Skip to content
arXiv cs.CV · Papers

BeCARE: Budgeted Cache Refresh for Diffusion Transformer Acceleration

arXiv:2605.27075v2 Announce Type: replace Abstract: Training-free feature caching accelerates diffusion transformer (DiT) inference by reusing or forecasting intermediate features. However, fixed schedules make compute predictable but ignore prompt- and timestep-dependent risk, whereas hand-tuned error thresholds adapt