Yanwei Li

Scaling from 120,000 hours of egocentric video boosts robotic manipulation success rates from 36.1% to 77.8% in zero-shot evaluations.

MLLMs can revolutionize video understanding by integrating watching, remembering, and reasoning into a cohesive framework that addresses long-range dependencies and sparse evidence.

The fragmented field of world modeling can now be unified under a "levels x laws" taxonomy, revealing critical gaps in autonomous model revision and decision-centric evaluation.