We propose black-box tests to detect harmful memorization in foundation models trained on structured EHR data. Validated on a public model, our toolkit supports privacy audits by distinguishing generalization from privacy-compromising memorization.

We learn non-gradient field dynamics by solving Schrödinger Bridge problem with non-zero reference process drift

State-of-the-art protein language models, under standard usage, cannot infer evolutionary relationships so we introduce Phyla, a model explicitly trained for evolutionary reasoning that achieves state-of-the-art phylogenetic performance.