I’m the creator of the tensorflow micropython examples project which is a micropython port of tensorflow lite for microcontrollers.
It supports ESP32, RP2040 and STM32 although the best support at the moment is ESP32 and RP2040.
The purpose of my project is to lower the bar of entry of what is needed to get started and experiment with TinyML and you can get started on a < $5 USD development board.
But also to support higher end microcontrollers like Teensy 4.1, Nucleo H743ZI2 and others like ESP32S3 that have SPIRAM and can be used to run larger < 10 MB models and explore running the smallest tensorflow lite models on a microcontroller.
Firmware for supported boards are downloadable via github actions.
We have working micropython implementations of hello-world, micro-speech and person-detection.
Some people are using it to run custom models built using edge impulse.
I’m posting to make more people aware of this project and maybe to try it and give feedback so I an improve it.