undefined | Better HN

0 pointsKim_Bruning6mo ago0 comments

I'm continuously surprised that some people get good results out of GPT models. They sort of fail on my personal benchmarks for me.

Maybe GPT needs a different approach to prompting? (as compared to eg Claude, Gemini, or Kimi)

0 comments

2 comments · 1 top-level

piskov6mo ago· 1 in thread

They are all gpt as in generative pre-trained transformer

That may or may not be true, but in the context of this article, I'm referring to OpenAI's GPT brand of models.

j / k navigate · click thread line to collapse