Hello there, I trained my own model using LSTM and a window size of 25 for temporal encoding, and then did the TensorFlow conversion for deployment on the edgeimpulse platform. I downloaded the model as a C++ library so that I can deploy it on the Thingy 53 device. I used the official edge impulse documentation to be able to use the necceasry functions to call my model. After building and flashing the thingy 53, whenever I run the prediction (regression), I end up getting a "nan" as the prediction result. Could you please giude me on why would I get this error of nan prediction? and what can I do to fix this issue?
