As part of my master thesis, I implemented a spiking neural network from scratch on a microcontroller. Because training was done separately on a PC, I only had to write the forward pass. This was a big relief because PyTorches autograd always seemed like black magic to me (even after learning the theoretical background). So seeing a post on HackerNews about Andrej Karpathys amazing video series on building neural networks from scratch, seemed like the perfect opportunity to fix this blind spo...
