It seems that some Python APIs about Int8 calibration are provided in TensorRT 4.0 but I can’t find any sample code in the documentation. Could you provide some examples showing whole workflow to quantize CNN models (LeNet, ResNet, etc.)? BTW, I wonder if TX2 is capable of INT8 inference?
TX2 doesn’t have the hardware to do INT8.
The python API for INT8 calibration is documented at
Python workflows are documented at
I am looking for examples on how to use Python Int8Calibrator, which I can’t find in the developer guild. There is only a C++ SampleINT8.
Has anyone found a good source on python int8 inference? Some methods in my calibrator are not working and since I dont have an access to the tensorrt source code I can’t figure out what’s wrong (my code is based on: https://devblogs.nvidia.com/int8-inference-autonomous-vehicles-tensorrt/).