How to select NPU instead of GPU on Intel laptop?

Where could I find a code sample showing how to run genai on NPU instead of GPU using `DmlExecutionProvider` in Python?

Conda environment:
```yaml
name: phi3-onnx-dml
channels:
- conda-forge
- nodefaults
dependencies:
- python=3.12
- vs2015_runtime=14.40.33810
- pip
- pip:
- ipykernel==6.29.5
- numpy==1.26.4
- huggingface-hub[cli]==0.26.2
- onnxruntime-directml==1.20.0
- onnxruntime-genai-directml==0.4.0
```

No file, test, or entry point is named. Start by locating the Python usage and DirectML execution-provider documentation related to the supplied onnxruntime-genai-directml environment; done should be a clear code sample showing how to run genai on an Intel NPU instead of the GPU.

Written by the indexing model from the issue text.