Out-of-Distribution Generalization
Building models that generalize beyond training distributions.
Projects, people & papersResearch
Building machine learning systems that understand, reason, and generalize.
We study how learning systems represent knowledge, reason over complex inputs, and remain dependable when the world differs from their training data.
Themes
Research Areas
Large Vision-Language Models
Compositional understanding and visual reasoning.
Projects, people & papersInterpretability
Tracing and explaining computations inside modern models.
Projects, people & papersLarge Language Models
Reasoning, agency, safety, and scientific discovery.
Projects, people & papersCompositional Text-to-Image Generation
Faithfully composing objects, attributes, and relations.
Projects, people & papersOngoing
Active Projects
No active projects listed.