Jul 2026
Evolving Cache Schedules for Fast Diffusion Policy Inference
Empirical benchmarks show that EVO preserves near-full performance while substantially reducing computation, achieving up to 8.05x action-generation speedup and reducing FLOPs from 15.77G to as low as 1.96G.
Si Wang, Kangye Ji, Dixiang Wang et al.
· arXiv.org · 1 citation