Hi,
I was looking to add Komi-Zyrian to Stanza, Stanford's NLP toolkit trained on UD. There's a newly released UD treebank which should be large enough to make some useful models. However, the fasttext vectors on 157 languages for Komi seem to be of rather low quality.
I found this paper:
https://hal.science/hal-01856178/document
In this paper, you mention retraining Fasttext word vectors for Komi. I can't find them on this repo, though. There are some links under "Pretrained monolingual ...", but those links are currently dead.
Are word vectors for Komi still available somewhere?
Thanks!
Hi,
I was looking to add Komi-Zyrian to Stanza, Stanford's NLP toolkit trained on UD. There's a newly released UD treebank which should be large enough to make some useful models. However, the fasttext vectors on 157 languages for Komi seem to be of rather low quality.
I found this paper:
https://hal.science/hal-01856178/document
In this paper, you mention retraining Fasttext word vectors for Komi. I can't find them on this repo, though. There are some links under "Pretrained monolingual ...", but those links are currently dead.
Are word vectors for Komi still available somewhere?
Thanks!