YOLOV8Detector Tflite conversion model size problem.

[Alparslan-Tamer]

Hello,

I converted the model to float32 and float16 with Tflite, but I expected that the float16 model would have about 2 times smaller size than float32, but the opposite happened. Is this normal behavior?

I have prepared a sample code on Colab: https://colab.research.google.com/drive/11eyUwMcTB_8X6oJrjSBjuRRd-gBDh3Du?usp=sharing

[Alparslan-Tamer]

false alarm!!! The conversion I thought was float32 is uint8, so the model sizes are correct. I updated the colab link, anyone who needs it can look there.