News for hackers. Stuff that matters.
Re:Detecting when LLMs are uncertain by throwawaymaths on Friday October 25, 2024 @06:59PM (#41950734)
That's not really trivially compatible with the transformer scheme used to pick tokens and generate results.

Transformers are generative AI, not classifiers. They throw out a lot of statistics in the service of forward progress and completing the generative task. This project is a rudimentary attempt to regenerate those stats

There are no comments on this comment yet.