mrm8488

/

distill-bert-base-spanish-wwm-cased-finetuned-spa-squad2-es

like 45

Hi!
I would like to know about usage in case the answer is not present in the context. I read that the squad2.0 model (used in this project) has both questions with and without answer.
But I test the model with several examples and even with data used for training (present in the squad2.0 dataset), it always responses something.
What am I doing wrong? How can I get a non-answer as the correct answer?
Thank you! Great job