dKV Forcing: Separating Generation and Memory Representations through a Differentiable KV Cache
cs.CV, cs.AI
Submitted: 2026-11-01
Updated: 2026-11-01
Code: https://github.com/ZhiqiLi-CG/dKV-Forcing
Project page: https://zhiqili-cg.github.io/dKV-Forcing-Project
Terminology
Sources
- Training Deep Nets with Sublinear Memory Cost
- Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion
- Rolling Forcing: Autoregressive Long Video Diffusion in Real Time
- Wan: Open and Advanced Large-Scale Video Generative Models
- WorldMem: Long-term Consistent World Simulation with Memory
- LongLive: Real-time Interactive Long Video Generation
- Improved Distribution Matching Distillation for Fast Image Synthesis
- Context as Memory: Scene-Consistent Interactive Long Video Generation with Memory Retrieval
- Connected Self Forcing: Beyond Local Learning in Video Autoregression
- Frame Context Packing and Drift Prevention in Next-Frame-Prediction Video Diffusion Models
- Causal Forcing++: Scalable Few-Step Autoregressive Diffusion Distillation for Real-Time Interactive Video Generation
- Causal Forcing: Autoregressive Diffusion Distillation Done Right for High-Quality Real-Time Interactive Video Generation
- Self Gradient Forcing: Native Long Video Extrapolation
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