Create optimize utility/quick train and eval loop

kmheckel / spyx

Spyx: Spiking Neural Networks in JAX

https://spyx.readthedocs.io/en/latest/

MIT License

102 stars 12 forks source link

Open kmheckel opened 8 months ago

kmheckel commented 8 months ago

Make spyx.optimize:

[ ] surrogate gradient training loop
[ ] take SNN, params, Dataset in form (#batches, batchsize, timesteps, ...), optimizer
[ ] also take number of epochs, maybe eval frequency.