This library currently lacks support for NPU (Neural Processing Unit) delegation, which limits its ability to leverage specialized hardware for faster and more efficient inference on devices with NPU capabilities. Integrating NPU delegate support will optimize inference performance for supported models.