Skip to main content
CoVSpec: Efficient Device-Edge Co-Inference for Vision-Language Models via Speculative Decoding | Signal Canvas | ScienceToStartup