Researchers introduced CARE, a certified accelerator-selection method that uses paired rollouts on a calibration set to give finite-sample guarantees that acceleration-induced failure risk stays below…
Jeff Cui and collaborators announced Patch Policy, a robot-control method that beats OpenVLA-OFT with 0.7% of its parameters and trains on a single NVIDIA RTX 5090 GPU. The approach freezes a Vision T…
Researchers have developed AURA-Mem, a memory system for robot policies that maintains a constant 4,224-byte inference state regardless of episode length, compared to a KV-cache that grows 6,061 times…
Researchers introduced ActQuant, a sub-4-bit quantization framework for Vision-Language-Action (VLA) models that reduces computational demands for edge deployment. The method achieved 95% performance …