用2比特量化压缩视频生成的键值缓存，大幅降低显存占用。
Quant VideoGen: Auto-Regressive Long Video Generation via 2-Bit KV-Cache Quantization
- 通过语义感知平滑与渐进残差量化，减少缓存数据量
- 显存降低7倍，延迟增加不足4%，长视频一致性显著提升
- 无需训练，适合部署在普通硬件上的长视频生成
自回归视频扩散模型面临关键瓶颈：键值（KV）缓存内存占用过高，生成过程中缓存持续增长，常超过30 GB，限制了在主流硬件上的部署。更严重的是，受限的缓存预算直接影响长时视频生成中身份、布局和运动的一致性。为此，我们提出Quant VideoGen（QVG），一个无需训练的自回归视频生成模型的KV缓存量化框架。QVG利用视频时空冗余，通过语义感知平滑生成低幅值、易于量化的残差；进一步引入渐进残差量化，采用粗到细的多阶段策略，在控制量化误差的同时实现质量-内存的平滑权衡。在LongCat Video、HY WorldPlay和Self Forcing基准上，QVG在质量和内存效率之间建立了新的帕累托前沿，最大可将KV缓存内存降低7.0倍，端到端延迟增加小于4%，且在生成质量上持续优于现有基线。代码已开源：https://github.com/svg-project/Quant-VideoGen
原文摘要 · Abstract (English)
Despite rapid progress in autoregressive video diffusion, an emerging system algorithm bottleneck limits both deployability and generation capability: KV cache memory. In autoregressive video generation models, the KV cache grows with generation history and quickly dominates GPU memory, often exceeding 30 GB, preventing deployment on widely available hardware. More critically, constrained KV cache budgets restrict the effective working memory, directly degrading long horizon consistency in identity, layout, and motion. To address this challenge, we present Quant VideoGen (QVG), a training free KV cache quantization framework for autoregressive video diffusion models. QVG leverages video spatiotemporal redundancy through Semantic Aware Smoothing, producing low magnitude, quantization friendly residuals. It further introduces Progressive Residual Quantization, a coarse to fine multi stage scheme that reduces quantization error while enabling a smooth quality memory trade off. Across LongCat Video, HY WorldPlay, and Self Forcing benchmarks, QVG establishes a new Pareto frontier between quality and memory efficiency, reducing KV cache memory by up to 7.0 times with less than 4% end to end latency overhead while consistently outperforming existing baselines in generation quality. Code is available at: https://github.com/svg-project/Quant-VideoGen
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。