InOpen IntelligencebyDr. Fadi Shaar·5d agoBeyond Page Limits: How Unlimited OCR 2 Processes Thousands of Pages in One PassHow Baidu’s open-source OCR model uses Sparse Reference Sliding Window Attention to process massive documents while keeping memory…
InAI Exploration JourneybyFlorian June·5d agoInfinity-Parser2: Document Parsing That Goes Beyond Text Recognition to Understand StructureIn a previous article, I introduced the principles behind Infinity-Parser (Rethinking Scanned Document Parsing with Layout-Aware RL — AI…A response icon1A response icon1
Binyang·6d agoThe Long-Overlooked Performance Cost in Document Parsing: Reading OrderWhere you put reading order decides how fast a parser runs and how much you have to label. Here is how OXR handles it.
InTowards AIbyBinyang·Oct 28Errors as Data: How RSI Helps OXR Achieve SOTA Accuracy and Speed in Document ParsingWhat if every document a model gets wrong told you what to train on next? That’s the loop behind OXR.A response icon1A response icon1
Binyang·Oct 26Why We’re Building OXR: From the Documents We Still Get WrongThe next batch of documents may come from an industry we’ve never served. This is how we’re getting OXR ready for it, in the open.A response icon2A response icon2
Mohitkundu·Oct 26Your RAG Pipeline Needs a Delete Button — and a Version Number ?How to design a retrieval pipeline that handles document changes, retries, and revoked access without serving yesterday’s truth.
InAI Exploration JourneybyFlorian June·Oct 16Why Reading a Book Is Still Hard for OCR ModelsMany OCR demos look impressive because they show one page at a time.A response icon1A response icon1
RevExOS·Oct 12I Made My Claude MCP Server Work in ChatGPT. Here’s What I Had to Change.One server, eight quote-to-cash tools, two AI assistants. Attach an invoice in ChatGPT and get the fields, line items and a totals check…
evolveyourself·Sep 29Only for JavaScript developersHey buddy, I know the pain when you need to add document parsing to a JavaScript project. You start looking for one package for PDF…
PyMuPDF·Sep 9The Benchmark Creator Is Almost Always on the PodiumWe aggregated thirteen public OCR and document-parsing leaderboards into a single table. Then, instead of asking the usual question “Which…
