hls4snn

hls4snn is a fork of hls4ml focused on spiking neural network (SNN) functionality in the PyTorch frontend, while keeping the upstream hls4ml workflow and documentation structure.

What's Different In This Fork

• Added initial SNN conversion support for PyTorch models using snntorch modules.
• Added SNN neuron/readout templates and related conversion backend plumbing.
• Added SNN-focused tests and documentation.

SNN Quick Start

Install SNN development dependencies from this repository:

pip install -r requirements-snn-dev.txt

SNN feature details, supported modules, and behavior notes are documented in:

• docs/advanced/snn.rst

A demo of the SNN functionality is in:

• hls4ml-snn-example.ipynb
• snn-hls4ml-demo.ipynb

These take you from building and training a simple SNN in snntorch, to running it through hls4ml and getting Vitis reports. The longer demo also includes optional NeuroBench metrics.