←back to thread

261 points david927 | 1 comments | | HN request time: 0.252s | source

What are you working on? Any new ideas that you're thinking about?
1. bambax ◴[] No.43159982[source]
I'm making an OCR website focused on outputting ascii text that follows the layout of the original, so that it doesn't need to understand or interpret zones in the source: it just resembles the source. This makes proofing easier and should also improve feeding documents to LLMs.