ReelCache: An Efficient KV Cache Compression and Management System for Multi-Round Video Generation
cs.CV, cs.DC
Submitted: 2026-10-31
Updated: 2026-10-31
Code: https://github.com/Tianyu9748/ReelCache
Terminology
Sources
- PyramidKV: Dynamic KV Cache Compression based on Pyramidal Information Funneling
- LongLive-2.0: An NVFP4 Parallel Infrastructure for Long Video Generation
- USP: A Unified Sequence Parallelism Approach for Long Context Generative AI
- xDiT: an Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelism
- DeepSpeed Ulysses: System Optimizations for Enabling Training of Extreme Long Sequence Transformer Models
- MemFlow: Flowing Adaptive Memory for Consistent and Efficient Long Video Narratives
- GEAR: An Efficient KV Cache Compression Recipe for Near-Lossless Generative Inference of LLM
- LMCache: An Efficient KV Cache Layer for Enterprise-Scale LLM Inference
- KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache
- Advancing Open-source World Models
- MAGI-1: Autoregressive Video Generation at Scale
- ShadowKV: KV Cache in Shadows for High-Throughput Long-Context LLM Inference
- Seedance 2.0: Advancing Video Generation for World Complexity
- Wan: Open and Advanced Large-Scale Video Generative Models
- Communication-Efficient Serving for Video Diffusion Models with Latent Parallelism
- LongLive: Real-time Interactive Long Video Generation
- vLLM-Omni: Fully Disaggregated Serving for Any-to-Any Multimodal Models
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