Drop the ON/OFF aside, the 8-replica anecdote and "next step" wording;
explain the per-line spike rule, define drawn-digit accuracy, separate
GoldenModel accuracy from counter verification, and phrase backend and
neuron-flag notes without project-internal references.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XFVgZ8Nct8BEh9k7bjuRdw
Definitions, system overview, architecture comparison and interactive
widgets: AltAILIF neuron, input coding, STDP, Altai core grid and
crossbar, quantisation, chip budget, and three recognisers (STDP
columns, native network, tick-trained spiking network) computed in
the browser with integer arithmetic on their real weights.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01XFVgZ8Nct8BEh9k7bjuRdw
