Tollef J

Why are there so few languages involved in the training of these models? You argue that this data mix was selected "to create a corpus of European and most widely spoken languages, representing a broad range of alphabets and cultures."
But what is the relevance in other alphabets when, for example, you do not include any Nordic languages with large and high-quality datasets?

Prefixing it "Euro" seems odd in this context. You have selected a tiny fraction of languages - so name it accordingly :-)
It would also make sense to refer to EuroEval https://euroeval.com/leaderboards/