ds_tcn convert to tflite model to be a large model

[kismeter]

Try to training a ds_tcs model with own datasets, after training completed the model is exported to onnx, then convert to tflite mode, using netron.app to view the model, it notify large model detected and the model become very complex

[mlxu995]

Is the architecture of the ONNX model the same as the TFLite model? If not, there may be something wrong with the conversion.