Show HN: LLMs think in hierarchies. Their interpretability tools should too

3 points | by visha1v 10 hours ago