|
|
--- |
|
|
language: |
|
|
- en |
|
|
metrics: |
|
|
- accuracy |
|
|
- f1 |
|
|
- precision |
|
|
- recall |
|
|
base_model: |
|
|
- distilbert/distilbert-base-uncased-finetuned-sst-2-english |
|
|
pipeline_tag: text-classification |
|
|
tags: |
|
|
- fake |
|
|
- real |
|
|
- news |
|
|
library_name: transformers |
|
|
--- |
|
|
|
|
|
# DistilBERT Fake News Classifier |
|
|
|
|
|
## Model Description |
|
|
This DistilBERT-based model achieves **97.18% accuracy** in classifying news articles as real or fake, with balanced precision (97.17%) and recall (97.30%). |
|
|
|
|
|
## Training Performance |
|
|
| Epoch | Training Loss | Validation Loss | Accuracy | F1 Score | |
|
|
|-------|---------------|-----------------|----------|----------| |
|
|
| 1 | - | 0.1115 | 96.08% | 96.09% | |
|
|
| 2 | 0.2026 | 0.1077 | 97.25% | 97.28% | |
|
|
| 3 | 0.0647 | 0.1119 | 97.45% | 97.50% | |
|
|
|
|
|
## Final Test Results |
|
|
| Metric | Score | |
|
|
|------------|--------| |
|
|
| Accuracy | 97.18% | |
|
|
| F1 Score | 97.23% | |
|
|
| Precision | 97.17% | |
|
|
| Recall | 97.30% | |
|
|
|
|
|
## Usage |
|
|
```python |
|
|
from transformers import pipeline |
|
|
|
|
|
classifier = pipeline("text-classification", |
|
|
model="KenLumod/ML-Project-DistilBERT-Fake-and-Real-Classifier") |
|
|
result = classifier("Scientists confirm climate change accelerating beyond previous estimates") |
|
|
# Output: {'label': 'REAL', 'score': 0.982} |
