Sinabs Is Not A Brain Simulator

It’s a deep learning library based on PyTorch for spiking neural networks, with a focus on simplicity, fast training and extendability. Sinabs works well for Vision models because of its support for weight transfer. If you’re looking to work with audio data or different backends, be sure to check out Rockpool as well.

Getting started

• Install Sinabs and potentially some plugins.

• Dive right into the syntax if you know your way around SNN simulators.

Speck/Dynap-CNN DevKits

• Overview of the development kits.

Tutorials

• Convert an existin ANN if you want to get started quickly.

• Run a first example using BPTT with this neuromorphic version of the MNIST dataset.

Plugins

• Training feed-forward models 10x faster: EXODUS.

API reference

• Complete reference overview.

• Supported neuron models.

• Weight transfer API.

About

• About Sinabs. How the project came about.

• Contribution guidelines. Please read this before opening a pull request.

• Release notes. Version changes.

Contact

• Contact us. For questions and bug reports!