Generate human faces with a diffusion model on an Alif Ensemble E8
Introduction
Understand the pico-faces workflow
Set up the development tools and DevKit
Export the pico-faces model for Ethos-U85
Prepare the DevKit for debugging
Generate and validate faces on the DevKit
Next Steps

  Log an issue

  Fork and edit

  Discuss on Discord

Clone the example branch

Clone the branch that contains the pico-faces application and check out the revision:

    

        
        
git clone --branch hackathon-pico-faces --single-branch https://github.com/Arm-Examples/CMSIS-Executorch.git
cd CMSIS-Executorch
git checkout bb82e8f0c67b9e603ba6b0a48984c2f72866afce
code .

When Visual Studio Code (VS Code) opens the workspace:

  1. Accept the Arm Tools Environment Manager prompt to activate the tools pinned in vcpkg-configuration.json.
  2. Accept the prompt to install the required CMSIS packs.
  3. Open the CMSIS view and select Manage Solution.
  4. Select the DevKit-E8 target type, then select Apply.

The managed environment supplies the following:

  • CMSIS-Toolbox
  • Arm Compiler for Embedded
  • GCC
  • CMake
  • Ninja

The CMSIS packs provide the Alif device support and the ExecuTorch runtime components.

Set up the Python environment

To set up the Python environment:

  1. Open the VS Code Command Palette.
  2. Select Tasks: Run Task.
  3. Select Setup Python virtual environment.

The task creates .venv and installs the pinned export packages. It downloads about 60 MB of pico-faces checkpoints into model/pico_faces/. The Python packages include several gigabytes of PyTorch dependencies, so the first setup can take some time.

Note

The exporter supports Python 3.10 through 3.13. If your system Python isn’t in that range, start by installing uv .

Restart VS Code after installation so that its tasks can find uv on PATH. Then, confirm that uv is available in the VS Code terminal:

    

        
        
uv --version

Select Setup Python virtual environment (uv) and enter 3.12 at the Python version prompt. This task can download a compatible interpreter. The physical DevKit was validated using Python 3.12.13.

Create the AI layer

To create the AI layer:

  1. Open the VS Code Command Palette.
  2. Select Tasks: Run Task.
  3. Select Create AI layer.

If the task reports that cmsis-executorch.cbuild-mlops.yml is missing, select Build once in the CMSIS view. The extension generates the target description. You can then run Create AI layer again.

The export performs the following operations:

  • Calibrates dit_step with 16-bit activations and 8-bit weights
  • Calibrates decode with 8-bit activations and weights
  • Delegates each method as one Ethos-U85 segment
  • Generates the CMSIS component selection and embedded model data

The task takes a few minutes. When it completes, confirm that ai_layer/ contains the following generated artifacts:

FilePurpose
model.pteExecuTorch program containing both methods
model_pte.cProgram embedded as a C array for the firmware build
model_params.hSampling schedule, image shapes, and conditioning data
ai_layer.clayer.ymlCMSIS components selected for the exported program

Review the task log. It should report one Ethos-U delegate for each model method. Only the quantize and dequantize boundaries remain on the CPU. The diffusion transformer and decoder run on the NPU.

Note

Run Create AI layer again after changing model/model.py or selecting a target with a different NPU configuration. The generated model data isn’t committed to the repository.

What you’ve accomplished and what’s next

You’ve exported the diffusion transformer and decoder as one ExecuTorch program for the DevKit’s Ethos-U85.

Next, you’ll install the boot configuration and debug stub needed to load the program on the high-performance Cortex-M55 core.

Back
Next