🛡️ PhishGuard: BERT-Based Phishing Detection

PhishGuard is an NLP-powered phishing detection model built using BERT.
It classifies text (e.g., emails, URLs, or messages) as phishing or legitimate.

The model is trained and hosted on Hugging Face, while this repository contains the training pipeline, preprocessing scripts, and evaluation code.

---

🚀 Model

The fine-tuned model is available on Hugging Face:
👉 bert-phishing-detector

You can load it directly in Python:

from transformers import AutoTokenizer, AutoModelForSequenceClassification

tokenizer = AutoTokenizer.from_pretrained("Swathi37/bert-phishing-detector")
model = AutoModelForSequenceClassification.from_pretrained("Swathi37/bert-phishing-detector")

text = "Your account has been suspended. Click here to verify."
inputs = tokenizer(text, return_tensors="pt")
outputs = model(**inputs)
print(outputs.logits)

📦 Installation

git clone https://github.com/SwathiPriya37/PhishGuard.git
cd PhishGuard
pip install -r requirements.txt

🏋️ Training

To fine-tune BERT on your phishing dataset:

python src/train_bert.py

📊 Evaluation

Evaluate the trained model:

python src/evaluate_model.py

📂 Using Your Own Dataset

To train on your own dataset, prepare a CSV file with the following format:

csv text,label "Your account is locked. Verify now.",phishing "Meeting is scheduled at 3 PM tomorrow.",legitimate Then run:

bash

python src/train_bert.py --data data/your_dataset.csv

🤝 Contributing

Pull requests are welcome! For major changes, please open an issue first to discuss what you’d like to add.

📜 License

This project is licensed under the MIT License.

👩‍💻 Author

Developed by Swathi Priya R Model: Swathi37/bert-phishing-detector