Open happyjohn360 opened 4 years ago
Hi! I'm also interested in uint8 quantization, any update?
I do an int8 quantization, 4x model size smaller, but the inference time is increasing. And you have to adjust the threshold to recognize correctly.
Any updates here? Can you share the uint8 quantized model? or the process you did to quantize it?
Do you have plans to quantize the model? Or can you provide the original .h5 file or saved model for me to do uint8 quantization?
Thanks.