back
user profile
felipemayamuniz
3karma·12submissions·March 1, 2026
recent activity (12 total)
comment
You have a point. Fixed!
comment
I trained a 354M decoder-only LLM called AletheionLLM-v2 on 1 billion tokens using a geometric cognitive architecture called ATIC (Adaptive Turing Intelligence Cognition), which adds per-token uncerta…