Overview

Beetorch is a PyTorch-based library designed to simplify testing the robustness of machine learning models.

Report: Robustness of High-Dimensional Deep Learning Models

Installation

To use Beetorch, you will need the following Python packages:

• numpy
• torch, torchmetrics
• mysql-connector-python
• pushbullet.py

You can install these packages using pip:

pip install numpy torch torchmetrics mysql-connector-python pushbullet.py

Features

PoisonClass

The PoisonClass in beetorch/init.py is used to handle poisoning of the data. It provides methods to initialize poison and convert poison to string.

Model

The Model class in beetorch/init.py is the main class for creating and training machine learning models.

Saver

The Saver class in beetorch/init.py is used to save the state of the model.

Pushbullet_saver

The Pushbullet_saver class in beetorch/pushbullet.py extends the Saver class and is used to send notifications about the model's progress via Pushbullet.

Contributing

Coming soon...