BBAI-64 what data format should I pass to compiled tflite model for libtidl_tfl_delegate

Hi there. Does anyone have experience compiling custom model with edgeai-tidl-tools for tfl_delegate or/and can help me with this task?
I have my test model trained, converted to tflite and working with tflite runtime.
I’m feeding the input tensor with images as float32 RGB array and model is working as expected.
However after compiling it using edgeai-tidl-tools:

cd examples/osrt_python/tfl
python3 tflrt_delegate.py -c

and running generated model (with all artifacts) using libtidl_tfl_delegate.so - the model stopped to work.
So my question is why? :slight_smile: