Spyker is a high-performance library written from scratch that simulates spiking neural networks. It has both C++ and Python interfaces and can be easily integrated with popular tools like Numpy and PyTorch.
Prebuilt packages will be available soon. For now, you can follow the instructions on how to build the library form source here.
You can see the documentation for the C++ and Python interfaces here.
You can take a look at the tutorials listed below to learn how to use the library.
- Tutorial 1: Spyker and PyTorch
- Tutorial 2: Spyker and Numpy
- Tutorial 3: Sparse Spyker
- Tutorial 4: Other Functionalities
- Tutorial 5: Rate Coding
You can checkout example implementations of some networks in the examples directory. The example use the MNIST dataset, which is expected to be inside the MNIST directory beside the files, and the name of the files is expected to be train-images-idx3-ubyte, train-labels-idx1-ubyte, t10k-images-idx3-ubyte, t10k-labels-idx1-ubyte.
You can report bugs and request featues on the issues page.
This library has a BSD 3-Clause permissive license. You can read it here.