Spam Detection Web Tool

Spam SMS Detection Project implemented using NLP & Transformers. roBerta - a hugging face Transformer model for text classification is used to fine-tune to best suit data to achieve the best results. The model was deployed on the Flask server.

Demo

Description

Training SimpleTransformers model on SMS Spam Collection Dataset, resulting in 0.9993 accuracy on the test results.

Getting Started

Dependencies

• Please refer to .The pyTorch website to install PyTorch
• Make the appropriate adjustments to enable Cuda on your PC for faster experience
• Python 3.8.6
• transformers 4.17.0
• torch 1.9.0

Installing

• git clone this repository
• Create a virtual environment using

Python -m venv .

• run

pip install -r /path/to/requirements.txt

• download The Trained Model and extract to the root repository

Executing program

CMD:

• Inside the folder load CMD
• run

python application.py

PyCharm

• load the projects folder and press Run.
• The website will automatically load.
• Insert your messages and press Predict.

Training the model yourself

• Download the "train_model" folder onto your PC
• Follow the installations command to set up the environment
• Execute

python training.py

• At the end of the training, 'outputs' folder will be created. This is your trained model.

Contributors

Asahel Cohen