@scaling01 Questions Training Models on Games

@scaling01, who runs LisanBench, a custom LLM reasoning benchmark, posted a reply stating that training models on certain games may be unwise. The commentator added that models are likely too smart and already recognize when they are inside a game. The post includes an attached 2x2 grid of photographs showing a young man with short blond hair in dark clothing and a young woman with curly light-brown hair.