A Controllable Examination for Long-Context Language Models | Read Paper on Bytez