Most foundational interpretability skills can be learned by analyzing pretrained models with lightweight experiments, modest hardware, and practical workflows rather...