We were using int8 model with calibration file for better accuracy and fast i want to know whether this calibration feature is available for FP16 model weights as well so that we can improve the accuracy of FP16
Reference used :
2 posts - 2 participants