I’m trying to execute a question answering model and for my use case, this model was working fine: https://huggingface.co/distilbert-base-cased-distilled-squad
However, when I tried to implement using: https://huggingface.co/transformers/model_doc/distilbert.html#distilbertforquestionanswering
the solution was different.
Are the ones in demo more trained than the ones I downloaded from here cause I get correct answer for a complex context only when I’m using the online demo. Maybe something else is amiss