EdgeVL

New framework boosts edge vision-language model efficiency

Researchers have developed a new framework to improve the efficiency of vision-language models (VLMs) on edge devices. This approach unifies distillation and quantization-aware training, ensuring consistent optimization…