Abstract

Abstract: Why don’t models generalize? Can we characterize this issue in current architectures like transformers or diffusion models? Do we have ideas on a path forward? In this talk, I’ll plan to discuss a few papers we have in this line of work and also fail to truly answer your questions.

Bio: Yonatan Bisk is an assistant professor of computer science at Carnegie Mellon University – Language Technologies Institute and Robotics Institute (courtesy). He received his PhD from the University of Illinois at Urbana-Champaign working on unsupervised Bayesian models of linguistic syntax.

He runs the CLAW lab, Connecting Language to Action and the World. His group works on grounded and embodied language and communication, placing perception and interaction as central to how language is learned and understood. He has held appointments at USC ’s ISI (working on grounding), the University of Washington (for commonsense research), Microsoft Research (for vision+language), and Meta Inc (for Embodied AI).