Preprint
Aug 2026
ChebBooster: A Training-Free Approach for Efficient Diffusion Transformer Inference via Chebyshev-Inspired Extrapolation
CebBooster is proposed, a training-free extrapolation framework based on Chebyshev polynomial theory that achieves stable and efficient acceleration for DiTs and outperforming existing training-free baselines under diverse generation tasks and resolutions.
Cheng-Jie Lu, Tianchi Deng, Zheng He et al.
· 0 citations