🛰️ Daily AI Frontier
‹ back to 2026-07-22

Evolving Cache Schedules for Fast Diffusion Policy Inference

Research Efficiency & Systems

Merged summary

TL;DR - EVO uses evolutionary search to optimize cache-refresh schedules for pretrained diffusion policies, accelerating visuomotor action generation without retraining. It achieves up to an 8.05× speedup while retaining near-full task performance.

  • Searches globally across transformer blocks and denoising timesteps to identify reusable activations.
  • Uses redundancy-aware initialization and target-conditioned early stopping to make offline schedule optimization practical.
  • Plugs optimized schedules directly into existing diffusion policies.
  • Reduces inference compute from 15.77G FLOPs to as little as 1.96G on manipulation benchmarks.

Sources (1)

Evolving Cache Schedules for Fast Diffusion Policy Inference

arXiv cs.CV Siying Wang, Kangye Ji, Di Wang, Fei Cheng 2026-07-22 arXiv:2607.20293

TL;DR - EVO uses evolutionary search to optimize cache-refresh schedules for pretrained diffusion policies, accelerating visuomotor action generation without retraining. It achieves up to an 8.05× speedup while retaining near-full task performance.

  • Searches globally across transformer blocks and denoising timesteps to identify reusable activations.
  • Uses redundancy-aware initialization and target-conditioned early stopping to make offline schedule optimization practical.
  • Plugs optimized schedules directly into existing diffusion policies.
  • Reduces inference compute from 15.77G FLOPs to as little as 1.96G on manipulation benchmarks.
item →