This implementation of scANNs relies on numpy, keras and tensorflow. Any recent versions should be compatible, but this code was tested with numpy 1.23.5, tensorflow 2.13 and keras 2.13.1. Additionally, this code uses keras, but tensorflow.keras should work as well. Just modify the code accordingly.
To clone a copy of this repository,
git clone https://github.com/sandialabs/scANNs
If you are interested in contributing or looking for support with this repository, please contact Brad Aimone (jbaimon@sandia.gov).
Copyright 2024 National Technology & Engineering Solutions of Sandia, LLC (NTESS). Under the terms of Contract DE-NA0003525 with NTESS, the U.S. Government retains certain rights in this software.
This code is licensed under BSD3.
This repository is designed to provide code to support the publication:
Aimone, James B., William Severa, and J. Darby Smith. "Synaptic Sampling of Neural Networks." 2023 IEEE International Conference on Rebooting Computing (ICRC). IEEE, 2023.
Continuous developments are not expected, but the repository may be updated sporadically as new results are published.