Show HN: I built "AI Wattpad" to eval LLMs on fiction

Show HN: Evaluating LLMs on creative writing via reader usage, not benchmarks