Reproduce “Training-Free Predictive Spatio-Temporal Token Pruning for Few-Step Autoregressive Video Diffusion” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jia Li, Xurui Peng, Xing Wang, Hayden Kwok-Hay So
PreSTP accelerates autoregressive video diffusion by pruning tokens that appear redundant across neighboring frames or denoising steps, without retraining. It retains the first frame as a reference and skips updates for static or converged tokens. On five-second Self-Forcing generation, it achieves 1.52× speedup, 83.92 VBench Total, and 0.109 LPIPS; gains vary with content and do not remove historical-context attention.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.