Use of Greek Bert for Question Answering

by Evaki - opened


We are trying to experiment on Question Answering in Greek. We are very new to all this, as this is not our actual field, so please forgive me if I say something totally irrelevant.

From my understanding what I should do is to run but instead of using Squad dataset , I should use a dataset containing greek, such as XNLI dataset or the bert-base-greek-uncased-v1 dataset.

However , the structure of Squad dataset is different in comparison to other two and i am confused on this.

Could you please give some instructions on how , if possible, I could use the bert-base-greek-uncased-v1 dataset for Question Answering? Is this possible, or a new greek dataset structure must be created?

Thank you in advance!

Sign up or log in to comment