Sublime

Meet Montelimar, an open-source toolbox for on-device OCR ! > https://t.co/3OKJJwkGdk... See more

To understand a new codebase, my workflow: 1. Convert the repo into markdown: https://t.co/BITB23Kwkb 2.... See more

We've been working on extracting information in templatized PDFs for the last couple of years, leveraging the best of LLMs and classical data extraction techniques. Our latest technique, TWIX, has the best of all worlds: beats Azure DI, AWS Textract, or LLM-based approaches by over 25% in precision AND recall, and is m... See more

have taken to git-cloning open source projects, concatenating all their documentation into a giant blob, then pasting into o1-pro almost invariably better than trying to scroll through the actual docs pages, asking on discord, etc