Code for the paper "Incorporating Post-Synaptic Effects into Online Training of Feed-Forward SNNs". This repo contains code to reproduce all figures and experiments in the paper. If you find this code useful please cite the paper
To reproduce Randman results, run python generate_randman_data.py
Set the settings you wish to produce in the beginning of script.
Then execute the script, which will save the data.
You can do the same for SHD with generate_SHD_data.py and generate_SHD_data-f.py
Use "generate_SHD_data-f.py" for online learning and generate_SHD_data.py for offline learning.
For plotting, use plots.ipynb.
For the loss landscape, set the matching settings of the generated Randman data in gen_l.py before running
To generate different seeds (0-3), change the value of "init_seed_val" in the scripts
The code is written using Python 3.10 the following packages:
- [jax] version 0.4.14
- [flax] 0.7.2
- [tonic] for loading Spiking Heidelberg Digits
- [randman] For randman dataset (install Randman from https://github.com/fzenke/randman)
- [optax] 0.1.7
- [numpy]
- [matplotlib]