Jiggly Jar

Introduces AnyFlow, the first any-step video diffusion framework using flow maps that allows a single model to adapt to arbitrary inference steps for high-quality video generation, enabling efficient predictive video modeling for world models.

A comprehensive talk outlining NVIDIA's roadmap to Physical AGI, proposing video world models as a second pretraining stage, introducing World Action Models (WAM) for joint video-action prediction, and describing systems like EgoScale and DreamDojo for robotics scaling.

Introduces World Action Models (WAMs) that jointly predict video and actions via a 14B-parameter autoregressive diffusion model, enabling state-of-the-art zero-shot generalization on manipulation benchmarks like MolmoSpaces and RoboArena through emergent internal world modeling.

TTTTT BBBBB