logoalt Hacker News

thomasfromcdnjstoday at 6:48 AM3 repliesview on HN

I've had a bit of fun working with low resource languages (aboriginal australian), and enjoying the result from Facebook's No Language Left Behind project -> https://huggingface.co/facebook/nllb-200-distilled-600M

I'd recommend giving it a squiz. (I assume Amish has a large corpus)


Replies

trollbridgetoday at 9:52 AM

“Amish” isn’t a language but Pennsylvania Dutch has plenty of written material, although most speakers of it prefer to write in English with the occasional person who prefers German.

I live adjacent to a few thousand speakers of it and I doubt there is a single person over the age of 8 who can’t speak English fluently.

Due to the lack of a standard orthography don’t expect LLMs to do anything remotely usable other than generate a few laughs.

cactaceatoday at 4:59 PM

> (I assume Amish has a large corpus)

Pennsylvania Dutch does not, it is primarily oral and the Amish generally don't allow themselves to be photographed or recorded.

show 1 reply
lproventoday at 10:38 AM

"squiz"?