But isn’t that what “training” is anyway? They train LLM today like that and the database becomes the parameters. You can post train on smaller corpus for purpo

But isn’t that what “training” is anyway? They train LLM today like that and the database becomes the parameters. You can post train on smaller corpus for purpose-built stuff.