I use tensorflow in the environment of WSL Ubuntu for higher version because Windows cannot support.
I use TF 2.14.0 version and training my deeplearning model.
However, As shown above, when training starts, training isn't going fast than Windows.
When model is learning on Windows, it is automatically learning fast even at first moment.
After some times, Cuda is using not fully and just 19%.
After more some times, then Cuda is starting to use lots of Cuda.
And almost 70% continuously.
Is it normal for WSL Ubuntu environment?