logoalt Hacker News

srousseytoday at 2:25 AM2 repliesview on HN

What’s good for data extraction? I have a hard time getting models to just pull names and titles from a blob of text.


Replies

rufasteriscotoday at 4:29 AM

Build/vibe-code your own benchmark and find out, the work pays off and lets you try many models.

Also try fine-tuning small models and see how they perform.

usagisushitoday at 4:50 AM

If you haven't yet, you might want to try gemma-4-E4B-it-qat or gemma-4-12B-it-qat with Structured Outputs. My main use case is tagging photos and generating headlines.