subsets for each langpair
#2
by ZJaume - opened
No description provided.
Previous dataset layout was not loading correctly due to different metadata in each language pair. Usually, translation datasets have each language pair in a different subset, which I think it is more convenient, so you can only load one and you don't need to load everything and filter based on JSON metadata. Therefore I moved each language pair into a subset and removed the language identifier columns.
ZJaume changed pull request status to open