We're finding through LLM research that smaller models aren't necessarily dumber, so much as their intelligence is more limited in terms of domains and knowledg

We're finding through LLM research that smaller models aren't necessarily dumber, so much as their intelligence is more limited in terms of domains and knowledge encoding ability. Those small brains might have limited knowledge encoding ability, but that doesn't mean they can't reason about the things they have encoded.