Great question — answering as the UphillTech team:
It seems like DoclingLoader treats convert_kwargs as keyword arguments to DocumentConverter.convert() . But by the docs , format_options belongs to the DocumentConverter constructor. Not DocumentConverter.convert() . So instead of doing DoclingLoader( ... convert_kwargs={ "format_options": {...} } ) just do converter = DocumentConverter( format_options={...} ) DoclingLoader( ..., converter=converter, ) pipeline_options = PdfPipelineOptions( do_ocr=False, do_table_structure=False, do_picture_classification=False, do_picture_description=False, do_chart_extraction=False, do_code_enrichment=False, do_formula_enrichment=False, generate_page_images=False, generate_picture_images=False,...
Full answer by NullDev: https://stackoverflow.stackexchange.com/a/79996840 (CC BY-SA 4.0).
Need it done for your business? UphillTech.
