: false
|
| 10 |
+
---
|
| 11 |
+
|
| 12 |
+
# Amharic Part-of-Speech Tagger
|
| 13 |
+
|
| 14 |
+
A neural network-based Part-of-Speech (POS) tagger for Amharic language using Bi-GRU with CRF architecture.
|
| 15 |
+
|
| 16 |
+
## Model Architecture
|
| 17 |
+
|
| 18 |
+
- **Embedding Layer**: 100-dimensional word embeddings
|
| 19 |
+
- **Encoder**: Bidirectional GRU with 64 hidden units
|
| 20 |
+
- **Decoder**: Conditional Random Field (CRF) for sequence labeling
|
| 21 |
+
- **Training**: Trained on Amharic text corpus with various POS tags
|
| 22 |
+
|
| 23 |
+
## Usage
|
| 24 |
+
|
| 25 |
+
1. Enter Amharic text in the input box
|
| 26 |
+
2. Click "Tag POS" to see the part-of-speech tags
|
| 27 |
+
3. Use the example buttons to try sample inputs
|
| 28 |
+
|
| 29 |
+
## POS Tags
|
| 30 |
+
|
| 31 |
+
The model predicts various POS tags including:
|
| 32 |
+
- Nouns (N)
|
| 33 |
+
- Verbs (V)
|
| 34 |
+
- Adjectives (ADJ)
|
| 35 |
+
- Adverbs (ADV)
|
| 36 |
+
- Prepositions (PREP)
|
| 37 |
+
- Pronouns (PRON)
|
| 38 |
+
- Determiners (DET)
|
| 39 |
+
- Conjunctions (CONJ)
|
| 40 |
+
|
| 41 |
+
## Technical Details
|
| 42 |
+
|
| 43 |
+
- Framework: PyTorch
|
| 44 |
+
- Interface: Gradio
|
| 45 |
+
- Language: Amharic (Ethiopia)
|