abinet2018 commited on
Commit
9a6fc7b
Β·
verified Β·
1 Parent(s): 67b73dc

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +8 -20
README.md CHANGED
@@ -11,14 +11,9 @@ pinned: false
11
 
12
  # πŸ“– Amharic Poetry Joint Theme Classifier
13
 
14
- Classify Amharic poetry into six thematic categories using transformer-based models.
15
 
16
- ## ✨ Features
17
-
18
- - 🎯 **Joint Classification**: Predicts both dominant theme and multiple overlapping themes
19
- - πŸ€– **Multi-Model Support**: Rasyosef-RoBERTa and AfroXLM-R models
20
- - 🎨 **Interactive UI**: User-friendly Gradio interface with sample poems
21
- - πŸ“Š **Visual Results**: Probability charts with confidence scores
22
 
23
  ## πŸ“š Thematic Categories
24
 
@@ -31,24 +26,17 @@ Classify Amharic poetry into six thematic categories using transformer-based mod
31
  | 5 | Historical Poetry | ታαˆͺαŠ«α‹Š ግαŒ₯ም |
32
  | 6 | Love Poetry | α‹¨αα‰…αˆ­ ግαŒ₯ም |
33
 
34
- ## πŸš€ How to Use
35
-
36
- 1. **Enter Text**: Paste an Amharic poem in the text box
37
- 2. **Adjust Threshold**: Use the slider (default: 0.5) to control additional theme sensitivity
38
- 3. **Submit**: Click the "Submit" button
39
- 4. **View Results**: See the predicted theme with confidence scores and probability chart
40
-
41
  ## πŸ“Š Model Performance
42
 
43
- | Model | Accuracy | Multi-label F1 | Parameters |
44
- |-------|----------|----------------|------------|
45
- | Rasyosef-RoBERTa | 77.15% | 81.79% | 112.5M |
46
- | AfroXLM-R | 77.05% | 80.23% | 279.9M |
 
47
 
48
  ## πŸ“ Citation
49
 
50
- If you use this work in your research, please cite:
51
-
52
  @article{desta2026transformer,
53
  title={Transformer-Based Joint Thematic Classification for Amharic Poetry},
54
  author={Desta, Abinet Bizuayehu and Beyene, Million Meshesha and
 
11
 
12
  # πŸ“– Amharic Poetry Joint Theme Classifier
13
 
14
+ **Transformer-Based Joint Thematic Classification for Amharic Poetry**
15
 
16
+ Classify Amharic poetry into six thematic categories using a multi-task transformer model.
 
 
 
 
 
17
 
18
  ## πŸ“š Thematic Categories
19
 
 
26
  | 5 | Historical Poetry | ታαˆͺαŠ«α‹Š ግαŒ₯ም |
27
  | 6 | Love Poetry | α‹¨αα‰…αˆ­ ግαŒ₯ም |
28
 
 
 
 
 
 
 
 
29
  ## πŸ“Š Model Performance
30
 
31
+ | Metric | Score |
32
+ |--------|-------|
33
+ | Dominant Theme Accuracy | 77.15% |
34
+ | Multi-label Macro F1 | 81.79% |
35
+ | Parameters | 112.5M |
36
 
37
  ## πŸ“ Citation
38
 
39
+ ```bibtex
 
40
  @article{desta2026transformer,
41
  title={Transformer-Based Joint Thematic Classification for Amharic Poetry},
42
  author={Desta, Abinet Bizuayehu and Beyene, Million Meshesha and