|
|
--- |
|
|
tags: |
|
|
- multilabel-classification |
|
|
- emotion-detection |
|
|
- text-classification |
|
|
- transformers |
|
|
- deberta |
|
|
- huggingface |
|
|
license: apache-2.0 |
|
|
datasets: |
|
|
- custom |
|
|
language: |
|
|
- en |
|
|
--- |
|
|
|
|
|
# DeBERTa-v3-Large for Multilabel Emotion Classification |
|
|
|
|
|
This model is a fine-tuned version of [`microsoft/deberta-v3-large`](https://huggingface.co/microsoft/deberta-v3-large) for multilabel emotion classification. |
|
|
It was trained on a custom dataset where each text sample may contain multiple emotions. |
|
|
|
|
|
## π Model Card |
|
|
|
|
|
- **Model**: `FurqonAryadana/deberta-emotion-multilabel-0.5007` |
|
|
- **Base**: DeBERTa-v3-Large |
|
|
- **Task**: Multilabel Emotion Classification |
|
|
- **License**: Apache 2.0 |
|
|
- **Language**: English |
|
|
- **Threshold Tuning**: Applied per-label |
|
|
|
|
|
## π Evaluation (Validation Set) |
|
|
|
|
|
**Detailed Classification Report (Threshold Tuned)**: |
|
|
|
|
|
| Emotion | Precision | Recall | F1-score | Support | |
|
|
|----------------|-----------|--------|----------|---------| |
|
|
| amusement | 0.61 | 0.72 | 0.66 | 851 | |
|
|
| anger | 0.43 | 0.58 | 0.49 | 999 | |
|
|
| annoyance | 0.34 | 0.70 | 0.46 | 1609 | |
|
|
| caring | 0.41 | 0.57 | 0.48 | 594 | |
|
|
| confusion | 0.49 | 0.71 | 0.58 | 800 | |
|
|
| disappointment | 0.30 | 0.47 | 0.37 | 990 | |
|
|
| disgust | 0.27 | 0.49 | 0.35 | 584 | |
|
|
| embarrassment | 0.33 | 0.32 | 0.33 | 308 | |
|
|
| excitement | 0.45 | 0.47 | 0.46 | 632 | |
|
|
| fear | 0.49 | 0.54 | 0.52 | 321 | |
|
|
| gratitude | 0.83 | 0.73 | 0.78 | 955 | |
|
|
| joy | 0.43 | 0.58 | 0.50 | 876 | |
|
|
| love | 0.64 | 0.79 | 0.71 | 701 | |
|
|
| sadness | 0.40 | 0.57 | 0.47 | 714 | |
|
|
|
|
|
- **F1 Micro**: `0.5101` |
|
|
- **F1 Macro**: `0.5107` |
|
|
|
|
|
## π§ Emotions (Label Order) |
|
|
|
|
|
```python |
|
|
[ |
|
|
'amusement', 'anger', 'annoyance', 'caring', 'confusion', 'disappointment', |
|
|
'disgust', 'embarrassment', 'excitement', 'fear', 'gratitude', 'joy', |
|
|
'love', 'sadness' |
|
|
] |
|
|
|