Federated Learning and onboard training using Raspberry Pi Pico RP2040 and Tensorflow

Question

I am trying to implement federated learning in a group of raspberry pi picos. I had the following queries:

1. What would be the best means for performing onboard ML training on the Raspberry Pi Pico, using real-time sensor data?

2. I was wondering whether tensorflow federated (tff) supports Raspberry Pi Pico’s RP2040. If it does then whether there are any examples that I can follow?

3. Can a tff model be converted to a tensorflow-lite (tf-lite) model. If it can be converted, then how can I perform onboard training using the tff tf-lite model?

Would highly appreciate a response.

I have yet to go forward with an implementation