Quant VideoGen: Auto-Regressive Long Video Generation via 2-Bit KV-Cache Quantization
This addresses a critical system bottleneck for deploying autoregressive video diffusion models on widely available hardware, with incremental improvements in efficiency.
The paper tackles the KV cache memory bottleneck in autoregressive video generation models, which limits deployability and long-horizon consistency, by introducing Quant VideoGen (QVG), a training-free KV cache quantization framework that reduces memory by up to 7.0 times with less than 4% latency overhead while improving generation quality.
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.