Chenyu Wang

LLMs can slash compute costs by up to 67% on reasoning tasks with a new online calibration method that adapts to each input at test time.

VLMs that excel at visual understanding can still fail at driving tasks requiring temporal reasoning, revealing an over-reliance on pretrained patterns instead of modeling dynamics.

LLMs can now generate more accurate and complex CAD models by pointing to existing geometric entities, rather than relying on discretized command sequences prone to topological errors.

Generative AI in systems design faces the same five challenges—no matter the layer—and surprisingly, the same five design principles keep emerging as solutions.