Several people have pointed out that it's inefficient for LLMs to *also* encode facts — build a good language model, and hook it up to Google/wikipedia/etc directly.But... has anyone demonstrated…

Posters: connerdelights. connerdelights

2 likes recorded by the source corpus.

Topics: AI & LLMs

View Post: https://x.com/connerdelights/status/1598011256971358208