samandar1105/news-classifier

DistilBERT Fine-Tuned on AG News

This model classifies news articles into 4 categories: World, Sports, Business, and Sci/Tech.

Model Details

• Base model: distilbert-base-cased
• Dataset: AG News (120,000 training samples)
• Task: Multi-class text classification (4 classes)
• Framework: PyTorch + HuggingFace Transformers

Performance

Metric	Score
Accuracy	0.9447
F1 Score	0.9447

Classes

ID	Label
0	World
1	Sports
2	Business
3	Sci/Tech

How to Use

from transformers import pipeline

classifier = pipeline(
    "text-classification",
    model="samandar1105/news-classifier"
)

result = classifier("Apple reported record quarterly earnings.")
print(result)  # [{'label': 'Business', 'score': 0.98}]

Training Details

• Learning rate: 2e-5
• Epochs: 4
• Batch size: 32
• Max sequence length: 128
• Warmup ratio: 0.1
• Weight decay: 0.01