Monte Carlo

CARV cuts diffusion-teacher gradient variance

CARV reduces Monte Carlo variance in diffusion-teacher pipelines by reusing expensive upstream work and smarter noise sampling.