Creating Model-specific assets for QNN EP after fine-tuning

[tc-wolf]

In https://onnxruntime.ai/docs/genai/howto/build-models-for-snapdragon.html#add-other-assets there is a reference to model-specific assets (tokenizer, quantizer/dequantizer/position processor ONNX models). These are provided for Phi-3.5-mini-instruct and Llama 3.2 3B Instruct, but how are these created?

If I wanted to deploy a fused fine-tuned model based on Llama 3.2 3B (or even a non-fine-tuned Llama 3.2 1B), I'd have to recreate these, and I can't find any information on how to do so.