onestardao

A Plain-Text Rea­son­ing Frame­work for Trans­par­ent AI Align­ment Experiments

Can For­mal “Solver Loops” Make LLMs Ac­tu­ally Rea­son? Four Math­e­mat­i­cal Proposals

Can Se­man­tic Com­pres­sion Be For­mal­ized for AGI-Scale In­ter­pretabil­ity? (Ini­tial ex­per­i­ments via an open-source rea­son­ing ker­nel)

Can a se­man­tic com­pres­sion ker­nel like WFGY im­prove LLM al­ign­ment and in­sti­tu­tional ro­bust­ness?

WFGY: A Self-Heal­ing Rea­son­ing Frame­work for LLMs — Open for Tech­ni­cal Scrutiny