We Found an Infinite Benchmark Glitch

What will happen when we have an LLM benchmark that will never be saturated by LLM, ever?

A project by Kreasof AI answers this need by providing an infinitely harder benchmark with N-dimensional Grid Regression, inspired by the ICLR (In-Context Learning of Representations) paper.