← sources

Lightest way to get a few thousand pdf into markdown on a mini PC with no GPU?

source: reddit (reddit:selfhosted)published 2026-10-31 10:06ingested 2026-11-01 02:38

retained text

looking for a parser that can live on an old mini pc (16gb and no gpu) and turn a pile of files into markdown for a local model and search

whats in the pile:

\- born digital manuals, the easy part

\- statements that are just pictures of pages

\- some docx and xlsx mixed in

docling comes up the most, but its layout and table models are a lot to ask from a cpu. liteparse skips the layout models and does plain text extraction plus tesseract so it should be quick, the open question is table

document fetched 2026-11-01 00:28 · canonical https://reddit.com/r/selfhosted/comments/1xk6djq/lightest_way_to_get_a_few_thousand_pdf_into