Important
When you compile models for Tensor RT, you must compile them for the specific hardware you are using. Meaning, if you compile a model for a Jetson Nano, it will not work on a Jetson Xavier or a desktop GPU. The same goes for the other way around.
Note
Whatever version of Tensor RT you use to compile the model, you must use the same version of Tensor RT to run the model to ensure compatibility.
There are guides available online on how to compile models for Tensor RT. Here is the official Nvidia TAR file docs