Document parsing

somarkai introduces OXR as posts highlight PDF formatting complaints

somarkai introduced OXR as an open-source document-parsing model, claiming leading results on olmOCR-Bench, OmniDocBench and DocParsingBench and the fastest parsing among open-source models. The announcement describes a 1.3B model that turns documents into data for agents.

Separate posts described frustrations with document conversion: olivvrszn called for PDF-to-Word conversion that preserves formatting, while aka_ssy singled out broken tables and paragraphs. AI_Workflow_Hub described OXR's training workflow as repeated parsing, VLM review of errors and uncertainty, clustering of issues and retraining on targeted data without human annotation. These posts describe parsing capabilities and formatting frustrations without demonstrating a PDF-to-Word repair.