SOTA in Open Source Document Understanding

Hello,

I’m looking for current models that can understand documents. There are some existing models like LayoutLMv3, but the latest are from 2022 or even older. Is there no current development?
Amazon published a new version of the DocFormer Paper (v2), but I can’t find an implementation yet.

Hello,
Recently Alibaba released mPLUG-DocOwl 1.5 and I think it’s currently holds SOTA for Document Understanding.
https://huggingface.co/papers/2403.12895

1 Like