cs.RONov 15, 2026

Cache or Denoise Less? Residual Reuse in a Flow-Matching World Action Model

Authors: Abhinav Gupta, Nissan Pow

Organizations: Independent Researcher

Abstract

Flow-matching robot policies generate each action chunk through repeated network evaluations, one per denoising step. Cross-cycle caches such as C3^3ache reuse computation from earlier replanning calls, but whether they beat simply taking fewer steps is unclear. We compare the two in Fast-WAM, a 6B-parameter world action model, on LIBERO-Goal. Two uncached steps keep the success of ten: on all 500 initial states they are non-inferior within two percentage points, for one checkpoint and noise seed. Under PyTorch's default compile mode, they also cut a replanning call on an RTX A6000 from 202 to 92 ms. This leaves little room for caching. A cache that keeps two fresh steps cannot be faster than two uncached steps: every call still encodes the images, runs the video expert and computes those two steps. Caching the first of two steps, refreshed every fourth call, saves at most 11--14% of a call by our cost model. Cached policies are also fragile. Their losses tracked the share of the flow path the cached steps cover: under Fast-WAM's current time shift, C3^3ache's eight cached steps cover 80% of the path instead of 44%, and the cached policy succeeds on 63% of states against 95% uncached. Resampling the noise on every call cost caches covering 60% or more of the path 14--40 points. These results argue for fewer-step baselines and for testing reuse under the deployed sampler and noise settings. We release our evaluation harness, logs and a checklist.

Figures & tables

Appendix figures & tables9 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. C3^3ache: Accelerating World Action Models with Cross Inference Chunk Cache

    Jun 8, 2026Weisen Zhao, Lam Nguyen, Zhicong Lu +1World Model LearningAction-Conditioned World Models

  2. WAM-Cache: Staleness-Bounded KV Reuse for Efficient World Action Models

    Oct 8, 2026Kai Ding, Yang He, Ruijie Quan +1Efficient Inference for World Action ModelsDiffusion Model Caching

  3. ΔΔ Flow Matching for Robot Foundation Models

    Oct 11, 2026Kuangji Zuo, Tuo An, Xinying Guo +9Flow Matching for Action GenerationRobot Foundation Models