Preprint
Oct 2026
Tikhonov-Stabilized Bezier Representation Forecasting for Training-free Diffusion Acceleration
BeziCast is proposed, a training-free diffusion acceleration framework that forecasts output-proximal denoising representations using low-order B'ezier trajectories using Tikhonov-stabilized fitting, providing a smooth, capacity-controlled temporal parameterization that avoids explicit high-order derivative estimation...
Lei Zhu, Mu-Jie Lin, Ruo-Chong Zheng et al.
· 0 citations