2 ms·
If LLMs from an internal world model, there are definitely things they "believe", and things they can pretend to believe.
by cubefox 3y ago
If LLMs from an internal world model, there are definitely things they "believe", and things they can pretend to believe.
- marci 3y agoIf LLMs have models (tokens), there are definitely tokens inside said models, and a program can manipulate them.