Wow - Best Document Image Question Answering Model?

#2
by awacke1 - opened

This is an amazing model and demo. Trying it with Q&A and images I see it is very accurate in extractive answers. This could save humans lots of time in extracting answers from complex documents. Thanks for amazing demo!!

Thanks a lot ๐Ÿ˜„ note that Transformers do scale very well, so increasing the size of the model and size of the dataset might give huge bumps in performance.

This one is a nice use case to show that Transformers really can learn anything :)

Sign up or log in to comment