Closed Johnson-yue closed 3 years ago
It's okay to deploy TF 1.x models and 2.x models because they are compatible. The only think you have to resolve is the difference of operators support.
If you want to use the TF 1.x operators, you can uninstall the TF in docker and install the new TensorFlow in your expected version.
I use your docker image : latest-gpu but it default install tensorflow 2.x
and my SavedModel is tf 1.x version so How I deploy it ??