Bug in Synaptic neuron forward function [solved]

[Roiwa]

• snntorch version: 0.9.1
• Python version: 3.12.3
• Operating System: Ubuntu x64

Description

During tutorial 4, I have found that Synaptic neuron model does not simulate correctly the expected behaviour. When passing the syn and mem value in the function call, lif1(spk_in[step], syn, mem), both values tend to infinity. Also, I ran the code that SNNTorch has for google colab and found the same issue.

What I Did

When reviewing the Synaptic code, I observed that when passing the syn and mem parameters the forward function overrides the syn attribute (self.syn) with the mem value instead of the syn value, we can see this behaviour in lines 218-219:

if not syn == None:
            self.syn = mem

So, I change this line with the following statement and the bug is removed.

if not syn == None:
            self.syn = syn