News for hackers. Stuff that matters.
Re:Extracting books from production language models (2026) by clbrmbr on Sunday January 11, 2026 @08:13AM (#46575462)
I found that Opus 4 was happy to regurgitate a random paragraph from the latter half of Wealth of Nations that nobody quotes. It was probably only in the training data once.

I was thinking we could use this technique to figure out which books were in / out of the training data for various models. Limitation is having to wrestle with refusals.

Re:Extracting books from production language models (2026) by carshodev on Sunday January 11, 2026 @01:52PM (#46578530)
Why would they filter non copyright material? Who cares if it repeats things that are already public/freely usable and available.