Spyx

Compact SNN package on DeepMind's Haiku library, based on JAX for JIT compilation on GPUs and TPUs.

Spyx

Overview

Spyx is a compact spiking neural network library built on top of DeepMind’s Haiku package. It aims to blend the flexibility and extensibility typical of PyTorch-based SNN libraries with efficient training capabilities found in frameworks with custom CUDA implementations. The library is optimized for high-performance simulations on GPUs and TPUs, which is critical for handling the computationally intensive nature of large-scale SNNs. Spyx is able to train models at comparable speeds to frameworks with custom CUDA implementations by extensively leveraging Just-In-Time compilation; another interesting feature is the ability to pack neuromorphic data into int8 datatypes and unpack them during training time, allowing datasets to be stored in completely in VRAM at a fraction of their normal size.

The library is designed to be a streamlined solution for SNN development, supporting diverse model structures and algorithms. Its documentation covers various aspects, including quick start guides, tutorials on surrogate gradients, training SNNs using neuroevolution, comparisons of spiking neuron models, and surrogate gradient functions. Additionally, it provides a template for creating surrogate gradients and a comprehensive API reference. The library is designed to be modular, allowing users to easily integrate custom models and algorithms, enhancing its flexibility. Spyx is a powerful and specialized tool with great strength in its high-performance simulation capabilities and versatility for various research applications.

Can you contribute tutorial guides or case studies?

Get Involved with ONM

Spiking Neurons: A Digital Hardware Implementation

Spiking Neurons: A Digital Hardware Implementation

  • Fabrizio Ottati

Learn how to model Leaky Integrate and Fire (LIF) neurons in digital hardware. Understand spike communication, synapse integration, and more for hardware implementation.

NorthPole, IBM's latest Neuromorphic AI Hardware

NorthPole, IBM's latest Neuromorphic AI Hardware

  • Fabrizio Ottati

Translating the NorthPole paper from IBM to human language.

Spiking Neural Network (SNN) Library Benchmarks

Spiking Neural Network (SNN) Library Benchmarks

  • Gregor Lenz, Kade Heckel, Sumit Bam Shrestha, Cameron Barker, Jens Egholm Pedersen

Discover the fastest Spiking Neural Network (SNN) frameworks for deep learning-based optimization. Performance, flexibility, and more analyzed in-depth