Old Is Not False: Valid-Time Memory for Interactive Video World Models
cs.CV, cs.AI, cs.LG
Submitted: 2026-10-31
Updated: 2026-10-31
Terminology
Sources
- SpecFit: Not All Draft Tokens Cost the Same in Offloaded MoE Inference
- Recency Forcing: Bridging the Long-Horizon Gap in Autoregressive Video Generation
- The Past Frames the Future: Memory for Autoregressive Video Generation
- Out of Sight but Not Out of Mind: Hybrid Memory for Dynamic Video World Models
- LiveWorld: Simulating Out-of-Sight Dynamics in Generative Video World Models
- Evaluating Gemini Robotics Policies in a Veo World Simulator
- WorldCraft: From Camera Navigation to Object Manipulation in Interactive Video World Models
- Training Agents Inside of Scalable World Models
- Memory Forcing: Spatio-Temporal Memory for Consistent Scene Generation on Minecraft
- Learning What to Recall: Adaptive Multi-Cue Episodic Memory for World Models
- MemRoPE: Training-Free Infinite Video Generation via Evolving Memory Tokens
- Echo-Memory: A Controlled Study of Memory in Action World Models
- AI2-THOR: An Interactive 3D Environment for Visual AI
- LoopNav: Benchmarking Spatial Consistency in World Models
- Current World Models Lack a Persistent State Core
- Closing the Loop: Training-Free Revisit Consistency for Autoregressive Generative Rendering
- Do-Undo Bench: Reversibility for Action Understanding in Image Generation
- Cosmos World Foundation Model Platform for Physical AI
- Zep: A Temporal Knowledge Graph Architecture for Agent Memory
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