how dropout could change the train/test MSE loss function

Question

Good morning, i’m trying to test my data with the following code. i tried to use dropout in my code but unfortunately my validation loss is lower than my training loss despi