MedConstBench: Benchmarking How Constitutions Steer LLMs in Medical Ethics.

arXiv:2610.36513v1 Announce Type: new Abstract: As large language models advance rapidly, they are increasingly used to answer medical questions, and many such models are now deployed locally. Once a model is deployed, its behavior is governed mainly by the et