Open silvioprog opened 7 months ago
Hey @silvioprog, the project has no true-casing support. This should be relatively simple to implement, and I am optimistic that the models can handle this quite well. I have two ideas how this feature could be implemented: A) You could add a second classification head, that classifies if the token starts in upper case or not. B) Or you can change the classification head to multi label classification and add an "upper case" class.
Best, Oliver
Hey.
Thanks a lot for this project!
So, I have a question. Do we have true-casing (capitalization) support in
fullstop-deep-punctuation-prediction
? If not, what do you recommend to solve it?TIA for any help!