Awesome

Improving Fractal Pre-training

<a href="https://pytorch.org/get-started/locally/"><img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-ee4c2c?logo=pytorch&logoColor=white"></a> <a href="https://pytorchlightning.ai/"><img alt="Lightning" src="https://img.shields.io/badge/-Lightning-792ee5?logo=pytorchlightning&logoColor=white"></a> <a href="https://hydra.cc/"><img alt="Config: Hydra" src="https://img.shields.io/badge/Config-Hydra-89b8cd"></a>

This is the official PyTorch code for Improving Fractal Pre-training (arXiv).

@article{anderson2021fractal,
  author  = {Connor Anderson and Ryan Farrell},
  title   = {Improving Fractal Pre-training},
  journal = {arXiv preprint arXiv:2110.03091},
  year    = {2021},
}

Setup

The code uses PyTorch-Lightning for training and Hydra for configuration. Other required packages are listed in install_requirements.sh.

# clone project
git clone https://github.com/catalys1/fractal-pretraining.git
cd fractal-pretraining

# [RECOMMENDED] set up a virtual environment
python3 -m venv venv_name  # choose your prefered venv name
source venv/bin/activate

# install requirements
bash install_requirements.sh
# install project in editable mode
pip install -e fractal_learning

Sample and Render Iterated Function Systems

See the fractals sub-package for details on sampling IFS codes and rendering fractal images.

Training

See the training sub-package for details on pre-training with fractal images, as well as finetuning on other datasets.