Essai Detect

A simple Windows application for detecting LLM-generated text in essays. The system uses either BERT and logistic regression model for predictions. Input the text in a chat like UI and get the result as response along with the model confidence.

Table of Contents

• Features
• Usage
• Installation
• License
• Acknowledgments

Features

• Message Box - Input text for LLM-generated or Human-written text classification.
• Chat History - Displays text input history, model response and confidence level.

Usage

1. Input text to be analyze in the Message Box.
2. Pressing the Send button or Enter Key in the keyboard, will feed it to the model.
3. Text must be more than 100 words to maintain a more accurate prediction.
4. Reloading the UI will cause the chat history to be reset.

Installation

1. Download and install the latest version of Essai Detect.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• LLM - Detect AI Generated Text Dataset: For LLM-generated and Human-written dataset.
• Phosphor Icons: For icon library.
• uv: For Python dependencies manager and virtual environment.
• Flask CORS: For handling resource sharing between Python and JavaScript.
• Waitress: For production-ready WSGI server.
• PyWebview: For standalone web app wrapper.
• PyInstaller: For building the app into executables.
• Inno Setup: For installer.
• Scikit-learn: For machine learning and statistical modeling.
• Joblib: For saving and loading trained machine learning models.
• Torch: For deep learning framework and prediction.
• Transformers: For utilizing pre-trained Bert model.
• Safetensors: For saving and loading of deep learning model weights.
• Numpy: For handling different types of arrays.
• Pandas: For data manipulation and analysis of Excel data.