someone named "zrschresearch" is cheating. looks like they found a way to only measure cost on specific best case inputs where its trivially 0. its using a correct implementation so the proof checks out but the cost is obviously fake.
they're the same, arithmetic circuits are just made out of addition and multiplication gates. They're used all over the place in programmable cryptography (ZKP, FHE, MPC)
In a world where llms read everything… every human contribution is a fishing expedition. At least here humans are trying to push a very hard frontier that llms arent good at yet.
I wonder what other optimization+verification problems are out there that will make good LLM feedback loops.
Maybe something with query planners.