I have a Tensorflow model that has been trained and saved successfully. Then I used it for a while, with no change to anything of the model or the environment. Now, after two month I am running the same model on the same data and it is giving results way worst than what is was before.
Has anyone had any similar experience with the Tensorflow?
I thought there are some changes to the environement perhaps that is causing this issue, so I installed the Tensorflow 2.5, CuDNN 8.1, and CUDA 11.2 as required by the original environment and tested the model again, however, it fails again.
I also tested it by Tensorflow 2.11, it fails in the same way.