An experiment to integrate ML4F into MicroPython.
Clone this repository and initialise the git submodules (recursive submodule init will unnecessarily initialise all submodules from all MicroPython ports):
git clone https://github.com/microbit-foundation/micropython-ml4f-poc.git
cd micropython-ml4f-poc
git submodule update --init
git -C micropython-microbit-v2 submodule update --init
git -C ml-module/pxt-ml-runner-poc submodule update --init
Changes to the micro:bit MicroPython port are required for this experiment:
git -C micropython-microbit-v2 apply ../micropython-microbit-v2.patch
Build MicroPython's mpy-cross:
make -C micropython-microbit-v2/lib/micropython/mpy-cross
And build MicroPython with the ML module (and example model) included:
make -C micropython-microbit-v2/src USER_C_MODULES=../../..
Copy the hex file to the micro:bit:
cp micropython-microbit-v2/src/build/MICROBIT.hex /Volumes/MICROBIT
Transfer the example Python code into the board with microFS:
ufs put ml-module/model-example/main.py