Hacker News Reader

Anyone have recommendations for chip datasheets? Ive explored a couple options so far, but getting some bitfields wrong is super annoying.

I see plenty of examples like the one here that are on easier extractions. A PDF to HTML or Markdown converter will probably get it right with OCR.

View on HN

No direct recommendation for that use case, but one strategy I've heard being used and that works with complex documents (or where hallucinations are Very Bad™ - like invoice processing) is using multiple techniques and models at once in a quorum approach. For example, direct ingestion of PDFs into Gemini, OCR and ingestion of text, plus perhaps using another model like GPT. If they all agree on a fact, you're (probably) good. If not, it can be bumped up to human correction.

Reply on news.ycombinator.com