Open lucasjinreal opened 3 years ago
@zhiqwang thanks for the information. I saw that most model deploy on mobile devices, does the QAT trained model able to deploy on GPU and accelerate on tensorrt?
For QAT, it seems that d2go only supports faster-rcnn now.
does the QAT trained model able to deploy on GPU and accelerate on tensorrt?
I didn't know either about this.
@zhiqwang Is that convinient to add QAT on other model trained upon d2 such as SOLOV2?
Since some model such MaskRCNN gains a lot speed up and very tiny precesion drop in int8 model, is there any plan to support automatically QAT training support in d2?