On: Apertus – Open Foundation Model for Sovereign AI

[8note]

it probably doesnt know what language each set of words is referencing.

i doubt they are including a lot of training data labeled with the language.

"how to say X in language Y" is a different task from saying X in language Y

[einpoklum → 8note]

Actually, it isn't all that different. There are only two words separating "how to say X in language Y" from "say X in language Y". And this "vulgar" metric is actually quite relevant for an LLM, which answers based on conversational context.