הפסקה, חלוקה, צמצום: חשיבה מחדש של הגוף הפעולי במדיה של דיפוזיה

תקציר מקורי באנגליתarXiv:2511.12101v3 Announce Type: replace-cross Abstract: Many recent Vision-Language-Action models employ diffusion or flow-matching backbones with hundreds of millions of parameters for action generation. However, unlike image synthesis where the output spans millions of diverse pixels, a manipulation policy generates only short sequences of low-dimensional, physically correlated action values, a far simpler target that may not require such capacity. We confirm this intuition and show that, in modulation-conditioned diffusion policies, task adaptation can be routed entirely through the conditioning pathway, leaving a frozen, observation-free backbone that serves as a reusable trajectory prior. To establish this, we introduce a decoupled training recipe: a general-purpose action head is f