WikiResearch on Nostr: "Beyond Aesthetics: Cultural Competence in Text-to-Image Models" with CUBE, a ...

"Beyond Aesthetics: Cultural Competence in Text-to-Image Models" with CUBE, a benchmark built out of @Wikidata, to evaluate cultural awareness and cultural diversity of text-to-image models.

(Kannen et al, 2024)

paper: https://t.co/jUDgO4wwge
data: https://t.co/qx32pJ8sKC https://t.co/LlcZ7G4iZB

via https://twitter.com/WikiResearch/status/1814226672360505484