Yeah, there's not a lot that's actionable here... mostly boils down to "try a lot of stuff yourself and find out what LLMs are good and bad at." Rs in strawberr

Yeah, there's not a lot that's actionable here... mostly boils down to "try a lot of stuff yourself and find out what LLMs are good and bad at." Rs in strawberry, generating some specific text in Nano Banana, what knowledge it knows, etc. Don't do those specific things because obviously (?) models are bad at them.