[06:51:07] <wm-bb>	 <olivneh> https://ai.facebook.com/research/no-language-left-behind/ is interesting
[09:03:36] <wm-bb>	 <Jandit> I also stumbled upon it, but I could not yet find out how their approach deals with the problem that for many languages, there is not much (digital) material to train an algorithm on (at least this is what I heard)
[18:01:59] <bd808>	 @Jandit: https://github.com/facebookresearch/fairseq/tree/nllb/examples/nllb/data has some info on the training datasets they used.