I know its not enough times to find the right weights. But the purpose of Bayesian is to find the right hyperparameters not weights so my question was if my approach is a good tradeoff to find the right hyperparameters. Once I find those I will obviously loop over the dataset more times to find the right weights. Is this a good approach or not? and why?

this link shows conflicting arguments for and against scaling target so I went with the safest bet. stats.stackexchange.com/questions/111467/…. Yes features are increasing.

Scaling both input and target. As I said in the title my data looks somewhat monotonically increasing. Why does the dimensions matter? All I am questioning is if the well-adopted method of making a scaler based on the training data and then scale the testing holds for a non shuffled dataset for lstm. just because all the testing data would be all > 1 due to this monotonically increasing pattern.