OPSD-V：用于训练后少步自回归视频生成器的策略内自蒸馏
OPSD-V: On-Policy Self-Distillation for Post-Training Few-Step Autoregressive Video Generators
AI总结：
研究针对训练后少步自回归视频生成器存在的问题，提出OPSD-V策略内自蒸馏范式。通过引入真实长视频数据提供监督，学生和教师模型按特定方式运行，应用该范式于相关模型后，实验及用户研究表明其在多方面有改进且更受青睐。
摘要：
我们提出了OPSD-V，一种用于训练后少步自回归（AR）视频扩散模型的策略内自蒸馏范式。现有的少步AR视频生成器能以低延迟生成长视频，但在长时间自回归展开过程中仍存在误差累积和运动动力学减弱的问题。OPSD-V减少了长时退化，同时保留了原始的少步推理路径。关键在于在训练期间引入真实长视频数据作为时间上下文，并利用它提供密集的轨迹级监督。具体而言，学生模型遵循推理时的展开方式，根据自身先前生成的KV缓存生成每个块。同时，教师模型在相同的去噪状态下进行评估，但使用更干净的与AR一致的时间缓存，其中旧历史可以被真实视频上下文替换。这在策略内AR缓存动态下提供了密集的去噪级校正目标，而不改变采样器、去噪步数或推理时的缓存机制。我们将OPSD-V应用于代表性的少步AR视频模型，包括Self-Forcing和LongLive。实验表明在视觉质量、运动动力学和VBenchLong分数方面有持续改进。一项有10名参与者比较20对视频的用户研究表明，在66.0%的总体偏好判断中（排除平局后为82.5%），OPSD-V比基础模型更受青睐。
英文摘要：
We propose OPSD-V, an on-policy self-distillation paradigm for post-training few-step autoregressive (AR) video diffusion models. Existing few-step AR video generators can produce long videos with low latency, but still suffer from error accumulation and weakened motion dynamics during long autoregressive rollout. OPSD-V reduces long-horizon degradation while preserving the original few-step inference path. The key idea is to introduce real long-video data as temporal context during training and use it to provide dense trajectory-level supervision. Specifically, the student follows the exact inference-time rollout, generating each chunk conditioned on its own previously generated KV cache. In parallel, the teacher is evaluated at the same student-visited denoising states, but uses a cleaner AR-consistent temporal cache in which older history can be replaced by real-video context. This provides dense denoising-level corrective targets under on-policy AR cache dynamics, without changing the sampler, number of denoising steps, or inference-time cache mechanism. We apply OPSD-V to representative few-step AR video models, including Self-Forcing and LongLive. Experiments show consistent improvements in visual quality, motion dynamics, and VBenchLong scores. A user study with 10 participants comparing 20 video pairs shows that OPSD-V is preferred over the base models in 66.0% of overall-preference judgments (82.5% excluding ties).
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
