Compositional Monte Carlo Tree Diffusion for Extendable Planning

A novel approach to extend the capabilities of Monte Carlo Tree Diffusion beyond trained trajectory lengths