tensorflow: TFLite: Cannot run inference on TF Lite Model: "Regular TensorFlow ops are not supported by this interpreter."

System information

• OSX
• TF 2.3.0-dev20200602

Command used to run the converter or code if you’re using the Python API If possible, please share a link to Colab/Jupyter/any notebook.

Conversion code:

    converter = tf.lite.TFLiteConverter.from_saved_model(curr_dir + "saved_model")
    tflite_model = converter.convert()

    # Save the TF Lite model.
    with tf.io.gfile.GFile(curr_dir + '/model.tflite', 'wb') as f:
        f.write(tflite_model)

Inference code:

    # Compare Inference
    import tensorflow as tf

    # Load the TFLite model and allocate tensors.
    interpreter = tf.lite.Interpreter(model_path="./model.tflite")
    interpreter.allocate_tensors()

    # Get input and output tensors.
    input_details = interpreter.get_input_details()
    output_details = interpreter.get_output_details()

The model I’m trying to convert to tflite and run inference on is SSDLite_MobileNetV2, obtained rom the Model Zoo:

http://download.tensorflow.org/models/object_detection/ssdlite_mobilenet_v2_coco_2018_05_09.tar.gz

Failure details

Conversion is successful, however I cannot run inference: Here is the error that I run into:

RuntimeError: Regular TensorFlow ops are not supported by this interpreter. 
Make sure you apply/link the Flex delegate before inference.Node number 3 (FlexTensorArrayV3) failed to prepare.

I’ve been playing around with converter settings with no luck i.e. combinations of:

    # converter.optimizations = [tf.lite.Optimize.DEFAULT]
    # converter.target_spec.supported_ops = [tf.lite.OpsSet.TFLITE_BUILTINS,
    #                                        tf.lite.OpsSet.SELECT_TF_OPS]

With none of the settings above set, or the supported_ops set, I can convert the model but cannot run inference, with a similar error as above. With optimizations set to default, it gives me an error in conversion