1 paper
Jonathan Sadeghi, Jenny Seidenschwarz, Jesse Allardice +3
Video world models approximate the stochastic distribution of physical outcomes through generative sampling, but existing benchmarks score individual generations or compare distrib…