quantization not working on newer versions of tflite

[SuranjanKTH]

Hi,

I ran the hello world example quantization script and it seems to increase the model size. This does not occue with pete wardens original notebook. He uses tensorflow 2.0.0. Using the 2.18.0 in the current example cause the quantization to increase the model size. Can the code be modified to still quantize properly with newer version of tensorflowlite to the same level as pete wardens work?