MORCA: Offline-to-Online Reinforcement Learning for Adaptive Cache Reuse in Video Diffusion Acceleration

arXiv:2610.10457 · cs.CV · Submitted 2026-10-07 · Read on arXiv

cs.CV

Submitted: 2026-10-07

Updated: 2026-10-07

Comments: 22 pages, 8 figures

Code: https://github.com/x10ngyx/MORCA

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Online Reinforcement Learning — 1× in this paper · explained in 2 episodes
• Reinforcement Learning — 1× in this paper · explained in 4 episodes

Sources

• $\Delta$-DiT: A Training-Free Acceleration Method Tailored for Diffusion Transformers
• HunyuanVideo: A Systematic Framework For Large Video Generative Models
• Wan: Open and Advanced Large-Scale Video Generative Models
• OnlineCache: Learning Dynamic Caching Policies with Error Correction for Efficient Diffusion Inference
• BAG: Budget-Aware Gating for Diffusion Caching

Related papers

• Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
• AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
• Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
• MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
• TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
• A Survey on Efficient Vision-Language-Action Models