WAM-Cache: Staleness-Bounded KV Reuse for Efficient World Action Models
WAM-Cache: Staleness-Bounded KV Reuse for Efficient World Action Models
WAM-Cache is a training-free framework that retains layerwise key-value representations across chunks and recomputes only a sparse refresh set of tokens. It combines cross-attention from the action expert with visual latent surprise and a strict age bound to suppress compounding error. The method reduces video DiT prefill FLOPs by 32-42% across multiple benchmarks while maintaining high action accuracy.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Environment
- SimulationOpen source S5
Reported numbers
percentage points
0.7–1.8
View original evidence
ing within 0.7-1.8 percentage points of the dense policy in simulation and 2.5 points on a real robot.
Open source S5percentage points
2.5
View original evidence
ing within 0.7-1.8 percentage points of the dense policy in simulation and 2.5 points on a real robot.
Open source S5
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ing within 0.7-1.8 percentage points of the dense policy in simulation and 2.5 points on a real robot.
Open source S5
ownstream action accuracy is instead governed by where the action expert attends, not by what moved. WAM-Cache therefore selects the refresh set by uniting the action expert's cross-attention with visual latent surprise, complemented by a strict age bound that suppresses compounding error. On Fast-WAM, WAM-Cache cuts video DiT prefill FLOPs by 32-42% across RoboTwin 2.0, LIBERO, and real-world experiments, while stay
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org