[vunderba]

Isn't this just a much larger "prompt based equivalent" of chain-of-reasoning systems like Qwq?

https://qwenlm.github.io/blog/qwq-32b-preview

[throwaway314155]

Yes, which is discussed several times in the article.