News for hackers. Stuff that matters.

A comment on Teaching Large Language Models to Self-Debug. Parent comment

Re:Teaching Large Language Models to Self-Debug by mirashii on Wednesday April 12, 2023 @05:33PM (#35547287)
> 100% accuracy up to about 13 digit addition

The graphs you just posted do not support that, they'd support at most 100% accuracy up to 4 digits.

[ Parent | on HN ]
Re:Teaching Large Language Models to Self-Debug by og_kalu on Wednesday April 12, 2023 @05:48PM (#35547477)
It's 100 at 13 and extremely close to it prior to that. Maybe basically 100 is better.
[ Parent | on HN ]
Re:Teaching Large Language Models to Self-Debug by sharemywin on Wednesday April 12, 2023 @06:32PM (#35547962)
it's GPT so 13=4
[ Parent | on HN ]