[a0zU]

They don't even use more methodological evaluations of fine-tuned LLMs, they use metrics that are specifically built to support a (false) contrarian conclusion in order to generate attention for their "paper."