Properties2
TypeIdea
Note createdNov 15, 2025

LLMs (and, for that matter, most of ML models) are grown, not build: we cannot tune the exact insides; instead, we have a process where we can tune some things and mostly hope that they have the effect we are expecting.