I am currently working a project to generate live video clips using biggan

I started by using the default code from the deepmind colab but i am now facing a optimization problem

I need to have as many fps but doing a sess.run() for every frame is very heavy, i tried looking on maybe converting the model to a TF Lite model, but i didn’t find any information on how to do it from a tensorflow hub module

I’m pretty new to tensorflow and i didn’t find any other idea to improve my code, so i would apreciate any help