Docling runs out of memory when converting a pdf

[marfago]

Docling fails to convert a pdf claiming all the available memory and eventually running OOM.
OS: Win11 Home

docling 2.15.1
docling-core 2.14.0
docling-ibm-models 3.1.2
docling-parse 3.1.0

from docling.document_converter import DocumentConverter

if __name__ == '__main__':
    print(DocumentConverter().convert("https://arxiv.org/pdf/2209.07650v1"))

[dolfim-ibm]

@marfago can you please provide a few more details on your system? How much memory? Do you know if the OOM was from the main system or the GPU?

[marfago]

@dolfim-ibm My system runs with 48GB of RAM on Win Home 11 and It fails the same way with CPU or CUDA. With CUDA it seems that the GPU is never really used, so maybe it is unrelated.

Please let me know what other details you need.

[dolfim-ibm]

We definitely have Docling running on systems with just a few GB of RAM. We need to think of a few other inputs which could help debugging your issue. Will keep you posted.

[marfago]

Please try the simple code above. My issue does not happen with all the PDFs, but only with https://arxiv.org/pdf/2209.07650v1.