Documentation

Learn how to use Synfire to publish, discover, and share neuromorphic models in the Neuromorphic Intermediate Representation (NIR) format.

What is Synfire?

Synfire is a public, open-source registry and web platform for hosting and sharing neuromorphic models in the NIR format. It provides a centralized place to discover, version, and distribute neuromorphic computing models.

Synfire consists of three main components:

Web App - Browser-based UI for publishing, discovery, and management
CLI (synfire) - Command-line tool for automation and CI/CD workflows
Python SDK (synfire) - Programmatic API for Python applications

Explore the Docs

Getting Started

Quick introduction to Synfire and how to start using the registry.

CLI Reference

Complete reference for the synfire command-line tool for automation and CI/CD.

Python SDK

Programmatic API for Python workflows using the synfire package.

Publishing Guide

Step-by-step guide to publishing your neuromorphic models to the registry.

Key Concepts

Organization / Namespace: A unique prefix like nrg-lab that groups related repositories. Must be lowercase alphanumeric with hyphens (3-39 characters).
Repository: A container for model releases, identified as <org>/<repo>.
Release: An immutable, versioned set of artifacts following SemVer (e.g., 1.0.0, 1.0.0-beta).
Hardware Targets: Supported platforms: Pulsar (Innatera), Loihi2, SpiNNaker2, BrainScaleS-2, Xylo, Speck, and Generic.

Next Steps

Ready to get started? Head over to the Getting Started guide to learn the basics, or jump straight to the Publishing Guide if you have a model ready to share.