PaperScope
LIVE · 2026-09-29 05:40 UTC

From Scores to Samples: Elastic Forcing for Autoregressive Video Generation

Chi Zhang, Yueyi Liu, Haoyang Shi, Ruichuan An, Haoyu Li, Yuhang Wu, Sen Cui, Miao Liu

Latestcs.CLcs.LGcs.AIcs.CV
arXiv ID
2609.35491 v1
Category
Submitted
2026-09-28

Abstract

Few-step autoregressive video generation commonly relies on Distribution Matching Distillation (DMD), requiring a bidirectional diffusion teacher and an online fake-score model. We instead learn the rollout distribution directly from reference videos, eliminating both score models during post-training. Our framework minimizes maximum mean discrepancy (MMD) in frozen self-supervised video representation spaces, using a hybrid Nyström--Monte Carlo estimator to balance approximation bias and sampling variance. Memory-efficient replay and gradient subsampling make this objective practical. Using the same architecture and initialization as Self-Forcing, our 1.3B model improves the VBench Total score from 83.80 to 84.64 while retaining 17 FPS. Removing auxiliary score models also enables 14B post-training on eight H200 GPUs. Beyond distillation, learning from reference videos enables the acquisition of new visual styles, semantic concepts, and spatial priors without a target-specific diffusion teacher.

arXiv abs page · PDF · same-day batch