How to run Phi3 on NPU with ORT+DML via OnnxRuntimeGenai library

[Gusha-nye]

Currently, in the example given in the official microsoft documentation(https://learn.microsoft.com/zh-cn/windows/ai/models/get-started-models-genai), it is possible to run phi3 models on GPUs using DML acceleration under the ORT framework with the OnnxRuntimeAI library. However, I now want to deploy it to reason on NPU, but I don't seem to see any parameter in the GeneratorParams class or Generator class of the OnnxRuntimeGenAI library where I can set the hardware (CPU, GPU, NPU) reasoning platform. Is it currently possible to implement ORT+DML based running Phi3 on NPU? If it is possible, please tell me how to set it up? If there is a demo can you send it for reference? (C#, python, C++ are all acceptable)