PING E/I networks trained with surrogate gradients; what sets the rate, the rhythm, and the sparsity.
Training the PING E/I network on the SHD event-based audio benchmark — from raw cochlear spikes to spoken-digit classification.
Model derivations, conventions, and how the toolkit works.
Odds and ends.
Operational notes for temporary compute, remote machines, schedulers, and cloud capacity.
Browse all entries · also available as a single pdf.