Hands-on tutorials for the RUHMI Framework AI Compiler for MCU.
| Tutorial | Description | Difficulty |
|---|---|---|
| CIFAR-10 Quantization Workflow | End-to-end workflow: train a CNN, quantize with MERA & Tflite, deploy to C-code, and validate accuracy across FP32 & INT8 variants for Tflite and Mera | Beginner |
All tutorials run as Jupyter Notebooks and require a MERA virtual environment.
If you don't have one set up yet, follow the Installation Guide first.
Tutorials may require extra Python packages beyond the base MERA installation.
Install them inside your activated mera-env:
source mera-env/bin/activate
pip install ipykernel scipy scikit-learn tensorflow matplotlib seabornRegister your MERA environment as a Jupyter kernel:
python -m ipykernel install --user --name=mera-env --display-name "Python (MERA Env)"
jupyter notebookImportant
In the Jupyter interface, select Kernel → Change Kernel → Python (MERA Env) to ensure the notebook uses the correct environment.