Hard problems that reduce to document ranking via noperator.dev
Noted From:
I like the idea of using LLMs to rank documents as opposed to generating facts. Ranking is inherently statistical but will still produce a correct answer when applied recursively and correctly.