HuggingFace Daily Papers（社区热门论文）· HuggingFace Daily Papers（社区热门论文）·· 2026-06-24精选AI 评分70
Causal-rCM：自回归视频扩散蒸馏的统一教师强制与自强制开源方案
AI 导读
Causal-rCM将扩散蒸馏框架rCM扩展至自回归视频扩散，提出教师强制(TF)与自强制(SF)互补训练范式，并发布统一开源算法与基础设施。通过自定义掩码FlashAttention-2 JVP内核，首次实现基于教师强制的连续时间一致性模型(sCM/MeanFlow)用于自回归视频扩散，收敛速度比离散时间版本快10倍。蒸馏后的2步因果Wan2.1-1.3B模型在仅1或2步采样下取得VBench-T2V分数84.63，仅使用合成数据即达到帧级和块级流式视频生成SOTA。该方法还应用于Cosmos 3全模态世界基础模型，实现动作条件生成的交互式世界模型。
推荐理由
把自回归视频扩散蒸馏到1-2步采样，VBench冲到84.63，这个配方让实时视频生成和交互世界模型从论文走进了工程落地，做视频产品的该看。
正文
Abstract:Autoregressive video diffusion with causal diffusion transformers has emerged as a major paradigm for real-time streaming video generation and action-conditioned interactive world models. In this work, we extend rCM, an advanced diffusion distillation framework, to autoregressive video diffusion. The core philosophy of rCM lies in the complementarity between forward and reverse divergences, represented by consistency models (CMs) and distribution matching distillation (DMD), respectively, in diffusion distillation. This philosophy naturally carries over to the autoregressive setting, where teacher-forcing (TF) provides an offline, forward-divergence causal training paradigm, while self-forcing (SF) corresponds to an on-policy, reverse-divergence refinement.
Our contributions are: (1) through extensive experiments, we show that teacher-forcing CM is currently the best complement to self-forcing DMD as an initialization strategy (2) we present the first implementation of teacher-forcing-based continuous-time CMs (e.g., sCM/MeanFlow) for autoregressive video diffusion, enabled by our custom-mask FlashAttention-2 JVP kernel, achieving 10$\times$ faster convergence compared to discrete-time CMs (dCMs) (3) we introduce Causal-rCM, a leading, unified, and scalable algorithm-infrastructure open recipe for diffusion distillation and causal training (4) we achieve state-of-the-art streaming video generation performance in both frame-wise and chunk-wise settings, using only synthetic data for training.
Notably, our distilled 2-step causal Wan2.1-1.3B model achieves a VBench-T2V score of 84.63 with only 1 or 2 sampling steps. We further apply Causal-rCM to Cosmos 3, an advanced omnimodal world foundation model for physical AI with action-conditioned generation capability, enabling an interactive world model.
| Comments: | Technical Report |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) |
| Cite as: | arXiv:2606.25473 [cs.CV] |
| (or arXiv:2606.25473v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2606.25473 arXiv-issued DOI via DataCite |
Submission history
From: Kaiwen Zheng [view email]
[v1]
Wed, 24 Jun 2026 06:58:02 UTC (2,753 KB)
来源：HuggingFace Daily Papers（社区热门论文） · arxiv.org