Open Neuromorphic
Join Discord (2100+ joined)

Neuromorphic Intermediate Representation (NIR)

A graph-based intermediate representation for computational graphs of spiking neural networks, enabling interoperability across different simulators and hardware.

Overview

By defining a common set of computational primitives (like Leaky-Integrate-and-Fire neurons and convolutions), NIR allows researchers and developers to define a model once and then translate it to run on different backends without having to rewrite the model from scratch for each platform. This decouples the model definition from the hardware or software-specific implementation details.

NIR is designed to be extensible and currently supports a range of popular SNN frameworks and hardware, including:

More information can be found in the NIR documentation and the Nature Communications paper.
It is actively being developed on GitHub with additional tools for PyTorch integrations.

Related Content, Events & Resources

Open-Source Neuromorphic Research Infrastructure: A Community Panel
Workshops

Join leading maintainers of neuromorphic software libraries for a panel discussion on building open-source infrastructure, sharing lessons learned, …

NIR: A unified instruction set for brain-inspired computing
Workshops

We show how to use the Neuromorphic Intermediate Representation to migrate your spiking model onto neuromorphic hardware.

Software

Framework for developing neuro-inspired applications, mapping them to neuromorphic hardware.

Software

Python package for building, testing, deploying neural networks, supporting many backends for SNN simulation.

Software

Exploits bio-inspired neural components, sparse and event-driven, expands PyTorch with primitives for bio-inspired neural components.

Software

Machine learning library for SNN applications, supports GPU, TPU, CPU acceleration, and neuromorphic compute hardware deployment.

Software

PyTorch-based DL library for SNNs, focusing on simplicity, fast training, extendability, and vision models.

Software

Focuses on gradient-based training of SNNs, based on PyTorch for GPU acceleration and gradient computation.

Software

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

Loihi 2 - Intel
Hardware

Learn about Intel's neuromorphic hardware: Loihi 2

Speck - SynSense
Hardware

Learn about SynSense's neuromorphic hardware: Speck

SpiNNaker 2 - University of Dresden
Hardware

Learn about TU Dresden's neuromorphic hardware: SpiNNaker2

Xylo - SynSense
Hardware

Learn about SynSense's neuromorphic hardware: Xylo

Help Improve this Software Guide

Our software guide is maintained by the community. If you have updates, see an error, or want to suggest a new tool, please let us know by opening an issue on our GitHub repository.

Open an Issue Edit this page