What is the best approach to convert a pytorch model to TVM?

Can i use torch==1.6.0 and torchvision== 0.7 ? becasue i trained my model using these versions…