[flir]

> You know what you know from nothing at all.

I've done it myself. Transcription of 19th century newspaper articles to markdown, mostly. Some earlier wills (which were an absolute pig - secretary hand). Oh, and categorisation of postcards. That's why I was poking around your pipeline - to see if I could learn anything. You're right, tabular data is hard. Also columns, and proper nouns.

Feeding the LLM a context-aware cheat sheet helped with the nouns. BTW, what I said about using multiple models for parallax was good advice.

Do you know you're very spiky?

[sweisman]

The biggest problem with hallucinations (or confabulations as Claude prefers haha) is from tables. Prose has context that anchors it.

Hebrew is also much worse than Latin, which uses Arabic numerals. Hebrew conventionally uses the letters for numeric representation too.

I am getting good (and improving results), but the token cost is heavy.

[sweisman]

No, because I don't know what spiky means.