Closed mocleiri closed 2 years ago
I tried to copy the Nucleo H743ZI2 board but I should cross-check with the out of tree example here: https://github.com/micropython/micropython-example-boards/tree/main/boards/CUSTOM_PYBV11
to make sure I am doing it right.
The port has been created but doesn't work right. I am going to close this issue and then use #33 to track finding the cause of inference not working.
Within the stm32 boards there are a wide variety of cortex M0, M3, M4 and M7's. Each needs to be built separately on the tensorflow side.
Make separate CI to build libtensorflow-microlite.a for the different permutations. Then find out how to encode the required flavor in the board configurations.
The board specific build can then download the right tensorflow artifact to use when building.