Compilation of keras-rl agent into MLIR code?

Hello everyone,

I am new here. I hope my question is not stupid.
Is there a way to process a reinforcement learning agent using the MLIR compilation chain (tf-opt, etc.) ?

Thank you very much

H