// PERSON
Yu Sun
ROLE RESEARCHERMENTIONS 1LAST SEEN JULY 16, 2026
// BIO
Yu Sun is a researcher at NVIDIA and a postdoctoral researcher at Stanford University, where he is hosted by Carlos Guestrin, Tatsu Hashimoto, and Sanmi Koyejo. He completed his PhD in EECS at UC Berkeley, advised by Alexei Efros and Moritz Hardt, with a thesis on test-time training. He is best known as the originator of the test-time training paradigm, which involves training a different model on-the-fly for each test instance using self-supervision.
// RECENT MENTIONS
// SIGNALS
1 SIGNAL
01
mention·arXiv Physical AI·JULY 16, 2026
“Originator of the TTT paradigm. His foundational work on fast weights updated by gradient descent at test time is the core mechanism RoboTTT builds upon”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.