How to handle explicit state for RNNs in mediapipe

Question

I am trying to run a RNN in mediapipe. Since I need to reset the model\'s state at certain times and didn\'t want to write a custom inference calculator, I figured the easie