Error in output prediction of LSTM model with EloquentTinyML version 2.4.3 about eloquenttinyml

[eloquentarduino]

I'm sorry, I'm having troubles replicating your project. If I use your sensor_model.h, I get "Only 1 subgraph is currently supported". Please add your complete Arduino sketch to your repository.

[koulnodji6056]

@eloquentarduino Thanks for the reply.
I have added the Arduino code.
Looking forward to hearing from you.

[eloquentarduino]

If you replace tf.begin(model) with while (!tf.begin(model)) Serial.print(tf.getErrorMessage());, you will see that the error is Only 1 subgraph supported.

I don't know how to solve this error (even Googling didn't helped me). I think the best move is to upgrade to a later version of TensorFlow for Micro and see if they fixed this. Since I need time to do this, you can try using the latest version of this library (which mine internally uses) and report if it works for you.

https://github.com/tanakamasayuki/Arduino_TensorFlowLite_ESP32

[koulnodji6056]

Thanks for the response. I will do that and let you know in case of any news.

[eloquentarduino]

Spent many hours trying to fix this. Upgraded Tf in Python to 2.11, upgraded to latest Arduino_TensorFlowLite library, but cannot solve this problem. Did you find anything related to a solution?

[koulnodji6056]

Unfortunately no.
Still trying but unsuccessful