Forget what it eats to continue improving. Realize what it already has. A foundational language model with no additional training is already quite powerful.

Forget what it eats to continue improving. Realize what it already has. A foundational language model with no additional training is already quite powerful. And that genie isn't going back into the bottle.