长时程自回归视频生成的展开边缘蒸馏
Rollout-Marginal Distillation for Long-Horizon Autoregressive Video Generation
- The Chinese University of Hong Kong（香港中文大学）
- Tencent AIPD（腾讯AIPD）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
针对自回归视频生成中长时程误差累积问题，提出Rollout-Marginal蒸馏，通过独立片段评分并辅以视频级蒸馏，在远超训练时域保持高视觉质量。
AI中文摘要：
自回归（AR）视频扩散技术能够实现低延迟、可流式传输的视频生成，但预测误差往往会在长时间展开过程中累积。让生成器在其自身的展开轨迹上训练，使其暴露于这些不完美的历史状态中。然而，现有的视频级分布匹配蒸馏（DMD）方法对整个展开轨迹进行联合评分。由于一个片段与其过去和未来的内容一同被评估，其修正可能倾向于匹配周围上下文中的伪影，仅仅为了保持时间一致性。为了提供更清晰的视觉质量信号，我们引入了展开边缘蒸馏（RMD）。RMD保留生成的过去历史用于AR预测，但每个片段独立地针对一个片段教师模型进行评分，确保其质量修正不受不完美时间上下文的干扰。为了弥补独立片段评分中缺乏时间上下文的问题，RMD随后应用视频级DMD来恢复时间连贯性。大量实验表明，RMD在远超其训练时域的情况下仍能保持高视觉质量，并优于视频级DMD基线。代码和视频结果可在该HTTPS URL获取。
英文摘要：
Autoregressive (AR) video diffusion enables low-latency, streamable video generation, but prediction errors often accumulate over long rollouts. Training the generator on its own rollouts exposes it to these imperfect histories. However, existing video-level distribution matching distillation (DMD) scores the whole rollout jointly. Because a chunk is evaluated together with its past and future, its correction can favor matching artifacts in the surrounding context merely to preserve temporal consistency. To provide a clearer visual-quality signal, we introduce Rollout-Marginal Distillation (RMD). RMD retains the generated history for AR prediction but scores each chunk independently against a chunk teacher, ensuring its quality correction is not compromised by an imperfect temporal context. To compensate for the lack of temporal context in independent chunk scoring, RMD subsequently applies video-level DMD to restore temporal coherence. Extensive experiments demonstrate that RMD maintains high visual quality far beyond its training horizon and outperforms video-level DMD baselines. Code and video results are available at https://cjeen.github.io/RMD
