Seamless Deployment of Pre-trained Spiking Neural Networks onto SpiNNaker2

The rise of spiking neural networks (SNNs) offers a promising alternative to traditional activation-based deep neural networks, advancing the pursuit of more sustainable and ethical computing architectures. Despite its potential, neuromorphic event-based computing, which promises to fully harness the benefits of spiking networks, remains largely confined to research settings, with commercialization hindered by the lack of user-friendly software tools for developers. While initiatives like Intel's Lava for Loihi and PySpiNNaker2 for deploying SNNs on the SpiNNaker 2 neuromorphic chip are promising, many researchers face challenges in adapting their existing simulations to these platforms.

In this work, we introduce a streamlined workflow designed to facilitate the deployment of trained SNNs from various frameworks, such as SNNTorch, directly onto the SpiNNaker2 chip. Our solution leverages PySpiNNaker2 as a backend, enabling users to load their trained models onto SpiNNaker2 while continuing to use their preferred frontends. This approach preserves researchers' ability to prototype rapidly with familiar software, without sacrificing the opportunity to test their architectures on advanced neuromorphic hardware.

By bridging the gap between popular SNN frameworks and neuromorphic hardware, our workflow aims to significantly benefit both computational neuroscience and neuromorphic computing. It empowers researchers with a fast, user-friendly prototyping process and seamless integration, promoting innovation and accelerating the development of both fields.