[bulbar]

Training of an LLM however is a lossy compressing algorithm to provide a copy of a variant of the data to the user later on.