What Helps Denoising Can Hurt Memory: Rethinking History in Long Video Generation
cs.CV
Submitted: 2026-10-31
Updated: 2026-10-31
Terminology
Sources
- Recency Forcing: Bridging the Long-Horizon Gap in Autoregressive Video Generation
- Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusion
- Self-Forcing++: Towards Minute-Scale High-Quality Video Generation
- Forcing-KV: Hybrid KV Cache Compression for Efficient Autoregressive Video Diffusion Models
- Rolling Sink: Bridging Limited-Horizon Training and Open-Ended Testing in Autoregressive Video Diffusion
- Attend Locally, Remember Linearly: Linear Attention as Cross-Frame Memory for Autoregressive Video Diffusion
- Rolling Forcing: Autoregressive Long Video Diffusion in Real Time
- Future Forcing: Future-aware Training-free KV Cache Policy for Autoregressive Video Generation
- TetherCache: Stabilizing Autoregressive Long-Form Video Generation with Gated Recall and Trusted Alignment
- In-Distribution Forcing for Long Video Generation at Test Time
- SGF+: Decoupling Gradient Flows for Autoregressive Video Generation
- WorldPlay: Towards Long-Term Geometric Consistency for Real-Time Interactive World Modeling
- Wan: Open and Advanced Large-Scale Video Generative Models
- Head Forcing: Long Autoregressive Video Generation via Head Heterogeneity
- Self-Aligned Forcing: Streaming Video Diffusion with Differentiable Noisy History
- In-Flight KV Cache with Clean Anchors for Faster Autoregressive Video Diffusion
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs