run DocumentConverter on GPU

[MHDataExpert]

I am very grateful that you created this amazing tool and made it accessible to everyone. How can we run this function on the GPU using DocumentConverter?

[cau-git]

@Stdev95 By default, Docling already uses a GPU when detected on most models (TableFormer, EasyOCR). You don't need to enable anything specific. We have work-in-progress on complete GPU support for all models, including the layout model (see DS4SD/docling-ibm-models#50) and will inform on our README and Changelog when it is available.

[MHDataExpert]

We are very thankful to you for creating such an amazing tool. If I do not need to use do_ocr, do_table_structure, or do_cell_matching and only want to extract the structure of scientific articles, what hardware configuration do I need to process 10 scientific articles per minute?