#multi-problem evaluation

REST is a new evaluation framework that tests large reasoning models with multiple problems simultaneously, revealing performance drops and providing deeper insights into their real-world multitasking abilities.