ZCU104 and Jetson Xavier AGX for online retraining

Dear all,

I would like to deploy a CNN on the ZCU 104 using pynq. However, the application I have in mind also requires the online retraining of the ML model. Given the convenience of a GPU’s parallelism, I would like to integrate the ZCU104 with the Jetson Xavier AGX.

Do you have any experience with the best practices for enabling the connection between these two boards?

Thank you so much for your attention and participation.