News_Classifier-bert-base-uncased

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
• Dataset: AG News
• Fine-tuning Approach: Sequence Classification

Training Results

Confusion Matrix

Confusion Matrix Values (True Label → Predicted Label)

How to Use

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

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. 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