Datasets:
Chinese locale setting
Currently, the Zh slice seems to be composed of traditional chinese.
is there a way to retrieve other versions of chinese?
The best way for access and storage would be like this
zh-Hans
zh-Hant
zh-CN
zh-TW
etc.
but if this is not feasible, atleast the first two would be valuable
Yes, the current zh subset names seem weird to me too, but I don't think I can do anything about them: that's how Wikipedia structured their encyclopedias and where people have been contributing articles 🤷♂️. In turn, this determines the dumps.
The reference page List of Wikipedias allows you to sort by language, and you can check that there are only 4 chinese locales: zh, zh-classical, zh-min-nan, zh-yue, with their corresponding encyclopedia pages: https://zh.wikipedia.org/wiki/, https://zh-classical.wikipedia.org/wiki/, etc.
Does that answer your question?
yes. I have done some indepdenent investigation and i agree that there is nothing you can do on your end , at least easily or intuitively.
I am currently trying to find downstream solutions, and if i find any that would be relevant to your codebase i'll bring it to your attention
Thanks! I leave it up to you to close this issue if you think an upcoming proposal would need a different issue.
Cheers