Forcing-KV：用于高效自回归视频扩散模型的混合KV缓存压缩
Forcing-KV: Hybrid KV Cache Compression for Efficient Autoregressive Video Diffusion Models
AI总结：
本文提出Forcing-KV方法，通过混合KV缓存压缩技术，提升自回归视频扩散模型的生成速度和内存效率，实现高达2.82倍的速度提升。
摘要：
自回归（AR）视频扩散模型采用流式生成框架，能够实现长视界视频生成并具有实时响应能力，如Self Forcing训练范式所示。然而，现有AR视频扩散模型仍因历史帧间冗余的键值（KV）缓存导致显著的注意力复杂度和严重的内存开销，限制了可扩展性。本文通过引入自回归视频扩散中的KV缓存压缩来解决这一挑战。我们发现主流AR扩散模型中的注意力头在样本和去噪步骤中表现出明显不同的注意力模式和功能角色。基于我们对头部功能专门化的经验研究，我们将注意力头分为两类：静态头，专注于自回归块之间的转换和帧内保真度；动态头，控制帧间运动和一致性。然后我们提出Forcing-KV，一种混合KV缓存压缩策略，对静态头执行结构化的静态剪枝，对动态头基于片段相似性进行动态剪枝。在保持输出质量的同时，我们的方法在单个NVIDIA H200 GPU上实现了超过29帧每秒的生成速度，并实现了30%的缓存内存减少，最高在480P分辨率下达到1.35倍和1.50倍的LongLive和Self Forcing速度提升，进一步扩展到1080P分辨率时达到2.82倍的速度提升。代码和演示视频可在https://zju-jiyicheng.github.io/Forcing-KV-Page获取。
英文摘要：
Autoregressive (AR) video diffusion models adopt a streaming generation framework, enabling long-horizon video generation with real-time responsiveness, as exemplified by the Self Forcing training paradigm. However, existing AR video diffusion models still suffer from significant attention complexity and severe memory overhead due to the redundant key-value (KV) caches across historical frames, which limits scalability. In this paper, we tackle this challenge by introducing KV cache compression into autoregressive video diffusion. We observe that attention heads in mainstream AR diffusion models exhibit markedly distinct attention patterns and functional roles that remain stable across samples and denoising steps. Building on our empirical study of head-wise functional specialization, we divide the attention heads into two categories: static heads, which focus on transitions across autoregressive chunks and intra-frame fidelity, and dynamic heads, which govern inter-frame motion and consistency. We then propose Forcing-KV, a hybrid KV cache compression strategy that performs structured static pruning for static heads and dynamic pruning based on segment-wise similarity for dynamic heads. While maintaining output quality, our method achieves a generation speed of over 29 frames per second on a single NVIDIA H200 GPU along with 30% cache memory reduction, delivering up to 1.35x and 1.50x speedups on LongLive and Self Forcing at 480P resolution, and further scaling to 2.82x speedup at 1080P resolution. Code and demo videos are provided at https://zju-jiyicheng.github.io/Forcing-KV-Page.
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。
