---
language: en
license: apache-2.0
datasets:
- ag_news
tags:
- text-classification
- bert
- ag-news
---

# BERT-base-uncased fine-tuned on AG News

This model is a fine-tuned version of `bert-base-uncased` on the AG News dataset, achieving **94.36% accuracy** on the test set.

## Model Details

- **Model Type:** Text Classification (BERT)
- **Base Model:** [bert-base-uncased](https://huggingface.co/bert-base-uncased)
- **Dataset:** [AG News](https://huggingface.co/datasets/ag_news)
- **Fine-tuning Approach:** Sequence Classification

## Training Results

| Epoch | Training Loss | Validation Loss | Accuracy | F1 (Weighted) |
|-------|---------------|-----------------|----------|---------------|
| 1     | 0.231600      | 0.212338        | 0.9359   | 0.9359        |
| 2     | 0.176300      | 0.213332        | 0.9439   | 0.9439        |
| 3     | 0.119100      | 0.230517        | 0.9450   | 0.9450        |
| 4     | 0.074500      | 0.286154        | 0.9447   | 0.9448        |
| 5     | 0.031700      | 0.344374        | 0.9436   | 0.9435        |

## Confusion Matrix


### Confusion Matrix Values (True Label → Predicted Label)

|             | World | Sports | Business | Sci/Tech |
|-------------|-------|--------|----------|----------|
| **World**   | 1812  | 13     | 43       | 32       |
| **Sports**  | 7     | 1880   | 7        | 6        |
| **Business**| 39    | 9      | 1728     | 124      |
| **Sci/Tech**| 34    | 10     | 105      | 1751     |

## How to Use

```python
from transformers import pipeline

classifier = pipeline("text-classification", model="ShahzaibAli-1/News_Classifier-bert-base-uncased")

result = classifier("Apple reported record profits last quarter.")

print(result)
```

## Performance

### Training Hyperparameters

- Learning Rate: 5e-5
- Batch Size: 8
- Epochs: 5
- Warmup Ratio: 0.1
- Max Sequence Length: 128

Final Test Accuracy: 94.36%

Final Test F1-Score (Weighted): 94.35%

### To watch a proper demo using Gradio

```python
from transformers import pipeline

classifier = pipeline("text-classification", model="ShahzaibAli-1/News_Classifier-bert-base-uncased")

result = classifier("Apple reported record profits last quarter.")

print(result)

import gradio as gr
from transformers import pipeline

# Load model
classifier = pipeline("text-classification", model="ShahzaibAli-1/News_Classifier-bert-base-uncased")

# Define label mapping (must match your training labels)
label_map = {
    0: "World",
    1: "Sports", 
    2: "Business",
    3: "Sci/Tech"
}

def predict(text):
    result = classifier(text)[0]
    # Extract numerical label (e.g., "LABEL_1" -> 1)
    label_num = int(result['label'].split("_")[-1])
    # Get corresponding text label
    label_text = label_map[label_num]
    return f"{label_text} (confidence: {result['score']:.2%})"

# Create interface
iface = gr.Interface(
    fn=predict,
    inputs=gr.Textbox(lines=2, placeholder="Enter news text here..."),
    outputs="text",
    title="AG News Classifier",
    description="Classify news articles into World, Sports, Business, or Sci/Tech categories"
)

iface.launch()
```

## Example Outputs

Here are some example outputs for various test cases:

- **Sports News**:  
  Prompt: `"Newzealand Won the Test Championship today"`  
  Output: `Sports (confidence: 99.99%)`

- **Business News**:  
  Prompt: `"The stock market saw a significant increase following the tech boom"`  
  Output: `Business (confidence: 98.50%)`

- **World News**:  
  Prompt: `"The political unrest in Eastern Europe has escalated this week"`  
  Output: `World (confidence: 97.70%)`

- **Sci/Tech News**:  
  Prompt: `"Scientists have developed a new battery that can last twice as long as current models"`  
  Output: `Sci/Tech (confidence: 96.30%)`

## Evaluation Metrics

The following evaluation metrics were used to assess the model's performance:

- **Accuracy**: The percentage of correct predictions over the total number of predictions.
- **Precision**: The proportion of positive predictions that were actually correct.
- **Recall**: The proportion of actual positives that were correctly identified.
- **F1-Score**: The harmonic mean of precision and recall.

The model demonstrated strong performance across all metrics, particularly with an accuracy of 94.36%.

---

## Citation

If you use this model in your research or projects, please cite it as follows:

```
@article{shahzaib2025news,
  title={Fine-Tuning BERT for AG News Classification},
  author={Shahzaib Ali},
  journal={Hugging Face Model Hub},
  year={2025},
  url={https://huggingface.co/ShahzaibAli-1/News_Classifier-bert-base-uncased}
}
```

## License

The model is released under the [Apache-2.0 License](https://opensource.org/licenses/Apache-2.0). Feel free to use it in your applications and research.

## Contact

For any questions or suggestions, feel free to open an issue or contact the model creator at:  
- **Hugging Face**: [ShahzaibAli-1](https://huggingface.co/ShahzaibAli-1)