Soft Decision Tree

An implementation of Frosst & Hinton's Distilling a Neural Network Into a Soft Decision Tree

Requirements

The project was developed using Python 3.6 and uses the following libraries:

• PyTorch 1.3.1
• Torchvision 0.4.2
• Tensorboard 1.15.0

One can install the required libraries by using pip in a Conda environment or virtualenv with the provided requirements.txt file:

pip install -r requirements.txt

Usage

main.py will fit a Soft Decision Tree on the MNIST dataset, acquired through Torchvision. One can simply run the script with default arguments with the following command: python main.py

Please refer to args.py to retrieve all the available arguments and alter to your liking :)