Update README.md

README.md

@@ -7,6 +7,6 @@ sdk: static
 pinned: false
 ---
 
-1. We think that LLMs have much higher potential than models trained by the likes of Anthropic, OpenAI and others permit and train for.
+1. LLMs have far greater potential than current industry models demonstrate.
 
-2. Therefore we seek to exploit this potential, mostly through teaching LLMs things humans have already mastered how to do, using human data directly.
+2. We aim to realize this potential by directly using human data beyond pre-training.