Is it possible to run models optimized for NPU with AMD processor?

[mpaulazamin]

Hi!

After your help, I was able to download Hugging Face ONNX models. It worked for running on GPU.

Now, I'm trying to run models on NPU and Hybrid mode (NPU + GPU). I downloaded amd/Llama-3.2-3B-Instruct-awq-g128-int4-asym-bf16-onnx-ryzen-strix. However, I received the following error:

I pasted all the .dll libraries necessary to run models on NPU inside the model folder, but it did not work:

Is there a way I can run models on NPU that are optimized for AMD processor?

[samuel100]

@mpaulazamin currently foundry local only supports QNN NPUs. We're working on automatic selection of execution providers that will open up more NPU providers, such as AMD. In the case of GPU - if you select the model-name-generic-gpu we'll use WebGPU that will select the GPU device.