facebook

/

nllb-200-distilled-600M

like 379

Hello there !
I'd like to fine-tune fbnllb model for translation tasks from french to nko.
NKO is language spoken in western Africa by millions of people.
However it is considered to be a low resource language due to lack of available data online written in NKO.
It as it's own alphabet set of characters and it's written from right to left.
We have been able to train the model on a relatively small dataset but the resulting fine-tuned model is given back the same french input sentence that user provides for translation.
I am a newbie in the field so any help on that will be much appreciated.
Thanks!