@feudjais On peut faire plus graphique en un seul caractère unicode, qui doit bien se taguer, à mon avis… Le #hangeul, c’est l’alphabet coréen, n’est-ce pas. https://fr.wikipedia.org/wiki/Hangeul?wprov=sfti1
Apart from word tokenization having to be adapted for Korean grammar, it's quite interesting how important understanding social and cultural context is to AI.
[Hello Hangeul] Race heats up for Hangeul AI
https://m.koreaherald.com/view.php?ud=20230618000149
Um. Yes, please. This will be a fun addition to my other #한글 studies. I've already watched two of those shows!
https://uhpress.hawaii.edu/title/learn-korean-through-k-dramas/