Fetching the paper…

MMLU-Pro+: Evaluating Higher-Order Reasoning and Shortcut Learning in LLMs · Around