Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Contrary to benchmarks, open weight models are way behind the frontier.


My point is that you don't want a big model for the kind of analysis being discussed here

Even if they were paying frontier prices they would be choosing 5 mini or nano with no thinking

At that point, a fine tuned open source model is going to be on the pareto frontier




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: