# ViRDM：无需教师与判别器的少步因果视频生成表示分布匹配

- 来源：HuggingFace Daily Papers（社区热门论文）
- 发布时间：2026-09-24 08:00
- AIHOT 分数：36
- AIHOT 链接：https://aihot.news/items/cmuge8kyg0pahrogvjagsfrq2
- 原文链接：https://arxiv.org/abs/2609.28923

## AI 摘要

ViRDM 提出一种无需教师模型和在线判别器的视频后训练方法，将表示分布匹配（RDM）迁移到少步因果视频生成，通过随机截断 clean-exit 监督、轻量 VAE 解码器和分阶段 vector-Jacobian 乘积解决显存瓶颈。仅需 20 次生成器更新，该方法在官方 VBench 上达到 84.87 分，超过此前最佳少步因果基线 0.36 分，训练消耗 16 A100 GPU-hours。

## 正文

Few-step autoregressive (AR) video diffusion enables low-latency streaming generation, but existing post-training methods predominantly rely on Distribution Matching Distillation (DMD), requiring both a large pretrained teacher and an online critic to estimate distributional discrepancies through diffusion scores. In this work, we ask whether this resource-intensive teacher--critic stack can be eliminated by post-training only the generator against a precomputed target distribution. Drawing inspiration from representation distribution matching (RDM) for one-step image generation, we systematically study its transfer to few-step causal video generation and identify three key barriers: a memory-intractable gradient path, a distinct video optimization regime, and representation distributions that underconstrain temporal dynamics.

We introduce ViRDM, a teacher- and critic-free video post-training recipe that addresses these barriers sequentially. By coupling RDM with stochastically truncated clean-exit supervision, a lightweight VAE decoder, and staged vector--Jacobian products, ViRDM makes representation distribution matching memory-feasible for multi-step causal video rollouts. We further establish effective generated-population and initialization regimes for video RDM, and introduce lightweight dynamics regularization to compensate for the underconstrained temporal dynamics.

ViRDM turns three-network distillation into generator-only post-training, reducing GPU memory use and training time while improving video quality. With only 20 generator updates, the recipe reaches 84.87 on the official VBench evaluation, outperforming the previous best few-step causal baseline by 0.36, while requiring 16 A100 GPU-hours. We additionally report exploratory results demonstrating the potential of the same recipe for lower causal sampling budget and for one-, two-, and four-step bidirectional generation.