Torchcelerate is a PyTorch add-on that accelerates the performance of your models. Just add a line of code to your script and you'll have better inference performance when you deploy.
And that's not all. Torchcelerate also makes it easy for you to package your PyTorch model for deployment to edge devices and embedding in Java, C++, C#, and Windows apps.
Check out the tutorial.ipynb